fix(deps): update github.com/tmc/langchaingo digest to 2c309cf (#1097)

[![Mend
Renovate](https://app.renovatebot.com/images/banner.svg)](https://renovatebot.com)

This PR contains the following updates:

| Package | Type | Update | Change |
|---|---|---|---|
| [github.com/tmc/langchaingo](https://togithub.com/tmc/langchaingo) |
require | digest | `9c8845b` -> `2c309cf` |

---

### Configuration

📅 **Schedule**: Branch creation - At any time (no schedule defined),
Automerge - At any time (no schedule defined).

🚦 **Automerge**: Disabled by config. Please merge this manually once you
are satisfied.

♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the
rebase/retry checkbox.

🔕 **Ignore**: Close this PR and you won't be reminded about this update
again.

---

- [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check
this box

---

This PR has been generated by [Mend
Renovate](https://www.mend.io/free-developer-tools/renovate/). View
repository job log
[here](https://developer.mend.io/github/go-skynet/LocalAI).

<!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiIzNi45Ny4xIiwidXBkYXRlZEluVmVyIjoiMzYuOTcuMSIsInRhcmdldEJyYW5jaCI6Im1hc3RlciJ9-->

Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>

.env

@@ -23,6 +23,12 @@ MODELS_PATH=/models
 ## Enable debug mode
 # DEBUG=true
 
+## Disables COMPEL (Diffusers)
+# COMPEL=0
+
+## Enable/Disable single backend (useful if only one GPU is available)
+# SINGLE_ACTIVE_BACKEND=true
+
 ## Specify a build type. Available: cublas, openblas, clblas.
 ## cuBLAS: This is a GPU-accelerated version of the complete standard BLAS (Basic Linear Algebra Subprograms) library. It's provided by Nvidia and is part of their CUDA toolkit.
 ## OpenBLAS: This is an open-source implementation of the BLAS library that aims to provide highly optimized code for various platforms. It includes support for multi-threading and can be compiled to use hardware-specific features for additional performance. OpenBLAS can run on many kinds of hardware, including CPUs from Intel, AMD, and ARM.
@@ -44,3 +50,20 @@ MODELS_PATH=/models
 
 ## Specify a default upload limit in MB (whisper)
 # UPLOAD_LIMIT
+
+## List of external GRPC backends (note on the container image this variable is already set to use extra backends available in extra/)
+# EXTERNAL_GRPC_BACKENDS=my-backend:127.0.0.1:9000,my-backend2:/usr/bin/backend.py
+
+### Advanced settings ###
+### Those are not really used by LocalAI, but from components in the stack ###
+##
+### Preload libraries
+# LD_PRELOAD=
+
+### Huggingface cache for models
+# HUGGINGFACE_HUB_CACHE=/usr/local/huggingface
+
+### Python backends GRPC max workers
+### Default number of workers for GRPC Python backends.
+### This actually controls wether a backend can process multiple requests or not.
+# PYTHON_GRPC_MAX_WORKERS=1

.github/PULL_REQUEST_TEMPLATE.md

@@ -8,16 +8,24 @@ This PR fixes #
 **[Signed commits](../CONTRIBUTING.md#signing-off-on-commits-developer-certificate-of-origin)**
 - [ ] Yes, I signed my commits.
  
-
 <!--
 Thank you for contributing to LocalAI! 
 
-Contributing Conventions:
+Contributing Conventions
+-------------------------
 
-1. Include descriptive PR titles with [<component-name>] prepended.
-2. Build and test your changes before submitting a PR. 
+The draft above helps to give a quick overview of your PR.
+
+Remember to remove this comment and to at least:
+
+1. Include descriptive PR titles with [<component-name>] prepended. We use [conventional commits](https://www.conventionalcommits.org/en/v1.0.0/).
+2. Build and test your changes before submitting a PR (`make build`). 
 3. Sign your commits
+4. **Tag maintainer:** for a quicker response, tag the relevant maintainer (see below).
+5. **X/Twitter handle:** we announce bigger features on X/Twitter. If your PR gets announced, and you'd like a mention, we'll gladly shout you out!
 
 By following the community's contribution conventions upfront, the review process will 
 be accelerated and your PR merged more quickly.
+
+If no one reviews your PR within a few days, please @-mention @mudler.
 -->

.github/workflows/image.yml

@@ -96,7 +96,7 @@ jobs:
 
       - name: Docker meta
         id: meta
-        uses: docker/metadata-action@v4
+        uses: docker/metadata-action@v5
         with:
           images: quay.io/go-skynet/local-ai
           tags: |
@@ -118,14 +118,14 @@ jobs:
 
       - name: Login to DockerHub
         if: github.event_name != 'pull_request'
-        uses: docker/login-action@v2
+        uses: docker/login-action@v3
         with:
           registry: quay.io
           username: ${{ secrets.LOCALAI_REGISTRY_USERNAME }}
           password: ${{ secrets.LOCALAI_REGISTRY_PASSWORD }}
 
       - name: Build and push
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           builder: ${{ steps.buildx.outputs.name }}
           build-args: |

.github/workflows/release.yaml

@@ -22,6 +22,9 @@ jobs:
         uses: actions/checkout@v3
         with:
           submodules: true
+      - uses: actions/setup-go@v4
+        with:
+          go-version: '>=1.21.0'
       - name: Dependencies
         run: |
           sudo apt-get update
@@ -60,6 +63,9 @@ jobs:
         uses: actions/checkout@v3
         with:
           submodules: true
+      - uses: actions/setup-go@v4
+        with:
+          go-version: '>=1.21.0'
       - name: Build
         id: build
         env:

.github/workflows/test.yml

@@ -16,12 +16,21 @@ concurrency:
 jobs:
   ubuntu-latest:
     runs-on: ubuntu-latest
-
+    strategy:
+      matrix:
+        go-version: ['1.21.x']
     steps:
       - name: Clone
         uses: actions/checkout@v3
         with: 
           submodules: true
+      - name: Setup Go ${{ matrix.go-version }}
+        uses: actions/setup-go@v4
+        with:
+          go-version: ${{ matrix.go-version }}
+      # You can test your matrix by printing the current Go version
+      - name: Display Go version
+        run: go version
       - name: Dependencies
         run: |
           sudo apt-get update
@@ -52,13 +61,21 @@ jobs:
 
   macOS-latest:
     runs-on: macOS-latest
-
+    strategy:
+      matrix:
+        go-version: ['1.21.x']
     steps:
       - name: Clone
         uses: actions/checkout@v3
         with: 
           submodules: true
-
+      - name: Setup Go ${{ matrix.go-version }}
+        uses: actions/setup-go@v4
+        with:
+          go-version: ${{ matrix.go-version }}
+      # You can test your matrix by printing the current Go version
+      - name: Display Go version
+        run: go version
       - name: Test
         run: |
           CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make test

.gitignore

@@ -1,5 +1,6 @@
 # go-llama build artifacts
 go-llama
+go-llama-stable
 /gpt4all
 go-stable-diffusion
 go-piper
@@ -22,6 +23,8 @@ LocalAI
 local-ai
 # prevent above rules from omitting the helm chart
 !charts/*
+# prevent above rules from omitting the api/localai folder
+!api/localai
 
 # Ignore models
 models/*

CONTRIBUTING.md

@@ -0,0 +1,72 @@
+# Contributing to localAI
+
+Thank you for your interest in contributing to LocalAI! We appreciate your time and effort in helping to improve our project. Before you get started, please take a moment to review these guidelines.
+
+## Table of Contents
+
+- [Getting Started](#getting-started)
+  - [Prerequisites](#prerequisites)
+  - [Setting up the Development Environment](#setting-up-the-development-environment)
+- [Contributing](#contributing)
+  - [Submitting an Issue](#submitting-an-issue)
+  - [Creating a Pull Request (PR)](#creating-a-pull-request-pr)
+- [Coding Guidelines](#coding-guidelines)
+- [Testing](#testing)
+- [Documentation](#documentation)
+- [Community and Communication](#community-and-communication)
+
+
+
+## Getting Started
+
+### Prerequisites
+
+- Golang [1.21]
+- Git
+- macOS/Linux
+
+### Setting up the Development Environment and running localAI in the local environment
+
+1. Clone the repository: `git clone https://github.com/go-skynet/LocalAI.git`
+2. Navigate to the project directory: `cd LocalAI`
+3. Install the required dependencies: `make prepare`
+4. Run LocalAI: `make run`
+
+## Contributing
+
+We welcome contributions from everyone! To get started, follow these steps:
+
+### Submitting an Issue
+
+If you find a bug, have a feature request, or encounter any issues, please check the [issue tracker](https://github.com/go-skynet/LocalAI/issues) to see if a similar issue has already been reported. If not, feel free to [create a new issue](https://github.com/go-skynet/LocalAI/issues/new) and provide as much detail as possible.
+
+### Creating a Pull Request (PR)
+
+1. Fork the repository.
+2. Create a new branch with a descriptive name: `git checkout -b [branch name]`
+3. Make your changes and commit them.
+4. Push the changes to your fork: `git push origin [branch name]`
+5. Create a new pull request from your branch to the main project's `main` or `master` branch.
+6. Provide a clear description of your changes in the pull request.
+7. Make any requested changes during the review process.
+8. Once your PR is approved, it will be merged into the main project.
+
+## Coding Guidelines
+
+- No specific coding guidelines at the moment. Please make sure the code can be tested. The most popular lint tools like []`golangci-lint`](https://golangci-lint.run) can help you here.
+
+## Testing
+
+`make test` cannot handle all the model now. Please be sure to add a test case for the new features or the part was changed.
+
+## Documentation
+
+- We are welcome the contribution of the documents, please open new PR in the official document repo [localai-website](https://github.com/go-skynet/localai-website)
+
+## Community and Communication
+
+- You can reach out via the Github issue tracker.
+- Open a new discussion at [Discussion](https://github.com/go-skynet/LocalAI/discussions)
+- Join the Discord channel [Discord](https://discord.gg/uJAeKSAGDy)
+
+---

Dockerfile

@@ -1,4 +1,4 @@
-ARG GO_VERSION=1.20-bullseye
+ARG GO_VERSION=1.21-bullseye
 
 FROM golang:$GO_VERSION as requirements
 
@@ -11,7 +11,7 @@ ARG TARGETARCH
 ARG TARGETVARIANT
 
 ENV BUILD_TYPE=${BUILD_TYPE}
-ENV EXTERNAL_GRPC_BACKENDS="huggingface-embeddings:/build/extra/grpc/huggingface/huggingface.py,autogptq:/build/extra/grpc/autogptq/autogptq.py,bark:/build/extra/grpc/bark/ttsbark.py,diffusers:/build/extra/grpc/diffusers/backend_diffusers.py,exllama:/build/extra/grpc/exllama/exllama.py"
+ENV EXTERNAL_GRPC_BACKENDS="huggingface-embeddings:/build/extra/grpc/huggingface/huggingface.py,autogptq:/build/extra/grpc/autogptq/autogptq.py,bark:/build/extra/grpc/bark/ttsbark.py,diffusers:/build/extra/grpc/diffusers/backend_diffusers.py,exllama:/build/extra/grpc/exllama/exllama.py,vall-e-x:/build/extra/grpc/vall-e-x/ttsvalle.py,vllm:/build/extra/grpc/vllm/backend_vllm.py"
 ENV GALLERIES='[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.yaml"}, {"url": "github:go-skynet/model-gallery/huggingface.yaml","name":"huggingface"}]'
 ARG GO_TAGS="stablediffusion tts"
 
@@ -43,10 +43,13 @@ RUN if [ "${TARGETARCH}" = "amd64" ]; then \
         pip install git+https://github.com/suno-ai/bark.git diffusers invisible_watermark transformers accelerate safetensors;\
     fi
 RUN if [ "${BUILD_TYPE}" = "cublas" ] && [ "${TARGETARCH}" = "amd64" ]; then \
-        pip install torch && pip install auto-gptq https://github.com/jllllll/exllama/releases/download/0.0.10/exllama-0.0.10+cu${CUDA_MAJOR_VERSION}${CUDA_MINOR_VERSION}-cp39-cp39-linux_x86_64.whl;\
+        pip install torch vllm && pip install auto-gptq https://github.com/jllllll/exllama/releases/download/0.0.10/exllama-0.0.10+cu${CUDA_MAJOR_VERSION}${CUDA_MINOR_VERSION}-cp39-cp39-linux_x86_64.whl;\
     fi
 RUN pip install -r /build/extra/requirements.txt && rm -rf /build/extra/requirements.txt
 
+# Vall-e-X
+RUN git clone https://github.com/Plachtaa/VALL-E-X.git /usr/lib/vall-e-x && cd /usr/lib/vall-e-x && pip install -r requirements.txt
+
 WORKDIR /build
 
 # OpenBLAS requirements
@@ -130,6 +133,10 @@ WORKDIR /build
 COPY . .
 RUN make prepare-sources
 COPY --from=builder /build/local-ai ./
+
+# Copy VALLE-X as it's not a real "lib"
+RUN cp -rfv /usr/lib/vall-e-x/* ./
+
 # To resolve exllama import error
 RUN if [ "${BUILD_TYPE}" = "cublas" ] && [ "${TARGETARCH:-$(go env GOARCH)}" = "amd64" ]; then \
         cp -rfv /usr/local/lib/python3.9/dist-packages/exllama extra/grpc/exllama/;\

Makefile

@@ -4,11 +4,13 @@ GOVET=$(GOCMD) vet
 BINARY_NAME=local-ai
 
 # llama.cpp versions
-GOLLAMA_VERSION?=50cee7712066d9e38306eccadcfbb44ea87df4b7
+GOLLAMA_VERSION?=d9f6176409de0a2b5ce798de502545c6721e346e
+
+GOLLAMA_STABLE_VERSION?=50cee7712066d9e38306eccadcfbb44ea87df4b7
 
 # gpt4all version
 GPT4ALL_REPO?=https://github.com/nomic-ai/gpt4all
-GPT4ALL_VERSION?=0f2bb506a8ee752afc06cbb832773bf85b97eef3
+GPT4ALL_VERSION?=27a8b020c36b0df8f8b82a252d261cda47cf44b8
 
 # go-ggml-transformers version
 GOGGMLTRANSFORMERS_VERSION?=ffb09d7dd71e2cbc6c5d7d05357d230eea6f369a
@@ -36,6 +38,7 @@ STABLEDIFFUSION_VERSION?=d89260f598afb809279bc72aa0107b4292587632
 GOGGLLM_VERSION?=862477d16eefb0805261c19c9b0d053e3b2b684b
 
 export BUILD_TYPE?=
+export CMAKE_ARGS?=
 CGO_LDFLAGS?=
 CUDA_LIBPATH?=/usr/local/cuda/lib64/
 GO_TAGS?=
@@ -62,9 +65,12 @@ ifndef UNAME_S
 UNAME_S := $(shell uname -s)
 endif
 
-# workaround for rwkv.cpp
 ifeq ($(UNAME_S),Darwin)
-        CGO_LDFLAGS += -lcblas -framework Accelerate 
+	CGO_LDFLAGS += -lcblas -framework Accelerate
+ifneq ($(BUILD_TYPE),metal)
+    # explicit disable metal if on Darwin and metal is disabled
+	CMAKE_ARGS+=-DLLAMA_METAL=OFF
+endif
 endif
 
 ifeq ($(BUILD_TYPE),openblas)
@@ -101,6 +107,8 @@ ifeq ($(findstring tts,$(GO_TAGS)),tts)
 	OPTIONAL_GRPC+=backend-assets/grpc/piper
 endif
 
+GRPC_BACKENDS?=backend-assets/grpc/langchain-huggingface backend-assets/grpc/falcon-ggml backend-assets/grpc/bert-embeddings backend-assets/grpc/falcon backend-assets/grpc/bloomz backend-assets/grpc/llama backend-assets/grpc/llama-stable backend-assets/grpc/gpt4all backend-assets/grpc/dolly backend-assets/grpc/gpt2 backend-assets/grpc/gptj backend-assets/grpc/gptneox backend-assets/grpc/mpt backend-assets/grpc/replit backend-assets/grpc/starcoder backend-assets/grpc/rwkv backend-assets/grpc/whisper $(OPTIONAL_GRPC)
+
 .PHONY: all test build vendor
 
 all: help
@@ -192,17 +200,23 @@ go-llama:
 	git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp go-llama
 	cd go-llama && git checkout -b build $(GOLLAMA_VERSION) && git submodule update --init --recursive --depth 1
 
+go-llama-stable:
+	git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp go-llama-stable
+	cd go-llama-stable && git checkout -b build $(GOLLAMA_STABLE_VERSION) && git submodule update --init --recursive --depth 1
+
 go-llama/libbinding.a: go-llama
 	$(MAKE) -C go-llama BUILD_TYPE=$(BUILD_TYPE) libbinding.a
 
+go-llama-stable/libbinding.a: go-llama-stable
+	$(MAKE) -C go-llama-stable BUILD_TYPE=$(BUILD_TYPE) libbinding.a
+
 go-piper/libpiper_binding.a:
 	$(MAKE) -C go-piper libpiper_binding.a example/main
 
-get-sources: go-llama go-ggllm go-ggml-transformers gpt4all go-piper go-rwkv whisper.cpp go-bert bloomz go-stable-diffusion
+get-sources: go-llama go-llama-stable go-ggllm go-ggml-transformers gpt4all go-piper go-rwkv whisper.cpp go-bert bloomz go-stable-diffusion
 	touch $@
 
 replace:
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(shell pwd)/go-llama
 	$(GOCMD) mod edit -replace github.com/nomic-ai/gpt4all/gpt4all-bindings/golang=$(shell pwd)/gpt4all/gpt4all-bindings/golang
 	$(GOCMD) mod edit -replace github.com/go-skynet/go-ggml-transformers.cpp=$(shell pwd)/go-ggml-transformers
 	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(shell pwd)/go-rwkv
@@ -220,6 +234,7 @@ prepare-sources: get-sources replace
 rebuild: ## Rebuilds the project
 	$(GOCMD) clean -cache
 	$(MAKE) -C go-llama clean
+	$(MAKE) -C go-llama-stable clean
 	$(MAKE) -C gpt4all/gpt4all-bindings/golang/ clean
 	$(MAKE) -C go-ggml-transformers clean
 	$(MAKE) -C go-rwkv clean
@@ -231,7 +246,7 @@ rebuild: ## Rebuilds the project
 	$(MAKE) -C go-ggllm clean
 	$(MAKE) build
 
-prepare: prepare-sources $(OPTIONAL_TARGETS) 
+prepare: prepare-sources $(OPTIONAL_TARGETS)
 	touch $@
 
 clean: ## Remove build related file
@@ -239,6 +254,7 @@ clean: ## Remove build related file
 	rm -f prepare
 	rm -rf ./go-llama
 	rm -rf ./gpt4all
+	rm -rf ./go-llama-stable
 	rm -rf ./go-gpt2
 	rm -rf ./go-stable-diffusion
 	rm -rf ./go-ggml-transformers
@@ -290,9 +306,10 @@ test: prepare test-models/testmodel grpcs
 	export GO_TAGS="tts stablediffusion"
 	$(MAKE) prepare-test
 	HUGGINGFACE_GRPC=$(abspath ./)/extra/grpc/huggingface/huggingface.py TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="!gpt4all && !llama" --flake-attempts 5 -v -r ./api ./pkg
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="!gpt4all && !llama && !llama-gguf" --flake-attempts 5 -v -r ./api ./pkg
 	$(MAKE) test-gpt4all
 	$(MAKE) test-llama
+	$(MAKE) test-llama-gguf
 	$(MAKE) test-tts
 	$(MAKE) test-stablediffusion
 
@@ -304,6 +321,10 @@ test-llama: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
 	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama" --flake-attempts 5 -v -r ./api ./pkg
 
+test-llama-gguf: prepare-test
+	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama-gguf" --flake-attempts 5 -v -r ./api ./pkg
+
 test-tts: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
 	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="tts" --flake-attempts 1 -v -r ./api ./pkg
@@ -340,6 +361,8 @@ protogen-python:
 	python3 -m grpc_tools.protoc -Ipkg/grpc/proto/ --python_out=extra/grpc/exllama/ --grpc_python_out=extra/grpc/exllama/ pkg/grpc/proto/backend.proto
 	python3 -m grpc_tools.protoc -Ipkg/grpc/proto/ --python_out=extra/grpc/bark/ --grpc_python_out=extra/grpc/bark/ pkg/grpc/proto/backend.proto
 	python3 -m grpc_tools.protoc -Ipkg/grpc/proto/ --python_out=extra/grpc/diffusers/ --grpc_python_out=extra/grpc/diffusers/ pkg/grpc/proto/backend.proto
+	python3 -m grpc_tools.protoc -Ipkg/grpc/proto/ --python_out=extra/grpc/vall-e-x/ --grpc_python_out=extra/grpc/vall-e-x/ pkg/grpc/proto/backend.proto
+	python3 -m grpc_tools.protoc -Ipkg/grpc/proto/ --python_out=extra/grpc/vllm/ --grpc_python_out=extra/grpc/vllm/ pkg/grpc/proto/backend.proto
 
 ## GRPC
 
@@ -351,6 +374,7 @@ backend-assets/grpc/falcon: backend-assets/grpc go-ggllm/libggllm.a
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/falcon ./cmd/grpc/falcon/
 
 backend-assets/grpc/llama: backend-assets/grpc go-llama/libbinding.a
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(shell pwd)/go-llama
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/go-llama LIBRARY_PATH=$(shell pwd)/go-llama \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/llama ./cmd/grpc/llama/
 # TODO: every binary should have its own folder instead, so can have different metal implementations
@@ -358,6 +382,11 @@ ifeq ($(BUILD_TYPE),metal)
 	cp go-llama/build/bin/ggml-metal.metal backend-assets/grpc/
 endif
 
+backend-assets/grpc/llama-stable: backend-assets/grpc go-llama-stable/libbinding.a
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(shell pwd)/go-llama-stable
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/go-llama-stable LIBRARY_PATH=$(shell pwd)/go-llama \
+	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/llama-stable ./cmd/grpc/llama-stable/
+
 backend-assets/grpc/gpt4all: backend-assets/grpc backend-assets/gpt4all gpt4all/gpt4all-bindings/golang/libgpt4all.a
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/gpt4all/gpt4all-bindings/golang/ LIBRARY_PATH=$(shell pwd)/gpt4all/gpt4all-bindings/golang/ \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/gpt4all ./cmd/grpc/gpt4all/
@@ -421,4 +450,4 @@ backend-assets/grpc/whisper: backend-assets/grpc whisper.cpp/libwhisper.a
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/whisper.cpp LIBRARY_PATH=$(shell pwd)/whisper.cpp \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/whisper ./cmd/grpc/whisper/
 
-grpcs: prepare backend-assets/grpc/langchain-huggingface backend-assets/grpc/falcon-ggml backend-assets/grpc/bert-embeddings backend-assets/grpc/falcon backend-assets/grpc/bloomz backend-assets/grpc/llama backend-assets/grpc/gpt4all backend-assets/grpc/dolly backend-assets/grpc/gpt2 backend-assets/grpc/gptj backend-assets/grpc/gptneox backend-assets/grpc/mpt backend-assets/grpc/replit backend-assets/grpc/starcoder backend-assets/grpc/rwkv backend-assets/grpc/whisper $(OPTIONAL_GRPC)
+grpcs: prepare $(GRPC_BACKENDS)

README.md

@@ -27,7 +27,7 @@
 
 [![tests](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml)[![Build and Release](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml)[![build container images](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml)[![Bump dependencies](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml)[![Artifact Hub](https://img.shields.io/endpoint?url=https://artifacthub.io/badge/repository/localai)](https://artifacthub.io/packages/search?repo=localai)
 
-**LocalAI** is a drop-in replacement REST API that's compatible with OpenAI API specifications for local inferencing. It allows you to run LLMs (and not only) locally or on-prem with consumer grade hardware, supporting multiple model families that are compatible with the ggml format. Does not require GPU.
+**LocalAI** is a drop-in replacement REST API that's compatible with OpenAI API specifications for local inferencing. It allows you to run LLMs (and not only) locally or on-prem with consumer grade hardware, supporting multiple model families that are compatible with the ggml format, pytorch and more. Does not require GPU.
 
 <p align="center"><b>Follow LocalAI </b></p>
 
@@ -104,15 +104,30 @@ Note that this started just as a [fun weekend project](https://localai.io/#backs
 
 Check out the [Getting started](https://localai.io/basics/getting_started/index.html) section in our documentation.
 
-### 💡 Example: Use GPT4ALL-J model
+### 💡 Example: Use Luna-AI Llama model
 
-See the [documentation](https://localai.io/basics/getting_started/#example-use-gpt4all-j-model-with-docker-compose)
+See the [documentation](https://localai.io/basics/getting_started)
 
 ### 🔗 Resources
 
 - [How to build locally](https://localai.io/basics/build/index.html)
 - [How to install in Kubernetes](https://localai.io/basics/getting_started/index.html#run-localai-in-kubernetes)
 - [Projects integrating LocalAI](https://localai.io/integrations/)
+- [How tos section](https://localai.io/howtos/) (curated by our community)
+  
+## Citation
+
+If you utilize this repository, data in a downstream project, please consider citing it with:
+
+```
+@misc{localai,
+  author = {Ettore Di Giacinto},
+  title = {LocalAI: The free, Open source OpenAI alternative},
+  year = {2023},
+  publisher = {GitHub},
+  journal = {GitHub repository},
+  howpublished = {\url{https://github.com/go-skynet/LocalAI}},
+```
 
 ## ❤️ Sponsors
 
@@ -127,6 +142,11 @@ A huge thank you to our generous sponsors who support this project:
 |  [Spectro Cloud](https://www.spectrocloud.com/)  |  
 |  Spectro Cloud kindly supports LocalAI by providing GPU and computing resources to run tests on lamdalabs!  |
 
+And a huge shout-out to individuals sponsoring the project by donating hardware or backing the project. 
+
+- [Sponsor list](https://github.com/sponsors/mudler)
+- JDAM00 (donating HW for the CI)
+
 ## 🌟 Star history
 
 [![LocalAI Star history Chart](https://api.star-history.com/svg?repos=go-skynet/LocalAI&type=Date)](https://star-history.com/#go-skynet/LocalAI&Date)

api/api.go

@@ -2,12 +2,14 @@ package api
 
 import (
 	"errors"
+	"fmt"
 	"strings"
 
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/localai"
 	"github.com/go-skynet/LocalAI/api/openai"
 	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	"github.com/go-skynet/LocalAI/internal"
 	"github.com/go-skynet/LocalAI/pkg/assets"
 
@@ -19,7 +21,7 @@ import (
 	"github.com/rs/zerolog/log"
 )
 
-func App(opts ...options.AppOption) (*fiber.App, error) {
+func Startup(opts ...options.AppOption) (*options.Option, *config.ConfigLoader, error) {
 	options := options.NewOptions(opts...)
 
 	zerolog.SetGlobalLevel(zerolog.InfoLevel)
@@ -27,6 +29,65 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 		zerolog.SetGlobalLevel(zerolog.DebugLevel)
 	}
 
+	log.Info().Msgf("Starting LocalAI using %d threads, with models path: %s", options.Threads, options.Loader.ModelPath)
+	log.Info().Msgf("LocalAI version: %s", internal.PrintableVersion())
+
+	cl := config.NewConfigLoader()
+	if err := cl.LoadConfigs(options.Loader.ModelPath); err != nil {
+		log.Error().Msgf("error loading config files: %s", err.Error())
+	}
+
+	if options.ConfigFile != "" {
+		if err := cl.LoadConfigFile(options.ConfigFile); err != nil {
+			log.Error().Msgf("error loading config file: %s", err.Error())
+		}
+	}
+
+	if options.Debug {
+		for _, v := range cl.ListConfigs() {
+			cfg, _ := cl.GetConfig(v)
+			log.Debug().Msgf("Model: %s (config: %+v)", v, cfg)
+		}
+	}
+
+	if options.AssetsDestination != "" {
+		// Extract files from the embedded FS
+		err := assets.ExtractFiles(options.BackendAssets, options.AssetsDestination)
+		log.Debug().Msgf("Extracting backend assets files to %s", options.AssetsDestination)
+		if err != nil {
+			log.Warn().Msgf("Failed extracting backend assets files: %s (might be required for some backends to work properly, like gpt4all)", err)
+		}
+	}
+
+	if options.PreloadJSONModels != "" {
+		if err := localai.ApplyGalleryFromString(options.Loader.ModelPath, options.PreloadJSONModels, cl, options.Galleries); err != nil {
+			return nil, nil, err
+		}
+	}
+
+	if options.PreloadModelsFromPath != "" {
+		if err := localai.ApplyGalleryFromFile(options.Loader.ModelPath, options.PreloadModelsFromPath, cl, options.Galleries); err != nil {
+			return nil, nil, err
+		}
+	}
+
+	// turn off any process that was started by GRPC if the context is canceled
+	go func() {
+		<-options.Context.Done()
+		log.Debug().Msgf("Context canceled, shutting down")
+		options.Loader.StopAllGRPC()
+	}()
+
+	return options, cl, nil
+}
+
+func App(opts ...options.AppOption) (*fiber.App, error) {
+
+	options, cl, err := Startup(opts...)
+	if err != nil {
+		return nil, fmt.Errorf("failed basic startup tasks with error %s", err.Error())
+	}
+
 	// Return errors as JSON responses
 	app := fiber.New(fiber.Config{
 		BodyLimit:             options.UploadLimitMB * 1024 * 1024, // this is the default limit of 4MB
@@ -44,8 +105,8 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 
 			// Send custom error page
 			return ctx.Status(code).JSON(
-				openai.ErrorResponse{
-					Error: &openai.APIError{Message: err.Error(), Code: code},
+				schema.ErrorResponse{
+					Error: &schema.APIError{Message: err.Error(), Code: code},
 				},
 			)
 		},
@@ -57,36 +118,6 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 		}))
 	}
 
-	log.Info().Msgf("Starting LocalAI using %d threads, with models path: %s", options.Threads, options.Loader.ModelPath)
-	log.Info().Msgf("LocalAI version: %s", internal.PrintableVersion())
-
-	cm := config.NewConfigLoader()
-	if err := cm.LoadConfigs(options.Loader.ModelPath); err != nil {
-		log.Error().Msgf("error loading config files: %s", err.Error())
-	}
-
-	if options.ConfigFile != "" {
-		if err := cm.LoadConfigFile(options.ConfigFile); err != nil {
-			log.Error().Msgf("error loading config file: %s", err.Error())
-		}
-	}
-
-	if options.Debug {
-		for _, v := range cm.ListConfigs() {
-			cfg, _ := cm.GetConfig(v)
-			log.Debug().Msgf("Model: %s (config: %+v)", v, cfg)
-		}
-	}
-
-	if options.AssetsDestination != "" {
-		// Extract files from the embedded FS
-		err := assets.ExtractFiles(options.BackendAssets, options.AssetsDestination)
-		log.Debug().Msgf("Extracting backend assets files to %s", options.AssetsDestination)
-		if err != nil {
-			log.Warn().Msgf("Failed extracting backend assets files: %s (might be required for some backends to work properly, like gpt4all)", err)
-		}
-	}
-
 	// Default middleware config
 	app.Use(recover.New())
 
@@ -116,18 +147,6 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 		return c.Next()
 	}
 
-	if options.PreloadJSONModels != "" {
-		if err := localai.ApplyGalleryFromString(options.Loader.ModelPath, options.PreloadJSONModels, cm, options.Galleries); err != nil {
-			return nil, err
-		}
-	}
-
-	if options.PreloadModelsFromPath != "" {
-		if err := localai.ApplyGalleryFromFile(options.Loader.ModelPath, options.PreloadModelsFromPath, cm, options.Galleries); err != nil {
-			return nil, err
-		}
-	}
-
 	if options.CORS {
 		var c func(ctx *fiber.Ctx) error
 		if options.CORSAllowOrigins == "" {
@@ -141,7 +160,7 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 
 	// LocalAI API endpoints
 	galleryService := localai.NewGalleryService(options.Loader.ModelPath)
-	galleryService.Start(options.Context, cm)
+	galleryService.Start(options.Context, cl)
 
 	app.Get("/version", auth, func(c *fiber.Ctx) error {
 		return c.JSON(struct {
@@ -149,36 +168,41 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 		}{Version: internal.PrintableVersion()})
 	})
 
-	app.Post("/models/apply", auth, localai.ApplyModelGalleryEndpoint(options.Loader.ModelPath, cm, galleryService.C, options.Galleries))
-	app.Get("/models/available", auth, localai.ListModelFromGalleryEndpoint(options.Galleries, options.Loader.ModelPath))
-	app.Get("/models/jobs/:uuid", auth, localai.GetOpStatusEndpoint(galleryService))
+	modelGalleryService := localai.CreateModelGalleryService(options.Galleries, options.Loader.ModelPath, galleryService)
+	app.Post("/models/apply", auth, modelGalleryService.ApplyModelGalleryEndpoint())
+	app.Get("/models/available", auth, modelGalleryService.ListModelFromGalleryEndpoint())
+	app.Get("/models/galleries", auth, modelGalleryService.ListModelGalleriesEndpoint())
+	app.Post("/models/galleries", auth, modelGalleryService.AddModelGalleryEndpoint())
+	app.Delete("/models/galleries", auth, modelGalleryService.RemoveModelGalleryEndpoint())
+	app.Get("/models/jobs/:uuid", auth, modelGalleryService.GetOpStatusEndpoint())
+	app.Get("/models/jobs", auth, modelGalleryService.GetAllStatusEndpoint())
 
 	// openAI compatible API endpoint
 
 	// chat
-	app.Post("/v1/chat/completions", auth, openai.ChatEndpoint(cm, options))
-	app.Post("/chat/completions", auth, openai.ChatEndpoint(cm, options))
+	app.Post("/v1/chat/completions", auth, openai.ChatEndpoint(cl, options))
+	app.Post("/chat/completions", auth, openai.ChatEndpoint(cl, options))
 
 	// edit
-	app.Post("/v1/edits", auth, openai.EditEndpoint(cm, options))
-	app.Post("/edits", auth, openai.EditEndpoint(cm, options))
+	app.Post("/v1/edits", auth, openai.EditEndpoint(cl, options))
+	app.Post("/edits", auth, openai.EditEndpoint(cl, options))
 
 	// completion
-	app.Post("/v1/completions", auth, openai.CompletionEndpoint(cm, options))
-	app.Post("/completions", auth, openai.CompletionEndpoint(cm, options))
-	app.Post("/v1/engines/:model/completions", auth, openai.CompletionEndpoint(cm, options))
+	app.Post("/v1/completions", auth, openai.CompletionEndpoint(cl, options))
+	app.Post("/completions", auth, openai.CompletionEndpoint(cl, options))
+	app.Post("/v1/engines/:model/completions", auth, openai.CompletionEndpoint(cl, options))
 
 	// embeddings
-	app.Post("/v1/embeddings", auth, openai.EmbeddingsEndpoint(cm, options))
-	app.Post("/embeddings", auth, openai.EmbeddingsEndpoint(cm, options))
-	app.Post("/v1/engines/:model/embeddings", auth, openai.EmbeddingsEndpoint(cm, options))
+	app.Post("/v1/embeddings", auth, openai.EmbeddingsEndpoint(cl, options))
+	app.Post("/embeddings", auth, openai.EmbeddingsEndpoint(cl, options))
+	app.Post("/v1/engines/:model/embeddings", auth, openai.EmbeddingsEndpoint(cl, options))
 
 	// audio
-	app.Post("/v1/audio/transcriptions", auth, openai.TranscriptEndpoint(cm, options))
-	app.Post("/tts", auth, localai.TTSEndpoint(cm, options))
+	app.Post("/v1/audio/transcriptions", auth, openai.TranscriptEndpoint(cl, options))
+	app.Post("/tts", auth, localai.TTSEndpoint(cl, options))
 
 	// images
-	app.Post("/v1/images/generations", auth, openai.ImageEndpoint(cm, options))
+	app.Post("/v1/images/generations", auth, openai.ImageEndpoint(cl, options))
 
 	if options.ImageDir != "" {
 		app.Static("/generated-images", options.ImageDir)
@@ -196,16 +220,14 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 	app.Get("/healthz", ok)
 	app.Get("/readyz", ok)
 
-	// models
-	app.Get("/v1/models", auth, openai.ListModelsEndpoint(options.Loader, cm))
-	app.Get("/models", auth, openai.ListModelsEndpoint(options.Loader, cm))
+	// Experimental Backend Statistics Module
+	backendMonitor := localai.NewBackendMonitor(cl, options) // Split out for now
+	app.Get("/backend/monitor", localai.BackendMonitorEndpoint(backendMonitor))
+	app.Post("/backend/shutdown", localai.BackendShutdownEndpoint(backendMonitor))
 
-	// turn off any process that was started by GRPC if the context is canceled
-	go func() {
-		<-options.Context.Done()
-		log.Debug().Msgf("Context canceled, shutting down")
-		options.Loader.StopGRPC()
-	}()
+	// models
+	app.Get("/v1/models", auth, openai.ListModelsEndpoint(options.Loader, cl))
+	app.Get("/models", auth, openai.ListModelsEndpoint(options.Loader, cl))
 
 	return app, nil
 }

api/api_test.go

@@ -296,7 +296,7 @@ var _ = Describe("API test", func() {
 				response := postModelApplyRequest("http://127.0.0.1:9090/models/apply", modelApplyRequest{
 					URL:       "github:go-skynet/model-gallery/openllama_3b.yaml",
 					Name:      "openllama_3b",
-					Overrides: map[string]interface{}{"backend": "llama", "mmap": true, "f16": true, "context_size": 128},
+					Overrides: map[string]interface{}{"backend": "llama-stable", "mmap": true, "f16": true, "context_size": 128},
 				})
 
 				Expect(response["uuid"]).ToNot(BeEmpty(), fmt.Sprint(response))
@@ -359,6 +359,82 @@ var _ = Describe("API test", func() {
 				Expect(string(resp2.Choices[0].FinishReason)).To(Equal("function_call"), fmt.Sprint(resp2.Choices[0].FinishReason))
 			})
 
+			It("runs openllama gguf", Label("llama-gguf"), func() {
+				if runtime.GOOS != "linux" {
+					Skip("test supported only on linux")
+				}
+				modelName := "codellama"
+				response := postModelApplyRequest("http://127.0.0.1:9090/models/apply", modelApplyRequest{
+					URL:       "github:go-skynet/model-gallery/codellama-7b-instruct.yaml",
+					Name:      modelName,
+					Overrides: map[string]interface{}{"backend": "llama", "mmap": true, "f16": true, "context_size": 128},
+				})
+
+				Expect(response["uuid"]).ToNot(BeEmpty(), fmt.Sprint(response))
+
+				uuid := response["uuid"].(string)
+
+				Eventually(func() bool {
+					response := getModelStatus("http://127.0.0.1:9090/models/jobs/" + uuid)
+					return response["processed"].(bool)
+				}, "360s", "10s").Should(Equal(true))
+
+				By("testing chat")
+				resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: modelName, Messages: []openai.ChatCompletionMessage{
+					{
+						Role:    "user",
+						Content: "How much is 2+2?",
+					},
+				}})
+				Expect(err).ToNot(HaveOccurred())
+				Expect(len(resp.Choices)).To(Equal(1))
+				Expect(resp.Choices[0].Message.Content).To(Or(ContainSubstring("4"), ContainSubstring("four")))
+
+				By("testing functions")
+				resp2, err := client.CreateChatCompletion(
+					context.TODO(),
+					openai.ChatCompletionRequest{
+						Model: modelName,
+						Messages: []openai.ChatCompletionMessage{
+							{
+								Role:    "user",
+								Content: "What is the weather like in San Francisco (celsius)?",
+							},
+						},
+						Functions: []openai.FunctionDefinition{
+							openai.FunctionDefinition{
+								Name:        "get_current_weather",
+								Description: "Get the current weather",
+								Parameters: jsonschema.Definition{
+									Type: jsonschema.Object,
+									Properties: map[string]jsonschema.Definition{
+										"location": {
+											Type:        jsonschema.String,
+											Description: "The city and state, e.g. San Francisco, CA",
+										},
+										"unit": {
+											Type: jsonschema.String,
+											Enum: []string{"celcius", "fahrenheit"},
+										},
+									},
+									Required: []string{"location"},
+								},
+							},
+						},
+					})
+				Expect(err).ToNot(HaveOccurred())
+				Expect(len(resp2.Choices)).To(Equal(1))
+				Expect(resp2.Choices[0].Message.FunctionCall).ToNot(BeNil())
+				Expect(resp2.Choices[0].Message.FunctionCall.Name).To(Equal("get_current_weather"), resp2.Choices[0].Message.FunctionCall.Name)
+
+				var res map[string]string
+				err = json.Unmarshal([]byte(resp2.Choices[0].Message.FunctionCall.Arguments), &res)
+				Expect(err).ToNot(HaveOccurred())
+				Expect(res["location"]).To(Equal("San Francisco"), fmt.Sprint(res))
+				Expect(res["unit"]).To(Equal("celcius"), fmt.Sprint(res))
+				Expect(string(resp2.Choices[0].FinishReason)).To(Equal("function_call"), fmt.Sprint(resp2.Choices[0].FinishReason))
+			})
+
 			It("runs gpt4all", Label("gpt4all"), func() {
 				if runtime.GOOS != "linux" {
 					Skip("test supported only on linux")

api/backend/embeddings.go

@@ -2,7 +2,6 @@ package backend
 
 import (
 	"fmt"
-	"sync"
 
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
@@ -22,17 +21,13 @@ func ModelEmbedding(s string, tokens []int, loader *model.ModelLoader, c config.
 	var inferenceModel interface{}
 	var err error
 
-	opts := []model.Option{
+	opts := modelOpts(c, o, []model.Option{
 		model.WithLoadGRPCLoadModelOpts(grpcOpts),
 		model.WithThreads(uint32(c.Threads)),
 		model.WithAssetDir(o.AssetsDestination),
 		model.WithModel(modelFile),
 		model.WithContext(o.Context),
-	}
-
-	for k, v := range o.ExternalGRPCBackends {
-		opts = append(opts, model.WithExternalBackend(k, v))
-	}
+	})
 
 	if c.Backend == "" {
 		inferenceModel, err = loader.GreedyLoader(opts...)
@@ -80,18 +75,6 @@ func ModelEmbedding(s string, tokens []int, loader *model.ModelLoader, c config.
 	}
 
 	return func() ([]float32, error) {
-		// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
-		mutexMap.Lock()
-		l, ok := mutexes[modelFile]
-		if !ok {
-			m := &sync.Mutex{}
-			mutexes[modelFile] = m
-			l = m
-		}
-		mutexMap.Unlock()
-		l.Lock()
-		defer l.Unlock()
-
 		embeds, err := fn()
 		if err != nil {
 			return embeds, err

api/backend/image.go

@@ -1,17 +1,15 @@
 package backend
 
 import (
-	"sync"
-
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
 	"github.com/go-skynet/LocalAI/pkg/grpc/proto"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 )
 
-func ImageGeneration(height, width, mode, step, seed int, positive_prompt, negative_prompt, dst string, loader *model.ModelLoader, c config.Config, o *options.Option) (func() error, error) {
+func ImageGeneration(height, width, mode, step, seed int, positive_prompt, negative_prompt, src, dst string, loader *model.ModelLoader, c config.Config, o *options.Option) (func() error, error) {
 
-	opts := []model.Option{
+	opts := modelOpts(c, o, []model.Option{
 		model.WithBackendString(c.Backend),
 		model.WithAssetDir(o.AssetsDestination),
 		model.WithThreads(uint32(c.Threads)),
@@ -21,12 +19,15 @@ func ImageGeneration(height, width, mode, step, seed int, positive_prompt, negat
 			CUDA:          c.Diffusers.CUDA,
 			SchedulerType: c.Diffusers.SchedulerType,
 			PipelineType:  c.Diffusers.PipelineType,
+			CFGScale:      c.Diffusers.CFGScale,
+			LoraAdapter:   c.LoraAdapter,
+			LoraBase:      c.LoraBase,
+			IMG2IMG:       c.Diffusers.IMG2IMG,
+			CLIPModel:     c.Diffusers.ClipModel,
+			CLIPSubfolder: c.Diffusers.ClipSubFolder,
+			CLIPSkip:      int32(c.Diffusers.ClipSkip),
 		}),
-	}
-
-	for k, v := range o.ExternalGRPCBackends {
-		opts = append(opts, model.WithExternalBackend(k, v))
-	}
+	})
 
 	inferenceModel, err := loader.BackendLoader(
 		opts...,
@@ -39,31 +40,20 @@ func ImageGeneration(height, width, mode, step, seed int, positive_prompt, negat
 		_, err := inferenceModel.GenerateImage(
 			o.Context,
 			&proto.GenerateImageRequest{
-				Height:         int32(height),
-				Width:          int32(width),
-				Mode:           int32(mode),
-				Step:           int32(step),
-				Seed:           int32(seed),
-				PositivePrompt: positive_prompt,
-				NegativePrompt: negative_prompt,
-				Dst:            dst,
+				Height:           int32(height),
+				Width:            int32(width),
+				Mode:             int32(mode),
+				Step:             int32(step),
+				Seed:             int32(seed),
+				CLIPSkip:         int32(c.Diffusers.ClipSkip),
+				PositivePrompt:   positive_prompt,
+				NegativePrompt:   negative_prompt,
+				Dst:              dst,
+				Src:              src,
+				EnableParameters: c.Diffusers.EnableParameters,
 			})
 		return err
 	}
 
-	return func() error {
-		// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
-		mutexMap.Lock()
-		l, ok := mutexes[c.Backend]
-		if !ok {
-			m := &sync.Mutex{}
-			mutexes[c.Backend] = m
-			l = m
-		}
-		mutexMap.Unlock()
-		l.Lock()
-		defer l.Unlock()
-
-		return fn()
-	}, nil
+	return fn, nil
 }

api/backend/llm.go

@@ -6,6 +6,7 @@ import (
 	"regexp"
 	"strings"
 	"sync"
+	"unicode/utf8"
 
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
@@ -15,7 +16,17 @@ import (
 	"github.com/go-skynet/LocalAI/pkg/utils"
 )
 
-func ModelInference(ctx context.Context, s string, loader *model.ModelLoader, c config.Config, o *options.Option, tokenCallback func(string) bool) (func() (string, error), error) {
+type LLMResponse struct {
+	Response string // should this be []byte?
+	Usage    TokenUsage
+}
+
+type TokenUsage struct {
+	Prompt     int
+	Completion int
+}
+
+func ModelInference(ctx context.Context, s string, loader *model.ModelLoader, c config.Config, o *options.Option, tokenCallback func(string, TokenUsage) bool) (func() (LLMResponse, error), error) {
 	modelFile := c.Model
 
 	grpcOpts := gRPCModelOpts(c)
@@ -23,17 +34,13 @@ func ModelInference(ctx context.Context, s string, loader *model.ModelLoader, c
 	var inferenceModel *grpc.Client
 	var err error
 
-	opts := []model.Option{
+	opts := modelOpts(c, o, []model.Option{
 		model.WithLoadGRPCLoadModelOpts(grpcOpts),
 		model.WithThreads(uint32(c.Threads)), // some models uses this to allocate threads during startup
 		model.WithAssetDir(o.AssetsDestination),
 		model.WithModel(modelFile),
 		model.WithContext(o.Context),
-	}
-
-	for k, v := range o.ExternalGRPCBackends {
-		opts = append(opts, model.WithExternalBackend(k, v))
-	}
+	})
 
 	if c.Backend != "" {
 		opts = append(opts, model.WithBackendString(c.Backend))
@@ -62,40 +69,71 @@ func ModelInference(ctx context.Context, s string, loader *model.ModelLoader, c
 	}
 
 	// in GRPC, the backend is supposed to answer to 1 single token if stream is not supported
-	fn := func() (string, error) {
+	fn := func() (LLMResponse, error) {
 		opts := gRPCPredictOpts(c, loader.ModelPath)
 		opts.Prompt = s
+
+		tokenUsage := TokenUsage{}
+
+		// check the per-model feature flag for usage, since tokenCallback may have a cost.
+		// Defaults to off as for now it is still experimental
+		if c.FeatureFlag.Enabled("usage") {
+			userTokenCallback := tokenCallback
+			if userTokenCallback == nil {
+				userTokenCallback = func(token string, usage TokenUsage) bool {
+					return true
+				}
+			}
+
+			promptInfo, pErr := inferenceModel.TokenizeString(ctx, opts)
+			if pErr == nil && promptInfo.Length > 0 {
+				tokenUsage.Prompt = int(promptInfo.Length)
+			}
+
+			tokenCallback = func(token string, usage TokenUsage) bool {
+				tokenUsage.Completion++
+				return userTokenCallback(token, tokenUsage)
+			}
+		}
+
 		if tokenCallback != nil {
 			ss := ""
-			err := inferenceModel.PredictStream(ctx, opts, func(s []byte) {
-				tokenCallback(string(s))
-				ss += string(s)
+
+			var partialRune []byte
+			err := inferenceModel.PredictStream(ctx, opts, func(chars []byte) {
+				partialRune = append(partialRune, chars...)
+
+				for len(partialRune) > 0 {
+					r, size := utf8.DecodeRune(partialRune)
+					if r == utf8.RuneError {
+						// incomplete rune, wait for more bytes
+						break
+					}
+
+					tokenCallback(string(r), tokenUsage)
+					ss += string(r)
+
+					partialRune = partialRune[size:]
+				}
 			})
-			return ss, err
+			return LLMResponse{
+				Response: ss,
+				Usage:    tokenUsage,
+			}, err
 		} else {
+			// TODO: Is the chicken bit the only way to get here? is that acceptable?
 			reply, err := inferenceModel.Predict(ctx, opts)
 			if err != nil {
-				return "", err
+				return LLMResponse{}, err
 			}
-			return string(reply.Message), err
+			return LLMResponse{
+				Response: string(reply.Message),
+				Usage:    tokenUsage,
+			}, err
 		}
 	}
 
-	return func() (string, error) {
-		// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
-		mutexMap.Lock()
-		l, ok := mutexes[modelFile]
-		if !ok {
-			m := &sync.Mutex{}
-			mutexes[modelFile] = m
-			l = m
-		}
-		mutexMap.Unlock()
-		l.Lock()
-		defer l.Unlock()
-
-		return fn()
-	}, nil
+	return fn, nil
 }
 
 var cutstrings map[string]*regexp.Regexp = make(map[string]*regexp.Regexp)

api/backend/lock.go

@@ -1,22 +0,0 @@
-package backend
-
-import "sync"
-
-// mutex still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
-var mutexMap sync.Mutex
-var mutexes map[string]*sync.Mutex = make(map[string]*sync.Mutex)
-
-func Lock(s string) *sync.Mutex {
-	// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
-	mutexMap.Lock()
-	l, ok := mutexes[s]
-	if !ok {
-		m := &sync.Mutex{}
-		mutexes[s] = m
-		l = m
-	}
-	mutexMap.Unlock()
-	l.Lock()
-
-	return l
-}

api/backend/options.go

@@ -5,21 +5,49 @@ import (
 	"path/filepath"
 
 	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
+	model "github.com/go-skynet/LocalAI/pkg/model"
 
 	config "github.com/go-skynet/LocalAI/api/config"
+	"github.com/go-skynet/LocalAI/api/options"
 )
 
+func modelOpts(c config.Config, o *options.Option, opts []model.Option) []model.Option {
+	if o.SingleBackend {
+		opts = append(opts, model.WithSingleActiveBackend())
+	}
+
+	if c.GRPC.Attempts != 0 {
+		opts = append(opts, model.WithGRPCAttempts(c.GRPC.Attempts))
+	}
+
+	if c.GRPC.AttemptsSleepTime != 0 {
+		opts = append(opts, model.WithGRPCAttemptsDelay(c.GRPC.AttemptsSleepTime))
+	}
+
+	for k, v := range o.ExternalGRPCBackends {
+		opts = append(opts, model.WithExternalBackend(k, v))
+	}
+
+	return opts
+}
+
 func gRPCModelOpts(c config.Config) *pb.ModelOptions {
 	b := 512
 	if c.Batch != 0 {
 		b = c.Batch
 	}
-	return &pb.ModelOptions{
-		ContextSize: int32(c.ContextSize),
-		Seed:        int32(c.Seed),
-		NBatch:      int32(b),
-		NGQA:        c.NGQA,
 
+	return &pb.ModelOptions{
+		ContextSize:   int32(c.ContextSize),
+		Seed:          int32(c.Seed),
+		NBatch:        int32(b),
+		NoMulMatQ:     c.NoMulMatQ,
+		DraftModel:    c.DraftModel,
+		AudioPath:     c.VallE.AudioPath,
+		Quantization:  c.Quantization,
+		LoraAdapter:   c.LoraAdapter,
+		LoraBase:      c.LoraBase,
+		NGQA:          c.NGQA,
 		RMSNormEps:    c.RMSNormEps,
 		F16Memory:     c.F16,
 		MLock:         c.MMlock,
@@ -38,6 +66,8 @@ func gRPCModelOpts(c config.Config) *pb.ModelOptions {
 		Device:           c.AutoGPTQ.Device,
 		UseTriton:        c.AutoGPTQ.Triton,
 		UseFastTokenizer: c.AutoGPTQ.UseFastTokenizer,
+		// RWKV
+		Tokenizer: c.Tokenizer,
 	}
 }
 
@@ -51,6 +81,7 @@ func gRPCPredictOpts(c config.Config, modelPath string) *pb.PredictOptions {
 	return &pb.PredictOptions{
 		Temperature:         float32(c.Temperature),
 		TopP:                float32(c.TopP),
+		NDraft:              c.NDraft,
 		TopK:                int32(c.TopK),
 		Tokens:              int32(c.Maxtokens),
 		Threads:             int32(c.Threads),

api/backend/transcript.go

@@ -5,25 +5,22 @@ import (
 	"fmt"
 
 	config "github.com/go-skynet/LocalAI/api/config"
+	"github.com/go-skynet/LocalAI/api/schema"
 
 	"github.com/go-skynet/LocalAI/api/options"
 	"github.com/go-skynet/LocalAI/pkg/grpc/proto"
-	"github.com/go-skynet/LocalAI/pkg/grpc/whisper/api"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 )
 
-func ModelTranscription(audio, language string, loader *model.ModelLoader, c config.Config, o *options.Option) (*api.Result, error) {
-	opts := []model.Option{
+func ModelTranscription(audio, language string, loader *model.ModelLoader, c config.Config, o *options.Option) (*schema.Result, error) {
+
+	opts := modelOpts(c, o, []model.Option{
 		model.WithBackendString(model.WhisperBackend),
 		model.WithModel(c.Model),
 		model.WithContext(o.Context),
 		model.WithThreads(uint32(c.Threads)),
 		model.WithAssetDir(o.AssetsDestination),
-	}
-
-	for k, v := range o.ExternalGRPCBackends {
-		opts = append(opts, model.WithExternalBackend(k, v))
-	}
+	})
 
 	whisperModel, err := o.Loader.BackendLoader(opts...)
 	if err != nil {

api/backend/tts.go

@@ -6,6 +6,7 @@ import (
 	"os"
 	"path/filepath"
 
+	api_config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
 	"github.com/go-skynet/LocalAI/pkg/grpc/proto"
 	model "github.com/go-skynet/LocalAI/pkg/model"
@@ -33,17 +34,12 @@ func ModelTTS(backend, text, modelFile string, loader *model.ModelLoader, o *opt
 	if bb == "" {
 		bb = model.PiperBackend
 	}
-	opts := []model.Option{
+	opts := modelOpts(api_config.Config{}, o, []model.Option{
 		model.WithBackendString(bb),
 		model.WithModel(modelFile),
 		model.WithContext(o.Context),
 		model.WithAssetDir(o.AssetsDestination),
-	}
-
-	for k, v := range o.ExternalGRPCBackends {
-		opts = append(opts, model.WithExternalBackend(k, v))
-	}
-
+	})
 	piperModel, err := o.Loader.BackendLoader(opts...)
 	if err != nil {
 		return "", nil, err

api/config/config.go

@@ -29,6 +29,7 @@ type Config struct {
 
 	FunctionsConfig Functions `yaml:"function"`
 
+	FeatureFlag FeatureFlag `yaml:"feature_flags"` // Feature Flag registry. We move fast, and features may break on a per model/backend basis. Registry for (usually temporary) flags that indicate aborting something early.
 	// LLM configs (GPT4ALL, Llama.cpp, ...)
 	LLMConfig `yaml:",inline"`
 
@@ -39,12 +40,40 @@ type Config struct {
 	Diffusers Diffusers `yaml:"diffusers"`
 
 	Step int `yaml:"step"`
+
+	// GRPC Options
+	GRPC GRPC `yaml:"grpc"`
+
+	// Vall-e-x
+	VallE VallE `yaml:"vall-e"`
+}
+
+type VallE struct {
+	AudioPath string `yaml:"audio_path"`
+}
+
+type FeatureFlag map[string]*bool
+
+func (ff FeatureFlag) Enabled(s string) bool {
+	v, exist := ff[s]
+	return exist && v != nil && *v
+}
+
+type GRPC struct {
+	Attempts          int `yaml:"attempts"`
+	AttemptsSleepTime int `yaml:"attempts_sleep_time"`
 }
 
 type Diffusers struct {
-	PipelineType  string `yaml:"pipeline_type"`
-	SchedulerType string `yaml:"scheduler_type"`
-	CUDA          bool   `yaml:"cuda"`
+	PipelineType     string  `yaml:"pipeline_type"`
+	SchedulerType    string  `yaml:"scheduler_type"`
+	CUDA             bool    `yaml:"cuda"`
+	EnableParameters string  `yaml:"enable_parameters"` // A list of comma separated parameters to specify
+	CFGScale         float32 `yaml:"cfg_scale"`         // Classifier-Free Guidance Scale
+	IMG2IMG          bool    `yaml:"img2img"`           // Image to Image Diffuser
+	ClipSkip         int     `yaml:"clip_skip"`         // Skip every N frames
+	ClipModel        string  `yaml:"clip_model"`        // Clip model to use
+	ClipSubFolder    string  `yaml:"clip_subfolder"`    // Subfolder to use for clip model
 }
 
 type LLMConfig struct {
@@ -69,6 +98,12 @@ type LLMConfig struct {
 	TrimSpace       []string `yaml:"trimspace"`
 	ContextSize     int      `yaml:"context_size"`
 	NUMA            bool     `yaml:"numa"`
+	LoraAdapter     string   `yaml:"lora_adapter"`
+	LoraBase        string   `yaml:"lora_base"`
+	NoMulMatQ       bool     `yaml:"no_mulmatq"`
+	DraftModel      string   `yaml:"draft_model"`
+	NDraft          int32    `yaml:"n_draft"`
+	Quantization    string   `yaml:"quantization"`
 }
 
 type AutoGPTQ struct {

api/config/prediction.go

@@ -41,4 +41,10 @@ type PredictionOptions struct {
 	NegativePromptScale float32 `json:"negative_prompt_scale" yaml:"negative_prompt_scale"`
 	// AutoGPTQ
 	UseFastTokenizer bool `json:"use_fast_tokenizer" yaml:"use_fast_tokenizer"`
+
+	// Diffusers
+	ClipSkip int `json:"clip_skip" yaml:"clip_skip"`
+
+	// RWKV (?)
+	Tokenizer string `json:"tokenizer" yaml:"tokenizer"`
 }

api/localai/backend_monitor.go

@@ -0,0 +1,163 @@
+package localai
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	config "github.com/go-skynet/LocalAI/api/config"
+	"github.com/go-skynet/LocalAI/pkg/grpc/proto"
+
+	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/gofiber/fiber/v2"
+	"github.com/rs/zerolog/log"
+
+	gopsutil "github.com/shirou/gopsutil/v3/process"
+)
+
+type BackendMonitorRequest struct {
+	Model string `json:"model" yaml:"model"`
+}
+
+type BackendMonitorResponse struct {
+	MemoryInfo    *gopsutil.MemoryInfoStat
+	MemoryPercent float32
+	CPUPercent    float64
+}
+
+type BackendMonitor struct {
+	configLoader *config.ConfigLoader
+	options      *options.Option // Taking options in case we need to inspect ExternalGRPCBackends, though that's out of scope for now, hence the name.
+}
+
+func NewBackendMonitor(configLoader *config.ConfigLoader, options *options.Option) BackendMonitor {
+	return BackendMonitor{
+		configLoader: configLoader,
+		options:      options,
+	}
+}
+
+func (bm *BackendMonitor) SampleLocalBackendProcess(model string) (*BackendMonitorResponse, error) {
+	config, exists := bm.configLoader.GetConfig(model)
+	var backend string
+	if exists {
+		backend = config.Model
+	} else {
+		// Last ditch effort: use it raw, see if a backend happens to match.
+		backend = model
+	}
+
+	if !strings.HasSuffix(backend, ".bin") {
+		backend = fmt.Sprintf("%s.bin", backend)
+	}
+
+	pid, err := bm.options.Loader.GetGRPCPID(backend)
+
+	if err != nil {
+		log.Error().Msgf("model %s : failed to find pid %+v", model, err)
+		return nil, err
+	}
+
+	// Name is slightly frightening but this does _not_ create a new process, rather it looks up an existing process by PID.
+	backendProcess, err := gopsutil.NewProcess(int32(pid))
+
+	if err != nil {
+		log.Error().Msgf("model %s [PID %d] : error getting process info %+v", model, pid, err)
+		return nil, err
+	}
+
+	memInfo, err := backendProcess.MemoryInfo()
+
+	if err != nil {
+		log.Error().Msgf("model %s [PID %d] : error getting memory info %+v", model, pid, err)
+		return nil, err
+	}
+
+	memPercent, err := backendProcess.MemoryPercent()
+	if err != nil {
+		log.Error().Msgf("model %s [PID %d] : error getting memory percent %+v", model, pid, err)
+		return nil, err
+	}
+
+	cpuPercent, err := backendProcess.CPUPercent()
+	if err != nil {
+		log.Error().Msgf("model %s [PID %d] : error getting cpu percent %+v", model, pid, err)
+		return nil, err
+	}
+
+	return &BackendMonitorResponse{
+		MemoryInfo:    memInfo,
+		MemoryPercent: memPercent,
+		CPUPercent:    cpuPercent,
+	}, nil
+}
+
+func (bm BackendMonitor) getModelLoaderIDFromCtx(c *fiber.Ctx) (string, error) {
+	input := new(BackendMonitorRequest)
+	// Get input data from the request body
+	if err := c.BodyParser(input); err != nil {
+		return "", err
+	}
+
+	config, exists := bm.configLoader.GetConfig(input.Model)
+	var backendId string
+	if exists {
+		backendId = config.Model
+	} else {
+		// Last ditch effort: use it raw, see if a backend happens to match.
+		backendId = input.Model
+	}
+
+	if !strings.HasSuffix(backendId, ".bin") {
+		backendId = fmt.Sprintf("%s.bin", backendId)
+	}
+
+	return backendId, nil
+}
+
+func BackendMonitorEndpoint(bm BackendMonitor) func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+
+		backendId, err := bm.getModelLoaderIDFromCtx(c)
+		if err != nil {
+			return err
+		}
+
+		client := bm.options.Loader.CheckIsLoaded(backendId)
+
+		if client == nil {
+			return fmt.Errorf("backend %s is not currently loaded", backendId)
+		}
+
+		status, rpcErr := client.Status(context.TODO())
+		if rpcErr != nil {
+			log.Warn().Msgf("backend %s experienced an error retrieving status info: %s", backendId, rpcErr.Error())
+			val, slbErr := bm.SampleLocalBackendProcess(backendId)
+			if slbErr != nil {
+				return fmt.Errorf("backend %s experienced an error retrieving status info via rpc: %s, then failed local node process sample: %s", backendId, rpcErr.Error(), slbErr.Error())
+			}
+			return c.JSON(proto.StatusResponse{
+				State: proto.StatusResponse_ERROR,
+				Memory: &proto.MemoryUsageData{
+					Total: val.MemoryInfo.VMS,
+					Breakdown: map[string]uint64{
+						"gopsutil-RSS": val.MemoryInfo.RSS,
+					},
+				},
+			})
+		}
+
+		return c.JSON(status)
+	}
+}
+
+func BackendShutdownEndpoint(bm BackendMonitor) func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		backendId, err := bm.getModelLoaderIDFromCtx(c)
+		if err != nil {
+			return err
+		}
+
+		return bm.options.Loader.ShutdownModel(backendId)
+	}
+}

api/localai/gallery.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"os"
+	"slices"
 	"strings"
 	"sync"
 
@@ -27,6 +28,7 @@ type galleryOp struct {
 }
 
 type galleryOpStatus struct {
+	FileName           string  `json:"file_name"`
 	Error              error   `json:"error"`
 	Processed          bool    `json:"processed"`
 	Message            string  `json:"message"`
@@ -50,7 +52,6 @@ func NewGalleryService(modelPath string) *galleryApplier {
 	}
 }
 
-// prepareModel applies a
 func prepareModel(modelPath string, req gallery.GalleryModel, cm *config.ConfigLoader, downloadStatus func(string, string, string, float64)) error {
 
 	config, err := gallery.GetGalleryConfigFromURL(req.URL)
@@ -76,6 +77,13 @@ func (g *galleryApplier) getStatus(s string) *galleryOpStatus {
 	return g.statuses[s]
 }
 
+func (g *galleryApplier) getAllStatus() map[string]*galleryOpStatus {
+	g.Lock()
+	defer g.Unlock()
+
+	return g.statuses
+}
+
 func (g *galleryApplier) Start(c context.Context, cm *config.ConfigLoader) {
 	go func() {
 		for {
@@ -94,7 +102,7 @@ func (g *galleryApplier) Start(c context.Context, cm *config.ConfigLoader) {
 
 				// displayDownload displays the download progress
 				progressCallback := func(fileName string, current string, total string, percentage float64) {
-					g.updateStatus(op.id, &galleryOpStatus{Message: "processing", Progress: percentage, TotalFileSize: total, DownloadedFileSize: current})
+					g.updateStatus(op.id, &galleryOpStatus{Message: "processing", FileName: fileName, Progress: percentage, TotalFileSize: total, DownloadedFileSize: current})
 					utils.DisplayDownloadFunction(fileName, current, total, percentage)
 				}
 
@@ -176,18 +184,12 @@ func ApplyGalleryFromString(modelPath, s string, cm *config.ConfigLoader, galler
 	return processRequests(modelPath, s, cm, galleries, requests)
 }
 
-/// Endpoints
+/// Endpoint Service
 
-func GetOpStatusEndpoint(g *galleryApplier) func(c *fiber.Ctx) error {
-	return func(c *fiber.Ctx) error {
-
-		status := g.getStatus(c.Params("uuid"))
-		if status == nil {
-			return fmt.Errorf("could not find any status for ID")
-		}
-
-		return c.JSON(status)
-	}
+type ModelGalleryService struct {
+	galleries      []gallery.Gallery
+	modelPath      string
+	galleryApplier *galleryApplier
 }
 
 type GalleryModel struct {
@@ -195,7 +197,31 @@ type GalleryModel struct {
 	gallery.GalleryModel
 }
 
-func ApplyModelGalleryEndpoint(modelPath string, cm *config.ConfigLoader, g chan galleryOp, galleries []gallery.Gallery) func(c *fiber.Ctx) error {
+func CreateModelGalleryService(galleries []gallery.Gallery, modelPath string, galleryApplier *galleryApplier) ModelGalleryService {
+	return ModelGalleryService{
+		galleries:      galleries,
+		modelPath:      modelPath,
+		galleryApplier: galleryApplier,
+	}
+}
+
+func (mgs *ModelGalleryService) GetOpStatusEndpoint() func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		status := mgs.galleryApplier.getStatus(c.Params("uuid"))
+		if status == nil {
+			return fmt.Errorf("could not find any status for ID")
+		}
+		return c.JSON(status)
+	}
+}
+
+func (mgs *ModelGalleryService) GetAllStatusEndpoint() func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		return c.JSON(mgs.galleryApplier.getAllStatus())
+	}
+}
+
+func (mgs *ModelGalleryService) ApplyModelGalleryEndpoint() func(c *fiber.Ctx) error {
 	return func(c *fiber.Ctx) error {
 		input := new(GalleryModel)
 		// Get input data from the request body
@@ -207,11 +233,11 @@ func ApplyModelGalleryEndpoint(modelPath string, cm *config.ConfigLoader, g chan
 		if err != nil {
 			return err
 		}
-		g <- galleryOp{
+		mgs.galleryApplier.C <- galleryOp{
 			req:         input.GalleryModel,
 			id:          uuid.String(),
 			galleryName: input.ID,
-			galleries:   galleries,
+			galleries:   mgs.galleries,
 		}
 		return c.JSON(struct {
 			ID        string `json:"uuid"`
@@ -220,11 +246,11 @@ func ApplyModelGalleryEndpoint(modelPath string, cm *config.ConfigLoader, g chan
 	}
 }
 
-func ListModelFromGalleryEndpoint(galleries []gallery.Gallery, basePath string) func(c *fiber.Ctx) error {
+func (mgs *ModelGalleryService) ListModelFromGalleryEndpoint() func(c *fiber.Ctx) error {
 	return func(c *fiber.Ctx) error {
-		log.Debug().Msgf("Listing models from galleries: %+v", galleries)
+		log.Debug().Msgf("Listing models from galleries: %+v", mgs.galleries)
 
-		models, err := gallery.AvailableGalleryModels(galleries, basePath)
+		models, err := gallery.AvailableGalleryModels(mgs.galleries, mgs.modelPath)
 		if err != nil {
 			return err
 		}
@@ -239,3 +265,56 @@ func ListModelFromGalleryEndpoint(galleries []gallery.Gallery, basePath string)
 		return c.Send(dat)
 	}
 }
+
+// NOTE: This is different (and much simpler!) than above! This JUST lists the model galleries that have been loaded, not their contents!
+func (mgs *ModelGalleryService) ListModelGalleriesEndpoint() func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		log.Debug().Msgf("Listing model galleries %+v", mgs.galleries)
+		dat, err := json.Marshal(mgs.galleries)
+		if err != nil {
+			return err
+		}
+		return c.Send(dat)
+	}
+}
+
+func (mgs *ModelGalleryService) AddModelGalleryEndpoint() func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		input := new(gallery.Gallery)
+		// Get input data from the request body
+		if err := c.BodyParser(input); err != nil {
+			return err
+		}
+		if slices.ContainsFunc(mgs.galleries, func(gallery gallery.Gallery) bool {
+			return gallery.Name == input.Name
+		}) {
+			return fmt.Errorf("%s already exists", input.Name)
+		}
+		dat, err := json.Marshal(mgs.galleries)
+		if err != nil {
+			return err
+		}
+		log.Debug().Msgf("Adding %+v to gallery list", *input)
+		mgs.galleries = append(mgs.galleries, *input)
+		return c.Send(dat)
+	}
+}
+
+func (mgs *ModelGalleryService) RemoveModelGalleryEndpoint() func(c *fiber.Ctx) error {
+	return func(c *fiber.Ctx) error {
+		input := new(gallery.Gallery)
+		// Get input data from the request body
+		if err := c.BodyParser(input); err != nil {
+			return err
+		}
+		if !slices.ContainsFunc(mgs.galleries, func(gallery gallery.Gallery) bool {
+			return gallery.Name == input.Name
+		}) {
+			return fmt.Errorf("%s is not currently registered", input.Name)
+		}
+		mgs.galleries = slices.DeleteFunc(mgs.galleries, func(gallery gallery.Gallery) bool {
+			return gallery.Name == input.Name
+		})
+		return c.Send(nil)
+	}
+}

api/openai/chat.go

@@ -10,6 +10,7 @@ import (
 	"github.com/go-skynet/LocalAI/api/backend"
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	"github.com/go-skynet/LocalAI/pkg/grammar"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 	"github.com/go-skynet/LocalAI/pkg/utils"
@@ -21,19 +22,24 @@ import (
 func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx) error {
 	emptyMessage := ""
 
-	process := func(s string, req *OpenAIRequest, config *config.Config, loader *model.ModelLoader, responses chan OpenAIResponse) {
-		initialMessage := OpenAIResponse{
+	process := func(s string, req *schema.OpenAIRequest, config *config.Config, loader *model.ModelLoader, responses chan schema.OpenAIResponse) {
+		initialMessage := schema.OpenAIResponse{
 			Model:   req.Model, // we have to return what the user sent here, due to OpenAI spec.
-			Choices: []Choice{{Delta: &Message{Role: "assistant", Content: &emptyMessage}}},
+			Choices: []schema.Choice{{Delta: &schema.Message{Role: "assistant", Content: &emptyMessage}}},
 			Object:  "chat.completion.chunk",
 		}
 		responses <- initialMessage
 
-		ComputeChoices(req, s, config, o, loader, func(s string, c *[]Choice) {}, func(s string) bool {
-			resp := OpenAIResponse{
+		ComputeChoices(req, s, config, o, loader, func(s string, c *[]schema.Choice) {}, func(s string, usage backend.TokenUsage) bool {
+			resp := schema.OpenAIResponse{
 				Model:   req.Model, // we have to return what the user sent here, due to OpenAI spec.
-				Choices: []Choice{{Delta: &Message{Content: &s}, Index: 0}},
+				Choices: []schema.Choice{{Delta: &schema.Message{Content: &s}, Index: 0}},
 				Object:  "chat.completion.chunk",
+				Usage: schema.OpenAIUsage{
+					PromptTokens:     usage.Prompt,
+					CompletionTokens: usage.Completion,
+					TotalTokens:      usage.Prompt + usage.Completion,
+				},
 			}
 
 			responses <- resp
@@ -231,17 +237,19 @@ func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 		}
 
 		if toStream {
-			responses := make(chan OpenAIResponse)
+			responses := make(chan schema.OpenAIResponse)
 
 			go process(predInput, input, config, o.Loader, responses)
 
 			c.Context().SetBodyStreamWriter(fasthttp.StreamWriter(func(w *bufio.Writer) {
 
+				usage := &schema.OpenAIUsage{}
+
 				for ev := range responses {
+					usage = &ev.Usage // Copy a pointer to the latest usage chunk so that the stop message can reference it
 					var buf bytes.Buffer
 					enc := json.NewEncoder(&buf)
 					enc.Encode(ev)
-
 					log.Debug().Msgf("Sending chunk: %s", buf.String())
 					_, err := fmt.Fprintf(w, "data: %v\n", buf.String())
 					if err != nil {
@@ -252,15 +260,16 @@ func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 					w.Flush()
 				}
 
-				resp := &OpenAIResponse{
+				resp := &schema.OpenAIResponse{
 					Model: input.Model, // we have to return what the user sent here, due to OpenAI spec.
-					Choices: []Choice{
+					Choices: []schema.Choice{
 						{
 							FinishReason: "stop",
 							Index:        0,
-							Delta:        &Message{Content: &emptyMessage},
+							Delta:        &schema.Message{Content: &emptyMessage},
 						}},
 					Object: "chat.completion.chunk",
+					Usage:  *usage,
 				}
 				respData, _ := json.Marshal(resp)
 
@@ -271,7 +280,7 @@ func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 			return nil
 		}
 
-		result, err := ComputeChoices(input, predInput, config, o, o.Loader, func(s string, c *[]Choice) {
+		result, tokenUsage, err := ComputeChoices(input, predInput, config, o, o.Loader, func(s string, c *[]schema.Choice) {
 			if processFunctions {
 				// As we have to change the result before processing, we can't stream the answer (yet?)
 				ss := map[string]interface{}{}
@@ -305,7 +314,7 @@ func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 								message = backend.Finetune(*config, predInput, message)
 								log.Debug().Msgf("Reply received from LLM(finetuned): %s", message)
 
-								*c = append(*c, Choice{Message: &Message{Role: "assistant", Content: &message}})
+								*c = append(*c, schema.Choice{Message: &schema.Message{Role: "assistant", Content: &message}})
 								return
 							}
 						}
@@ -327,28 +336,33 @@ func ChatEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 						return
 					}
 
-					prediction = backend.Finetune(*config, predInput, prediction)
-					*c = append(*c, Choice{Message: &Message{Role: "assistant", Content: &prediction}})
+					fineTunedResponse := backend.Finetune(*config, predInput, prediction.Response)
+					*c = append(*c, schema.Choice{Message: &schema.Message{Role: "assistant", Content: &fineTunedResponse}})
 				} else {
 					// otherwise reply with the function call
-					*c = append(*c, Choice{
+					*c = append(*c, schema.Choice{
 						FinishReason: "function_call",
-						Message:      &Message{Role: "assistant", FunctionCall: ss},
+						Message:      &schema.Message{Role: "assistant", FunctionCall: ss},
 					})
 				}
 
 				return
 			}
-			*c = append(*c, Choice{FinishReason: "stop", Index: 0, Message: &Message{Role: "assistant", Content: &s}})
+			*c = append(*c, schema.Choice{FinishReason: "stop", Index: 0, Message: &schema.Message{Role: "assistant", Content: &s}})
 		}, nil)
 		if err != nil {
 			return err
 		}
 
-		resp := &OpenAIResponse{
+		resp := &schema.OpenAIResponse{
 			Model:   input.Model, // we have to return what the user sent here, due to OpenAI spec.
 			Choices: result,
 			Object:  "chat.completion",
+			Usage: schema.OpenAIUsage{
+				PromptTokens:     tokenUsage.Prompt,
+				CompletionTokens: tokenUsage.Completion,
+				TotalTokens:      tokenUsage.Prompt + tokenUsage.Completion,
+			},
 		}
 		respData, _ := json.Marshal(resp)
 		log.Debug().Msgf("Response: %s", respData)

api/openai/completion.go

@@ -7,8 +7,10 @@ import (
 	"errors"
 	"fmt"
 
+	"github.com/go-skynet/LocalAI/api/backend"
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 	"github.com/gofiber/fiber/v2"
 	"github.com/rs/zerolog/log"
@@ -17,17 +19,22 @@ import (
 
 // https://platform.openai.com/docs/api-reference/completions
 func CompletionEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx) error {
-	process := func(s string, req *OpenAIRequest, config *config.Config, loader *model.ModelLoader, responses chan OpenAIResponse) {
-		ComputeChoices(req, s, config, o, loader, func(s string, c *[]Choice) {}, func(s string) bool {
-			resp := OpenAIResponse{
+	process := func(s string, req *schema.OpenAIRequest, config *config.Config, loader *model.ModelLoader, responses chan schema.OpenAIResponse) {
+		ComputeChoices(req, s, config, o, loader, func(s string, c *[]schema.Choice) {}, func(s string, usage backend.TokenUsage) bool {
+			resp := schema.OpenAIResponse{
 				Model: req.Model, // we have to return what the user sent here, due to OpenAI spec.
-				Choices: []Choice{
+				Choices: []schema.Choice{
 					{
 						Index: 0,
 						Text:  s,
 					},
 				},
 				Object: "text_completion",
+				Usage: schema.OpenAIUsage{
+					PromptTokens:     usage.Prompt,
+					CompletionTokens: usage.Completion,
+					TotalTokens:      usage.Prompt + usage.Completion,
+				},
 			}
 			log.Debug().Msgf("Sending goroutine: %s", s)
 
@@ -84,7 +91,7 @@ func CompletionEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 				log.Debug().Msgf("Template found, input modified to: %s", predInput)
 			}
 
-			responses := make(chan OpenAIResponse)
+			responses := make(chan schema.OpenAIResponse)
 
 			go process(predInput, input, config, o.Loader, responses)
 
@@ -100,9 +107,9 @@ func CompletionEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 					w.Flush()
 				}
 
-				resp := &OpenAIResponse{
+				resp := &schema.OpenAIResponse{
 					Model: input.Model, // we have to return what the user sent here, due to OpenAI spec.
-					Choices: []Choice{
+					Choices: []schema.Choice{
 						{
 							Index:        0,
 							FinishReason: "stop",
@@ -119,7 +126,10 @@ func CompletionEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 			return nil
 		}
 
-		var result []Choice
+		var result []schema.Choice
+
+		totalTokenUsage := backend.TokenUsage{}
+
 		for k, i := range config.PromptStrings {
 			// A model can have a "file.bin.tmpl" file associated with a prompt template prefix
 			templatedInput, err := o.Loader.EvaluateTemplateForPrompt(model.CompletionPromptTemplate, templateFile, model.PromptTemplateData{
@@ -131,20 +141,29 @@ func CompletionEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 				log.Debug().Msgf("Template found, input modified to: %s", i)
 			}
 
-			r, err := ComputeChoices(input, i, config, o, o.Loader, func(s string, c *[]Choice) {
-				*c = append(*c, Choice{Text: s, FinishReason: "stop", Index: k})
-			}, nil)
+			r, tokenUsage, err := ComputeChoices(
+				input, i, config, o, o.Loader, func(s string, c *[]schema.Choice) {
+					*c = append(*c, schema.Choice{Text: s, FinishReason: "stop", Index: k})
+				}, nil)
 			if err != nil {
 				return err
 			}
 
+			totalTokenUsage.Prompt += tokenUsage.Prompt
+			totalTokenUsage.Completion += tokenUsage.Completion
+
 			result = append(result, r...)
 		}
 
-		resp := &OpenAIResponse{
+		resp := &schema.OpenAIResponse{
 			Model:   input.Model, // we have to return what the user sent here, due to OpenAI spec.
 			Choices: result,
 			Object:  "text_completion",
+			Usage: schema.OpenAIUsage{
+				PromptTokens:     totalTokenUsage.Prompt,
+				CompletionTokens: totalTokenUsage.Completion,
+				TotalTokens:      totalTokenUsage.Prompt + totalTokenUsage.Completion,
+			},
 		}
 
 		jsonResult, _ := json.Marshal(resp)

api/openai/edit.go

@@ -4,10 +4,13 @@ import (
 	"encoding/json"
 	"fmt"
 
+	"github.com/go-skynet/LocalAI/api/backend"
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 	"github.com/gofiber/fiber/v2"
+
 	"github.com/rs/zerolog/log"
 )
 
@@ -31,7 +34,9 @@ func EditEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 			templateFile = config.TemplateConfig.Edit
 		}
 
-		var result []Choice
+		var result []schema.Choice
+		totalTokenUsage := backend.TokenUsage{}
+
 		for _, i := range config.InputStrings {
 			// A model can have a "file.bin.tmpl" file associated with a prompt template prefix
 			templatedInput, err := o.Loader.EvaluateTemplateForPrompt(model.EditPromptTemplate, templateFile, model.PromptTemplateData{
@@ -44,20 +49,28 @@ func EditEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx)
 				log.Debug().Msgf("Template found, input modified to: %s", i)
 			}
 
-			r, err := ComputeChoices(input, i, config, o, o.Loader, func(s string, c *[]Choice) {
-				*c = append(*c, Choice{Text: s})
+			r, tokenUsage, err := ComputeChoices(input, i, config, o, o.Loader, func(s string, c *[]schema.Choice) {
+				*c = append(*c, schema.Choice{Text: s})
 			}, nil)
 			if err != nil {
 				return err
 			}
 
+			totalTokenUsage.Prompt += tokenUsage.Prompt
+			totalTokenUsage.Completion += tokenUsage.Completion
+
 			result = append(result, r...)
 		}
 
-		resp := &OpenAIResponse{
+		resp := &schema.OpenAIResponse{
 			Model:   input.Model, // we have to return what the user sent here, due to OpenAI spec.
 			Choices: result,
 			Object:  "edit",
+			Usage: schema.OpenAIUsage{
+				PromptTokens:     totalTokenUsage.Prompt,
+				CompletionTokens: totalTokenUsage.Completion,
+				TotalTokens:      totalTokenUsage.Prompt + totalTokenUsage.Completion,
+			},
 		}
 
 		jsonResult, _ := json.Marshal(resp)

api/openai/embeddings.go

@@ -6,6 +6,8 @@ import (
 
 	"github.com/go-skynet/LocalAI/api/backend"
 	config "github.com/go-skynet/LocalAI/api/config"
+	"github.com/go-skynet/LocalAI/api/schema"
+
 	"github.com/go-skynet/LocalAI/api/options"
 	"github.com/gofiber/fiber/v2"
 	"github.com/rs/zerolog/log"
@@ -25,7 +27,7 @@ func EmbeddingsEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 		}
 
 		log.Debug().Msgf("Parameter Config: %+v", config)
-		items := []Item{}
+		items := []schema.Item{}
 
 		for i, s := range config.InputToken {
 			// get the model function to call for the result
@@ -38,7 +40,7 @@ func EmbeddingsEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 			if err != nil {
 				return err
 			}
-			items = append(items, Item{Embedding: embeddings, Index: i, Object: "embedding"})
+			items = append(items, schema.Item{Embedding: embeddings, Index: i, Object: "embedding"})
 		}
 
 		for i, s := range config.InputStrings {
@@ -52,10 +54,10 @@ func EmbeddingsEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fibe
 			if err != nil {
 				return err
 			}
-			items = append(items, Item{Embedding: embeddings, Index: i, Object: "embedding"})
+			items = append(items, schema.Item{Embedding: embeddings, Index: i, Object: "embedding"})
 		}
 
-		resp := &OpenAIResponse{
+		resp := &schema.OpenAIResponse{
 			Model:  input.Model, // we have to return what the user sent here, due to OpenAI spec.
 			Data:   items,
 			Object: "list",

api/openai/image.go

@@ -1,9 +1,11 @@
 package openai
 
 import (
+	"bufio"
 	"encoding/base64"
 	"encoding/json"
 	"fmt"
+	"github.com/go-skynet/LocalAI/api/schema"
 	"os"
 	"path/filepath"
 	"strconv"
@@ -49,6 +51,31 @@ func ImageEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx
 			return fmt.Errorf("failed reading parameters from request:%w", err)
 		}
 
+		src := ""
+		if input.File != "" {
+			//base 64 decode the file and write it somewhere
+			// that we will cleanup
+			decoded, err := base64.StdEncoding.DecodeString(input.File)
+			if err != nil {
+				return err
+			}
+			// Create a temporary file
+			outputFile, err := os.CreateTemp(o.ImageDir, "b64")
+			if err != nil {
+				return err
+			}
+			// write the base64 result
+			writer := bufio.NewWriter(outputFile)
+			_, err = writer.Write(decoded)
+			if err != nil {
+				outputFile.Close()
+				return err
+			}
+			outputFile.Close()
+			src = outputFile.Name()
+			defer os.RemoveAll(src)
+		}
+
 		log.Debug().Msgf("Parameter Config: %+v", config)
 
 		// XXX: Only stablediffusion is supported for now
@@ -73,8 +100,8 @@ func ImageEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx
 		if input.ResponseFormat == "b64_json" {
 			b64JSON = true
 		}
-
-		var result []Item
+		// src and clip_skip
+		var result []schema.Item
 		for _, i := range config.PromptStrings {
 			n := input.N
 			if input.N == 0 {
@@ -121,7 +148,7 @@ func ImageEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx
 
 				baseURL := c.BaseURL()
 
-				fn, err := backend.ImageGeneration(height, width, mode, step, input.Seed, positive_prompt, negative_prompt, output, o.Loader, *config, o)
+				fn, err := backend.ImageGeneration(height, width, mode, step, input.Seed, positive_prompt, negative_prompt, src, output, o.Loader, *config, o)
 				if err != nil {
 					return err
 				}
@@ -129,7 +156,7 @@ func ImageEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx
 					return err
 				}
 
-				item := &Item{}
+				item := &schema.Item{}
 
 				if b64JSON {
 					defer os.RemoveAll(output)
@@ -147,7 +174,7 @@ func ImageEndpoint(cm *config.ConfigLoader, o *options.Option) func(c *fiber.Ctx
 			}
 		}
 
-		resp := &OpenAIResponse{
+		resp := &schema.OpenAIResponse{
 			Data: result,
 		}
 

api/openai/inference.go

@@ -4,12 +4,20 @@ import (
 	"github.com/go-skynet/LocalAI/api/backend"
 	config "github.com/go-skynet/LocalAI/api/config"
 	"github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 )
 
-func ComputeChoices(req *OpenAIRequest, predInput string, config *config.Config, o *options.Option, loader *model.ModelLoader, cb func(string, *[]Choice), tokenCallback func(string) bool) ([]Choice, error) {
-	n := req.N
-	result := []Choice{}
+func ComputeChoices(
+	req *schema.OpenAIRequest,
+	predInput string,
+	config *config.Config,
+	o *options.Option,
+	loader *model.ModelLoader,
+	cb func(string, *[]schema.Choice),
+	tokenCallback func(string, backend.TokenUsage) bool) ([]schema.Choice, backend.TokenUsage, error) {
+	n := req.N // number of completions to return
+	result := []schema.Choice{}
 
 	if n == 0 {
 		n = 1
@@ -18,20 +26,25 @@ func ComputeChoices(req *OpenAIRequest, predInput string, config *config.Config,
 	// get the model function to call for the result
 	predFunc, err := backend.ModelInference(req.Context, predInput, loader, *config, o, tokenCallback)
 	if err != nil {
-		return result, err
+		return result, backend.TokenUsage{}, err
 	}
 
+	tokenUsage := backend.TokenUsage{}
+
 	for i := 0; i < n; i++ {
 		prediction, err := predFunc()
 		if err != nil {
-			return result, err
+			return result, backend.TokenUsage{}, err
 		}
 
-		prediction = backend.Finetune(*config, predInput, prediction)
-		cb(prediction, &result)
+		tokenUsage.Prompt += prediction.Usage.Prompt
+		tokenUsage.Completion += prediction.Usage.Completion
+
+		finetunedResponse := backend.Finetune(*config, predInput, prediction.Response)
+		cb(finetunedResponse, &result)
 
 		//result = append(result, Choice{Text: prediction})
 
 	}
-	return result, err
+	return result, tokenUsage, err
 }

api/openai/list.go

@@ -4,6 +4,7 @@ import (
 	"regexp"
 
 	config "github.com/go-skynet/LocalAI/api/config"
+	"github.com/go-skynet/LocalAI/api/schema"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 	"github.com/gofiber/fiber/v2"
 )
@@ -16,7 +17,7 @@ func ListModelsEndpoint(loader *model.ModelLoader, cm *config.ConfigLoader) func
 		}
 		var mm map[string]interface{} = map[string]interface{}{}
 
-		dataModels := []OpenAIModel{}
+		dataModels := []schema.OpenAIModel{}
 
 		var filterFn func(name string) bool
 		filter := c.Query("filter")
@@ -45,7 +46,7 @@ func ListModelsEndpoint(loader *model.ModelLoader, cm *config.ConfigLoader) func
 			}
 
 			if filterFn(c.Name) {
-				dataModels = append(dataModels, OpenAIModel{ID: c.Name, Object: "model"})
+				dataModels = append(dataModels, schema.OpenAIModel{ID: c.Name, Object: "model"})
 			}
 		}
 
@@ -53,13 +54,13 @@ func ListModelsEndpoint(loader *model.ModelLoader, cm *config.ConfigLoader) func
 		for _, m := range models {
 			// And only adds them if they shouldn't be skipped.
 			if _, exists := mm[m]; !exists && filterFn(m) {
-				dataModels = append(dataModels, OpenAIModel{ID: m, Object: "model"})
+				dataModels = append(dataModels, schema.OpenAIModel{ID: m, Object: "model"})
 			}
 		}
 
 		return c.JSON(struct {
-			Object string        `json:"object"`
-			Data   []OpenAIModel `json:"data"`
+			Object string               `json:"object"`
+			Data   []schema.OpenAIModel `json:"data"`
 		}{
 			Object: "list",
 			Data:   dataModels,

api/openai/request.go

@@ -10,14 +10,15 @@ import (
 
 	config "github.com/go-skynet/LocalAI/api/config"
 	options "github.com/go-skynet/LocalAI/api/options"
+	"github.com/go-skynet/LocalAI/api/schema"
 	model "github.com/go-skynet/LocalAI/pkg/model"
 	"github.com/gofiber/fiber/v2"
 	"github.com/rs/zerolog/log"
 )
 
-func readInput(c *fiber.Ctx, o *options.Option, randomModel bool) (string, *OpenAIRequest, error) {
+func readInput(c *fiber.Ctx, o *options.Option, randomModel bool) (string, *schema.OpenAIRequest, error) {
 	loader := o.Loader
-	input := new(OpenAIRequest)
+	input := new(schema.OpenAIRequest)
 	ctx, cancel := context.WithCancel(o.Context)
 	input.Context = ctx
 	input.Cancel = cancel
@@ -60,7 +61,7 @@ func readInput(c *fiber.Ctx, o *options.Option, randomModel bool) (string, *Open
 	return modelFile, input, nil
 }
 
-func updateConfig(config *config.Config, input *OpenAIRequest) {
+func updateConfig(config *config.Config, input *schema.OpenAIRequest) {
 	if input.Echo {
 		config.Echo = input.Echo
 	}
@@ -75,6 +76,10 @@ func updateConfig(config *config.Config, input *OpenAIRequest) {
 		config.Backend = input.Backend
 	}
 
+	if input.ClipSkip != 0 {
+		config.Diffusers.ClipSkip = input.ClipSkip
+	}
+
 	if input.ModelBaseName != "" {
 		config.AutoGPTQ.ModelBaseName = input.ModelBaseName
 	}
@@ -214,7 +219,7 @@ func updateConfig(config *config.Config, input *OpenAIRequest) {
 	}
 }
 
-func readConfig(modelFile string, input *OpenAIRequest, cm *config.ConfigLoader, loader *model.ModelLoader, debug bool, threads, ctx int, f16 bool) (*config.Config, *OpenAIRequest, error) {
+func readConfig(modelFile string, input *schema.OpenAIRequest, cm *config.ConfigLoader, loader *model.ModelLoader, debug bool, threads, ctx int, f16 bool) (*config.Config, *schema.OpenAIRequest, error) {
 	// Load a config file if present after the model name
 	modelConfig := filepath.Join(loader.ModelPath, modelFile+".yaml")
 

api/options/options.go

@@ -33,6 +33,8 @@ type Option struct {
 	ExternalGRPCBackends map[string]string
 
 	AutoloadGalleries bool
+
+	SingleBackend bool
 }
 
 type AppOption func(*Option)
@@ -58,6 +60,10 @@ func WithCors(b bool) AppOption {
 	}
 }
 
+var EnableSingleBackend = func(o *Option) {
+	o.SingleBackend = true
+}
+
 var EnableGalleriesAutoload = func(o *Option) {
 	o.AutoloadGalleries = true
 }
@@ -93,6 +99,7 @@ func WithStringGalleries(galls string) AppOption {
 	return func(o *Option) {
 		if galls == "" {
 			log.Debug().Msgf("no galleries to load")
+			o.Galleries = []gallery.Gallery{}
 			return
 		}
 		var galleries []gallery.Gallery

api/schema/openai.go

@@ -1,4 +1,4 @@
-package openai
+package schema
 
 import (
 	"context"

api/schema/whisper.go

@@ -1,4 +1,4 @@
-package api
+package schema
 
 import "time"
 

cmd/grpc/bert-embeddings/main.go

@@ -5,8 +5,8 @@ package main
 import (
 	"flag"
 
+	bert "github.com/go-skynet/LocalAI/pkg/backend/llm/bert"
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
-	bert "github.com/go-skynet/LocalAI/pkg/grpc/llm/bert"
 )
 
 var (

cmd/grpc/bloomz/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	bloomz "github.com/go-skynet/LocalAI/pkg/grpc/llm/bloomz"
+	bloomz "github.com/go-skynet/LocalAI/pkg/backend/llm/bloomz"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/dolly/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/falcon-ggml/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/falcon/main.go

@@ -7,7 +7,7 @@ package main
 import (
 	"flag"
 
-	falcon "github.com/go-skynet/LocalAI/pkg/grpc/llm/falcon"
+	falcon "github.com/go-skynet/LocalAI/pkg/backend/llm/falcon"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/gpt2/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/gpt4all/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	gpt4all "github.com/go-skynet/LocalAI/pkg/grpc/llm/gpt4all"
+	gpt4all "github.com/go-skynet/LocalAI/pkg/backend/llm/gpt4all"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/gptj/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/gptneox/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/langchain-huggingface/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	langchain "github.com/go-skynet/LocalAI/pkg/grpc/llm/langchain"
+	langchain "github.com/go-skynet/LocalAI/pkg/backend/llm/langchain"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/llama-stable/main.go

@@ -0,0 +1,21 @@
+package main
+
+import (
+	"flag"
+
+	llama "github.com/go-skynet/LocalAI/pkg/backend/llm/llama-stable"
+
+	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
+)
+
+var (
+	addr = flag.String("addr", "localhost:50051", "the address to connect to")
+)
+
+func main() {
+	flag.Parse()
+
+	if err := grpc.StartServer(*addr, &llama.LLM{}); err != nil {
+		panic(err)
+	}
+}

cmd/grpc/llama/main.go

@@ -7,7 +7,7 @@ package main
 import (
 	"flag"
 
-	llama "github.com/go-skynet/LocalAI/pkg/grpc/llm/llama"
+	llama "github.com/go-skynet/LocalAI/pkg/backend/llm/llama"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/mpt/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/piper/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	tts "github.com/go-skynet/LocalAI/pkg/grpc/tts"
+	tts "github.com/go-skynet/LocalAI/pkg/backend/tts"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/replit/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/rwkv/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	rwkv "github.com/go-skynet/LocalAI/pkg/grpc/llm/rwkv"
+	rwkv "github.com/go-skynet/LocalAI/pkg/backend/llm/rwkv"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/stablediffusion/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	image "github.com/go-skynet/LocalAI/pkg/grpc/image"
+	image "github.com/go-skynet/LocalAI/pkg/backend/image"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/starcoder/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transformers "github.com/go-skynet/LocalAI/pkg/grpc/llm/transformers"
+	transformers "github.com/go-skynet/LocalAI/pkg/backend/llm/transformers"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

cmd/grpc/whisper/main.go

@@ -5,7 +5,7 @@ package main
 import (
 	"flag"
 
-	transcribe "github.com/go-skynet/LocalAI/pkg/grpc/transcribe"
+	transcribe "github.com/go-skynet/LocalAI/pkg/backend/transcribe"
 
 	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
 )

examples/README.md

@@ -157,6 +157,26 @@ Allows to run any LocalAI-compatible model as a backend on the servers of https:
 
 [Check it out here](https://runpod.io/gsc?template=uv9mtqnrd0&ref=984wlcra)
 
+### Continue
+
+_by [@gruberdev](https://github.com/gruberdev)_
+
+<img src="continue/img/screen.png" width="600" height="200" alt="Screenshot">
+
+Demonstrates how to integrate an open-source copilot alternative that enhances code analysis, completion, and improvements. This approach seamlessly integrates with any LocalAI model, offering a more user-friendly experience.
+
+[Check it out here](https://github.com/go-skynet/LocalAI/tree/master/examples/continue/)
+
+### Streamlit bot
+
+_by [@majoshi1](https://github.com/majoshi1)_
+
+![Screenshot](streamlit-bot/streamlit-bot.png)
+
+A chat bot made using `Streamlit` & LocalAI.
+
+[Check it out here](https://github.com/go-skynet/LocalAI/tree/master/examples/streamlit-bot/)
+
 ## Want to contribute?
 
 Create an issue, and put `Example: <description>` in the title! We will post your examples here.

examples/chatbot-ui-manual/README.md

@@ -24,10 +24,13 @@ docker-compose up -d --pull always
 # docker-compose up -d --build
 ```
 
+Then browse to `http://localhost:3000` to view the Web UI.
+
 ## Pointing chatbot-ui to a separately managed LocalAI service
 
-If you want to use the [chatbot-ui example](https://github.com/go-skynet/LocalAI/tree/master/examples/chatbot-ui) with an externally managed LocalAI service, you can alter the `docker-compose` file so that it looks like the below. You will notice the file is smaller, because we have removed the section that would normally start the LocalAI service. Take care to update the IP address (or FQDN) that the chatbot-ui service tries to access (marked `<<LOCALAI_IP>>` below):
-```
+If you want to use the [chatbot-ui example](https://github.com/go-skynet/LocalAI/tree/master/examples/chatbot-ui) with an externally managed LocalAI service, you can alter the `docker-compose.yaml` file so that it looks like the below. You will notice the file is smaller, because we have removed the section that would normally start the LocalAI service. Take care to update the IP address (or FQDN) that the chatbot-ui service tries to access (marked `<<LOCALAI_IP>>` below):
+
+```yaml
 version: '3.6'
 
 services:
@@ -40,9 +43,8 @@ services:
       - 'OPENAI_API_HOST=http://<<LOCALAI_IP>>:8080'
 ```
 
-Once you've edited the Dockerfile, you can start it with `docker compose up`, then browse to `http://localhost:3000`.
+Once you've edited the `docker-compose.yaml`, you can start it with `docker compose up`, then browse to `http://localhost:3000` to view the Web UI.
 
 ## Accessing chatbot-ui
 
 Open http://localhost:3000 for the Web UI.
-

examples/chatbot-ui/README.md

@@ -20,10 +20,13 @@ docker-compose up --pull always
 # docker-compose up -d --build
 ```
 
+Then browse to `http://localhost:3000` to view the Web UI.
+
 ## Pointing chatbot-ui to a separately managed LocalAI service
 
-If you want to use the [chatbot-ui example](https://github.com/go-skynet/LocalAI/tree/master/examples/chatbot-ui) with an externally managed LocalAI service, you can alter the `docker-compose` file so that it looks like the below. You will notice the file is smaller, because we have removed the section that would normally start the LocalAI service. Take care to update the IP address (or FQDN) that the chatbot-ui service tries to access (marked `<<LOCALAI_IP>>` below):
-```
+If you want to use the [chatbot-ui example](https://github.com/go-skynet/LocalAI/tree/master/examples/chatbot-ui) with an externally managed LocalAI service, you can alter the `docker-compose.yaml` file so that it looks like the below. You will notice the file is smaller, because we have removed the section that would normally start the LocalAI service. Take care to update the IP address (or FQDN) that the chatbot-ui service tries to access (marked `<<LOCALAI_IP>>` below):
+
+```yaml
 version: '3.6'
 
 services:
@@ -36,9 +39,8 @@ services:
       - 'OPENAI_API_HOST=http://<<LOCALAI_IP>>:8080'
 ```
 
-Once you've edited the Dockerfile, you can start it with `docker compose up`, then browse to `http://localhost:3000`.
+Once you've edited the `docker-compose.yaml`, you can start it with `docker compose up`, then browse to `http://localhost:3000` to view the Web UI.
 
 ## Accessing chatbot-ui
 
 Open http://localhost:3000 for the Web UI.
-

examples/continue/README.md

@@ -0,0 +1,53 @@
+# Continue
+
+![logo](https://continue.dev/docs/assets/images/continue-cover-logo-aa135cc83fe8a14af480d1633ed74eb5.png)
+
+This document presents an example of integration with [continuedev/continue](https://github.com/continuedev/continue).
+
+![Screenshot](https://continue.dev/docs/assets/images/continue-screenshot-1f36b99467817f755739d7f4c4c08fe3.png)
+
+For a live demonstration, please click on the link below:
+
+- [How it works (Video demonstration)](https://www.youtube.com/watch?v=3Ocrc-WX4iQ)
+
+## Integration Setup Walkthrough
+
+1. [As outlined in `continue`'s documentation](https://continue.dev/docs/getting-started), install the [Visual Studio Code extension from the marketplace](https://marketplace.visualstudio.com/items?itemName=Continue.continue) and open it.
+2. In this example, LocalAI will download the gpt4all model and set it up as "gpt-3.5-turbo". Refer to the `docker-compose.yaml` file for details.
+
+    ```bash
+    # Clone LocalAI
+    git clone https://github.com/go-skynet/LocalAI
+
+    cd LocalAI/examples/continue
+
+    # Start with docker-compose
+    docker-compose up --build -d
+    ```
+
+3. Type `/config` within Continue's VSCode extension, or edit the file located at `~/.continue/config.py` on your system with the following configuration:
+
+    ```py
+    from continuedev.src.continuedev.libs.llm.openai import OpenAI
+
+    config = ContinueConfig(
+       ...
+       models=Models(
+            default=OpenAI(
+               api_key="my-api-key",
+               model="gpt-3.5-turbo",
+               api_base="http://localhost:8080",
+            )
+       ),
+    )
+    ```
+
+This setup enables you to make queries directly to your model running in the Docker container. Note that the `api_key` does not need to be properly set up; it is included here as a placeholder.
+
+If editing the configuration seems confusing, you may copy and paste the provided default `config.py` file over the existing one in `~/.continue/config.py` after initializing the extension in the VSCode IDE.
+
+## Additional Resources
+
+- [Official Continue documentation](https://continue.dev/docs/intro)
+- [Documentation page on using self-hosted models](https://continue.dev/docs/customization#self-hosting-an-open-source-model)
+- [Official extension link](https://marketplace.visualstudio.com/items?itemName=Continue.continue)

examples/continue/config.py

@@ -0,0 +1,148 @@
+"""
+This is the Continue configuration file.
+
+See https://continue.dev/docs/customization to learn more.
+"""
+
+import subprocess
+
+from continuedev.src.continuedev.core.main import Step
+from continuedev.src.continuedev.core.sdk import ContinueSDK
+from continuedev.src.continuedev.core.models import Models
+from continuedev.src.continuedev.core.config import CustomCommand, SlashCommand, ContinueConfig
+from continuedev.src.continuedev.plugins.context_providers.github import GitHubIssuesContextProvider
+from continuedev.src.continuedev.plugins.context_providers.google import GoogleContextProvider
+from continuedev.src.continuedev.plugins.policies.default import DefaultPolicy
+from continuedev.src.continuedev.libs.llm.openai import OpenAI, OpenAIServerInfo
+from continuedev.src.continuedev.libs.llm.ggml import GGML
+
+from continuedev.src.continuedev.plugins.steps.open_config import OpenConfigStep
+from continuedev.src.continuedev.plugins.steps.clear_history import ClearHistoryStep
+from continuedev.src.continuedev.plugins.steps.feedback import FeedbackStep
+from continuedev.src.continuedev.plugins.steps.comment_code import CommentCodeStep
+from continuedev.src.continuedev.plugins.steps.share_session import ShareSessionStep
+from continuedev.src.continuedev.plugins.steps.main import EditHighlightedCodeStep
+from continuedev.src.continuedev.plugins.context_providers.search import SearchContextProvider
+from continuedev.src.continuedev.plugins.context_providers.diff import DiffContextProvider
+from continuedev.src.continuedev.plugins.context_providers.url import URLContextProvider
+
+class CommitMessageStep(Step):
+    """
+    This is a Step, the building block of Continue.
+    It can be used below as a slash command, so that
+    run will be called when you type '/commit'.
+    """
+    async def run(self, sdk: ContinueSDK):
+
+        # Get the root directory of the workspace
+        dir = sdk.ide.workspace_directory
+
+        # Run git diff in that directory
+        diff = subprocess.check_output(
+            ["git", "diff"], cwd=dir).decode("utf-8")
+
+        # Ask the LLM to write a commit message,
+        # and set it as the description of this step
+        self.description = await sdk.models.default.complete(
+            f"{diff}\n\nWrite a short, specific (less than 50 chars) commit message about the above changes:")
+
+
+config = ContinueConfig(
+
+    # If set to False, we will not collect any usage data
+    # See here to learn what anonymous data we collect: https://continue.dev/docs/telemetry
+    allow_anonymous_telemetry=True,
+
+    models = Models(
+        default = OpenAI(
+            api_key = "my-api-key",
+            model = "gpt-3.5-turbo",
+            openai_server_info = OpenAIServerInfo(
+                api_base = "http://localhost:8080",
+                model = "gpt-3.5-turbo"
+            )
+        )
+    ),
+    # Set a system message with information that the LLM should always keep in mind
+    # E.g. "Please give concise answers. Always respond in Spanish."
+    system_message=None,
+
+    # Set temperature to any value between 0 and 1. Higher values will make the LLM
+    # more creative, while lower values will make it more predictable.
+    temperature=0.5,
+
+    # Custom commands let you map a prompt to a shortened slash command
+    # They are like slash commands, but more easily defined - write just a prompt instead of a Step class
+    # Their output will always be in chat form
+    custom_commands=[
+        # CustomCommand(
+        #     name="test",
+        #     description="Write unit tests for the higlighted code",
+        #     prompt="Write a comprehensive set of unit tests for the selected code. It should setup, run tests that check for correctness including important edge cases, and teardown. Ensure that the tests are complete and sophisticated. Give the tests just as chat output, don't edit any file.",
+        # )
+    ],
+
+    # Slash commands let you run a Step from a slash command
+    slash_commands=[
+        # SlashCommand(
+        #     name="commit",
+        #     description="This is an example slash command. Use /config to edit it and create more",
+        #     step=CommitMessageStep,
+        # )
+        SlashCommand(
+            name="edit",
+            description="Edit code in the current file or the highlighted code",
+            step=EditHighlightedCodeStep,
+        ),
+        SlashCommand(
+            name="config",
+            description="Customize Continue - slash commands, LLMs, system message, etc.",
+            step=OpenConfigStep,
+        ),
+        SlashCommand(
+            name="comment",
+            description="Write comments for the current file or highlighted code",
+            step=CommentCodeStep,
+        ),
+        SlashCommand(
+            name="feedback",
+            description="Send feedback to improve Continue",
+            step=FeedbackStep,
+        ),
+        SlashCommand(
+            name="clear",
+            description="Clear step history",
+            step=ClearHistoryStep,
+        ),
+        SlashCommand(
+            name="share",
+            description="Download and share the session transcript",
+            step=ShareSessionStep,
+        )
+    ],
+
+    # Context providers let you quickly select context by typing '@'
+    # Uncomment the following to
+    # - quickly reference GitHub issues
+    # - show Google search results to the LLM
+    context_providers=[
+        # GitHubIssuesContextProvider(
+        #     repo_name="<your github username or organization>/<your repo name>",
+        #     auth_token="<your github auth token>"
+        # ),
+        # GoogleContextProvider(
+        #     serper_api_key="<your serper.dev api key>"
+        # )
+        SearchContextProvider(),
+        DiffContextProvider(),
+        URLContextProvider(
+            preset_urls = [
+                # Add any common urls you reference here so they appear in autocomplete
+            ]
+        )
+    ],
+
+    # Policies hold the main logic that decides which Step to take next
+    # You can use them to design agents, or deeply customize Continue
+    policy=DefaultPolicy()
+)

examples/continue/docker-compose.yml

@@ -0,0 +1,27 @@
+version: '3.6'
+
+services:
+  api:
+    image: quay.io/go-skynet/local-ai:latest
+    # As initially LocalAI will download the models defined in PRELOAD_MODELS
+    # you might need to tweak the healthcheck values here according to your network connection.
+    # Here we give a timespan of 20m to download all the required files.
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8080/readyz"]
+      interval: 1m
+      timeout: 20m
+      retries: 20
+    build:
+      context: ../../
+      dockerfile: Dockerfile
+    ports:
+      - 8080:8080
+    environment:
+      - DEBUG=true
+      - MODELS_PATH=/models
+      # You can preload different models here as well.
+      # See: https://github.com/go-skynet/model-gallery
+      - 'PRELOAD_MODELS=[{"url": "github:go-skynet/model-gallery/gpt4all-j.yaml", "name": "gpt-3.5-turbo"}]'
+    volumes:
+      - ./models:/models:cached
+    command: ["/usr/bin/local-ai" ]

examples/insomnia/README.md

@@ -0,0 +1,17 @@
+# Insomnia
+
+Developer Testing Request Collection for [Insomnia](https://insomnia.rest/), an open-source REST client
+
+## Instructions
+
+* Install Insomnia as normal
+* [Import](https://docs.insomnia.rest/insomnia/import-export-data) `Insomnia_LocalAI.json`
+* Control + E opens the environment settings - 
+
+| **Parameter Name** | **Default Value** | **Description**                          |
+|--------------------|-------------------|------------------------------------------|
+| HOST               | localhost         | LocalAI base URL                         |
+| PORT               | 8080              | LocalAI port                             |
+| DEFAULT_MODEL      | gpt-3.5-turbo     | Name of the model used on most requests. |
+
+** you may want to duplicate localhost into a "Private" environment to avoid saving private settings back to this file **

examples/query_data/query.py

@@ -15,7 +15,7 @@ llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="gpt-3.5-turbo
 # Configure prompt parameters and initialise helper
 max_input_size = 500
 num_output = 256
-max_chunk_overlap = 20
+max_chunk_overlap = 0.2
 
 prompt_helper = PromptHelper(max_input_size, num_output, max_chunk_overlap)
 

examples/query_data/store.py

@@ -15,7 +15,7 @@ llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="gpt-3.5-turbo
 # Configure prompt parameters and initialise helper
 max_input_size = 400
 num_output = 400
-max_chunk_overlap = 30
+max_chunk_overlap = 0.3
 
 prompt_helper = PromptHelper(max_input_size, num_output, max_chunk_overlap)
 

examples/streamlit-bot/.gitignore

@@ -0,0 +1 @@
+installer_files

examples/streamlit-bot/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Manohar Joshi
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

examples/streamlit-bot/Main.py

@@ -0,0 +1,70 @@
+import streamlit as st
+import time
+import requests
+import json
+
+def ask(prompt):
+    url = 'http://localhost:8080/v1/chat/completions'
+    myobj = {
+        "model": "ggml-gpt4all-j.bin",
+        "messages": [{"role": "user", "content": prompt}],
+        "temperature": 0.9
+    }
+    myheaders = { "Content-Type" : "application/json" }  
+
+    x = requests.post(url, json = myobj, headers=myheaders)
+    
+    print(x.text)
+    
+    json_data = json.loads(x.text)
+
+    return json_data["choices"][0]["message"]["content"]
+
+
+def main():
+    # Page setup
+    st.set_page_config(page_title="Ask your LLM")
+    st.header("Ask your Question 💬")
+
+    # Initialize chat history
+    if "messages" not in st.session_state:
+        st.session_state.messages = []
+
+    # Display chat messages from history on app rerun
+    for message in st.session_state.messages:
+        with st.chat_message(message["role"]):
+            st.markdown(message["content"])
+
+    # Scroll to bottom
+    st.markdown(
+        """
+        <script>
+        var element = document.getElementById("end-of-chat");
+        element.scrollIntoView({behavior: "smooth"});
+        </script>
+        """,
+        unsafe_allow_html=True,
+    )   
+
+    # React to user input
+    if prompt := st.chat_input("What is up?"):
+        # Display user message in chat message container
+        st.chat_message("user").markdown(prompt)
+        # Add user message to chat history
+        st.session_state.messages.append({"role": "user", "content": prompt})
+        print(f"User has asked the following question: {prompt}")
+        
+        # Process
+        response = ""
+        with st.spinner('Processing...'):
+            response = ask(prompt)
+            
+        #response = f"Echo: {prompt}"
+        # Display assistant response in chat message container
+        with st.chat_message("assistant"):
+            st.markdown(response)
+        # Add assistant response to chat history
+        st.session_state.messages.append({"role": "assistant", "content": response})     
+
+if __name__ == "__main__":
+    main()        

examples/streamlit-bot/README.md

@@ -0,0 +1,54 @@
+## Streamlit bot
+
+![Screenshot](streamlit-bot.png)
+
+This is an example to deploy a Streamlit bot with LocalAI instead of OpenAI. Instructions are for Windows.
+
+```bash
+# Install & run Git Bash
+
+# Clone LocalAI
+git clone https://github.com/go-skynet/LocalAI.git
+cd LocalAI
+
+# (optional) Checkout a specific LocalAI tag
+# git checkout -b build <TAG>
+
+# Use a template from the examples
+cp -rf prompt-templates/ggml-gpt4all-j.tmpl models/
+
+# (optional) Edit the .env file to set things like context size and threads
+# vim .env
+# Download model
+curl --progress-bar -C - -O https://gpt4all.io/models/ggml-gpt4all-j.bin > models/ggml-gpt4all-j.bin
+
+# Install & Run Docker Desktop for Windows
+https://www.docker.com/products/docker-desktop/
+
+# start with docker-compose
+docker-compose up -d --pull always
+# or you can build the images with:
+# docker-compose up -d --build
+# Now API is accessible at localhost:8080
+curl http://localhost:8080/v1/models
+# {"object":"list","data":[{"id":"ggml-gpt4all-j","object":"model"}]}
+
+curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+     "model": "ggml-gpt4all-j",
+     "messages": [{"role": "user", "content": "How are you?"}],
+     "temperature": 0.9
+   }'
+
+# {"model":"ggml-gpt4all-j","choices":[{"message":{"role":"assistant","content":"I'm doing well, thanks. How about you?"}}]}
+
+cd examples/streamlit-bot
+
+install_requirements.bat
+
+# run the bot
+start_windows.bat
+
+# UI will be launched automatically (http://localhost:8501/) in browser.
+
+```
+

examples/streamlit-bot/cmd_windows.bat

@@ -0,0 +1,31 @@
+@echo off
+
+cd /D "%~dp0"
+
+set PATH=%PATH%;%SystemRoot%\system32
+
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+
+@rem fix failed install when installing to a separate drive
+set TMP=%cd%\installer_files
+set TEMP=%cd%\installer_files
+
+@rem config
+set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
+set INSTALL_ENV_DIR=%cd%\installer_files\env
+
+@rem environment isolation
+set PYTHONNOUSERSITE=1
+set PYTHONPATH=
+set PYTHONHOME=
+set "CUDA_PATH=%INSTALL_ENV_DIR%"
+set "CUDA_HOME=%CUDA_PATH%"
+
+@rem activate installer env
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+
+@rem enter commands
+cmd /k "%*"
+
+:end
+pause

examples/streamlit-bot/install_requirements.bat

@@ -0,0 +1,81 @@
+@echo off
+
+cd /D "%~dp0"
+
+set PATH=%PATH%;%SystemRoot%\system32
+
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+
+@rem Check for special characters in installation path
+set "SPCHARMESSAGE="WARNING: Special characters were detected in the installation path!" "         This can cause the installation to fail!""
+echo "%CD%"| findstr /R /C:"[!#\$%&()\*+,;<=>?@\[\]\^`{|}~]" >nul && (
+	call :PrintBigMessage %SPCHARMESSAGE%
+)
+set SPCHARMESSAGE=
+
+@rem fix failed install when installing to a separate drive
+set TMP=%cd%\installer_files
+set TEMP=%cd%\installer_files
+
+@rem config
+set INSTALL_DIR=%cd%\installer_files
+set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
+set INSTALL_ENV_DIR=%cd%\installer_files\env
+set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-Windows-x86_64.exe
+set conda_exists=F
+
+@rem figure out whether git and conda needs to be installed
+call "%CONDA_ROOT_PREFIX%\_conda.exe" --version >nul 2>&1
+if "%ERRORLEVEL%" EQU "0" set conda_exists=T
+
+@rem (if necessary) install git and conda into a contained environment
+@rem download conda
+if "%conda_exists%" == "F" (
+	echo Downloading Miniconda from %MINICONDA_DOWNLOAD_URL% to %INSTALL_DIR%\miniconda_installer.exe
+
+	mkdir "%INSTALL_DIR%"
+	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
+
+	echo Installing Miniconda to %CONDA_ROOT_PREFIX%
+	start /wait "" "%INSTALL_DIR%\miniconda_installer.exe" /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%CONDA_ROOT_PREFIX%
+
+	@rem test the conda binary
+	echo Miniconda version:
+	call "%CONDA_ROOT_PREFIX%\_conda.exe" --version || ( echo. && echo Miniconda not found. && goto end )
+)
+
+@rem create the installer env
+if not exist "%INSTALL_ENV_DIR%" (
+  echo Packages to install: %PACKAGES_TO_INSTALL%
+  call "%CONDA_ROOT_PREFIX%\_conda.exe" create --no-shortcuts -y -k --prefix "%INSTALL_ENV_DIR%" python=3.10 || ( echo. && echo Conda environment creation failed. && goto end )
+)
+
+@rem check if conda environment was actually created
+if not exist "%INSTALL_ENV_DIR%\python.exe" ( echo. && echo Conda environment is empty. && goto end )
+
+@rem environment isolation
+set PYTHONNOUSERSITE=1
+set PYTHONPATH=
+set PYTHONHOME=
+set "CUDA_PATH=%INSTALL_ENV_DIR%"
+set "CUDA_HOME=%CUDA_PATH%"
+
+@rem activate installer env
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+
+@rem setup installer env
+call pip install -r requirements.txt
+
+@rem below are functions for the script   next line skips these during normal execution
+goto end
+
+:PrintBigMessage
+echo. && echo.
+echo *******************************************************************
+for %%M in (%*) do echo * %%~M
+echo *******************************************************************
+echo. && echo.
+exit /b
+
+:end
+pause

examples/streamlit-bot/requirements.txt

@@ -0,0 +1,2 @@
+streamlit==1.26.0
+requests

examples/streamlit-bot/start_windows.bat

@@ -0,0 +1,81 @@
+@echo off
+
+cd /D "%~dp0"
+
+set PATH=%PATH%;%SystemRoot%\system32
+
+echo "%CD%"| findstr /C:" " >nul && echo This script relies on Miniconda which can not be silently installed under a path with spaces. && goto end
+
+@rem Check for special characters in installation path
+set "SPCHARMESSAGE="WARNING: Special characters were detected in the installation path!" "         This can cause the installation to fail!""
+echo "%CD%"| findstr /R /C:"[!#\$%&()\*+,;<=>?@\[\]\^`{|}~]" >nul && (
+	call :PrintBigMessage %SPCHARMESSAGE%
+)
+set SPCHARMESSAGE=
+
+@rem fix failed install when installing to a separate drive
+set TMP=%cd%\installer_files
+set TEMP=%cd%\installer_files
+
+@rem config
+set INSTALL_DIR=%cd%\installer_files
+set CONDA_ROOT_PREFIX=%cd%\installer_files\conda
+set INSTALL_ENV_DIR=%cd%\installer_files\env
+set MINICONDA_DOWNLOAD_URL=https://repo.anaconda.com/miniconda/Miniconda3-py310_23.3.1-0-Windows-x86_64.exe
+set conda_exists=F
+
+@rem figure out whether git and conda needs to be installed
+call "%CONDA_ROOT_PREFIX%\_conda.exe" --version >nul 2>&1
+if "%ERRORLEVEL%" EQU "0" set conda_exists=T
+
+@rem (if necessary) install git and conda into a contained environment
+@rem download conda
+if "%conda_exists%" == "F" (
+	echo Downloading Miniconda from %MINICONDA_DOWNLOAD_URL% to %INSTALL_DIR%\miniconda_installer.exe
+
+	mkdir "%INSTALL_DIR%"
+	call curl -Lk "%MINICONDA_DOWNLOAD_URL%" > "%INSTALL_DIR%\miniconda_installer.exe" || ( echo. && echo Miniconda failed to download. && goto end )
+
+	echo Installing Miniconda to %CONDA_ROOT_PREFIX%
+	start /wait "" "%INSTALL_DIR%\miniconda_installer.exe" /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%CONDA_ROOT_PREFIX%
+
+	@rem test the conda binary
+	echo Miniconda version:
+	call "%CONDA_ROOT_PREFIX%\_conda.exe" --version || ( echo. && echo Miniconda not found. && goto end )
+)
+
+@rem create the installer env
+if not exist "%INSTALL_ENV_DIR%" (
+  echo Packages to install: %PACKAGES_TO_INSTALL%
+  call "%CONDA_ROOT_PREFIX%\_conda.exe" create --no-shortcuts -y -k --prefix "%INSTALL_ENV_DIR%" python=3.10 || ( echo. && echo Conda environment creation failed. && goto end )
+)
+
+@rem check if conda environment was actually created
+if not exist "%INSTALL_ENV_DIR%\python.exe" ( echo. && echo Conda environment is empty. && goto end )
+
+@rem environment isolation
+set PYTHONNOUSERSITE=1
+set PYTHONPATH=
+set PYTHONHOME=
+set "CUDA_PATH=%INSTALL_ENV_DIR%"
+set "CUDA_HOME=%CUDA_PATH%"
+
+@rem activate installer env
+call "%CONDA_ROOT_PREFIX%\condabin\conda.bat" activate "%INSTALL_ENV_DIR%" || ( echo. && echo Miniconda hook not found. && goto end )
+
+@rem setup installer env
+streamlit run Main.py
+
+@rem below are functions for the script   next line skips these during normal execution
+goto end
+
+:PrintBigMessage
+echo. && echo.
+echo *******************************************************************
+for %%M in (%*) do echo * %%~M
+echo *******************************************************************
+echo. && echo.
+exit /b
+
+:end
+pause

extra/grpc/autogptq/autogptq.py

@@ -15,6 +15,9 @@ from transformers import TextGenerationPipeline
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
     def Health(self, request, context):
@@ -77,7 +80,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
 
 
 def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
     server.add_insecure_port(address)
     server.start()

extra/grpc/autogptq/backend_pb2_grpc.py

@@ -54,6 +54,16 @@ class BackendStub(object):
                 request_serializer=backend__pb2.TTSRequest.SerializeToString,
                 response_deserializer=backend__pb2.Result.FromString,
                 )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
 
 
 class BackendServicer(object):
@@ -107,6 +117,18 @@ class BackendServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_BackendServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -150,6 +172,16 @@ def add_BackendServicer_to_server(servicer, server):
                     request_deserializer=backend__pb2.TTSRequest.FromString,
                     response_serializer=backend__pb2.Result.SerializeToString,
             ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'backend.Backend', rpc_method_handlers)
@@ -295,3 +327,37 @@ class Backend(object):
             backend__pb2.Result.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/bark/backend_pb2_grpc.py

@@ -54,6 +54,16 @@ class BackendStub(object):
                 request_serializer=backend__pb2.TTSRequest.SerializeToString,
                 response_deserializer=backend__pb2.Result.FromString,
                 )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
 
 
 class BackendServicer(object):
@@ -107,6 +117,18 @@ class BackendServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_BackendServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -150,6 +172,16 @@ def add_BackendServicer_to_server(servicer, server):
                     request_deserializer=backend__pb2.TTSRequest.FromString,
                     response_serializer=backend__pb2.Result.SerializeToString,
             ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'backend.Backend', rpc_method_handlers)
@@ -295,3 +327,37 @@ class Backend(object):
             backend__pb2.Result.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/bark/ttsbark.py

@@ -15,6 +15,9 @@ from scipy.io.wavfile import write as write_wav
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
     def Health(self, request, context):
@@ -51,7 +54,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         return backend_pb2.Result(success=True)
 
 def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
     server.add_insecure_port(address)
     server.start()

extra/grpc/diffusers/backend_diffusers.py

@@ -12,9 +12,114 @@ import os
 # import diffusers
 import torch
 from torch import autocast
-from diffusers import StableDiffusionXLPipeline, DPMSolverMultistepScheduler, StableDiffusionPipeline, DiffusionPipeline, EulerAncestralDiscreteScheduler
-
+from diffusers import StableDiffusionXLPipeline, StableDiffusionDepth2ImgPipeline, DPMSolverMultistepScheduler, StableDiffusionPipeline, DiffusionPipeline, EulerAncestralDiscreteScheduler
+from diffusers.pipelines.stable_diffusion import safety_checker
+from compel import Compel
+from PIL import Image
+from io import BytesIO
+from diffusers import StableDiffusionImg2ImgPipeline
+from transformers import CLIPTextModel
+from enum import Enum
+from collections import defaultdict
+from safetensors.torch import load_file
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
+COMPEL=os.environ.get("COMPEL", "1") == "1"
+CLIPSKIP=os.environ.get("CLIPSKIP", "1") == "1"
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# https://github.com/CompVis/stable-diffusion/issues/239#issuecomment-1627615287
+def sc(self, clip_input, images) : return images, [False for i in images]
+# edit the StableDiffusionSafetyChecker class so that, when called, it just returns the images and an array of True values
+safety_checker.StableDiffusionSafetyChecker.forward = sc
+
+from diffusers.schedulers import (
+    DDIMScheduler,
+    DPMSolverMultistepScheduler,
+    DPMSolverSinglestepScheduler,
+    EulerAncestralDiscreteScheduler,
+    EulerDiscreteScheduler,
+    HeunDiscreteScheduler,
+    KDPM2AncestralDiscreteScheduler,
+    KDPM2DiscreteScheduler,
+    LMSDiscreteScheduler,
+    PNDMScheduler,
+    UniPCMultistepScheduler,
+)
+# The scheduler list mapping was taken from here: https://github.com/neggles/animatediff-cli/blob/6f336f5f4b5e38e85d7f06f1744ef42d0a45f2a7/src/animatediff/schedulers.py#L39
+# Credits to https://github.com/neggles
+# See https://github.com/huggingface/diffusers/issues/4167 for more details on sched mapping from A1111
+class DiffusionScheduler(str, Enum):
+    ddim = "ddim"  # DDIM
+    pndm = "pndm"  # PNDM
+    heun = "heun"  # Heun
+    unipc = "unipc"  # UniPC
+    euler = "euler"  # Euler
+    euler_a = "euler_a"  # Euler a
+
+    lms = "lms"  # LMS
+    k_lms = "k_lms"  # LMS Karras
+
+    dpm_2 = "dpm_2"  # DPM2
+    k_dpm_2 = "k_dpm_2"  # DPM2 Karras
+
+    dpm_2_a = "dpm_2_a"  # DPM2 a
+    k_dpm_2_a = "k_dpm_2_a"  # DPM2 a Karras
+
+    dpmpp_2m = "dpmpp_2m"  # DPM++ 2M
+    k_dpmpp_2m = "k_dpmpp_2m"  # DPM++ 2M Karras
+
+    dpmpp_sde = "dpmpp_sde"  # DPM++ SDE
+    k_dpmpp_sde = "k_dpmpp_sde"  # DPM++ SDE Karras
+
+    dpmpp_2m_sde = "dpmpp_2m_sde"  # DPM++ 2M SDE
+    k_dpmpp_2m_sde = "k_dpmpp_2m_sde"  # DPM++ 2M SDE Karras
+
+
+def get_scheduler(name: str, config: dict = {}):
+    is_karras = name.startswith("k_")
+    if is_karras:
+        # strip the k_ prefix and add the karras sigma flag to config
+        name = name.lstrip("k_")
+        config["use_karras_sigmas"] = True
+
+    if name == DiffusionScheduler.ddim:
+        sched_class = DDIMScheduler
+    elif name == DiffusionScheduler.pndm:
+        sched_class = PNDMScheduler
+    elif name == DiffusionScheduler.heun:
+        sched_class = HeunDiscreteScheduler
+    elif name == DiffusionScheduler.unipc:
+        sched_class = UniPCMultistepScheduler
+    elif name == DiffusionScheduler.euler:
+        sched_class = EulerDiscreteScheduler
+    elif name == DiffusionScheduler.euler_a:
+        sched_class = EulerAncestralDiscreteScheduler
+    elif name == DiffusionScheduler.lms:
+        sched_class = LMSDiscreteScheduler
+    elif name == DiffusionScheduler.dpm_2:
+        # Equivalent to DPM2 in K-Diffusion
+        sched_class = KDPM2DiscreteScheduler
+    elif name == DiffusionScheduler.dpm_2_a:
+        # Equivalent to `DPM2 a`` in K-Diffusion
+        sched_class = KDPM2AncestralDiscreteScheduler
+    elif name == DiffusionScheduler.dpmpp_2m:
+        # Equivalent to `DPM++ 2M` in K-Diffusion
+        sched_class = DPMSolverMultistepScheduler
+        config["algorithm_type"] = "dpmsolver++"
+        config["solver_order"] = 2
+    elif name == DiffusionScheduler.dpmpp_sde:
+        # Equivalent to `DPM++ SDE` in K-Diffusion
+        sched_class = DPMSolverSinglestepScheduler
+    elif name == DiffusionScheduler.dpmpp_2m_sde:
+        # Equivalent to `DPM++ 2M SDE` in K-Diffusion
+        sched_class = DPMSolverMultistepScheduler
+        config["algorithm_type"] = "sde-dpmsolver++"
+    else:
+        raise ValueError(f"Invalid scheduler '{'k_' if is_karras else ''}{name}'")
+
+    return sched_class.from_config(config)
 
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
@@ -28,61 +133,223 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             if request.F16Memory:
                 torchType = torch.float16
 
+            local = False
+            modelFile = request.Model
+
+            cfg_scale = 7
+            if request.CFGScale != 0:
+                cfg_scale = request.CFGScale
+            
+            clipmodel = "runwayml/stable-diffusion-v1-5"
+            if request.CLIPModel != "":
+                clipmodel = request.CLIPModel
+            clipsubfolder = "text_encoder"
+            if request.CLIPSubfolder != "":
+                clipsubfolder = request.CLIPSubfolder
+            
+            # Check if ModelFile exists
+            if request.ModelFile != "":
+                if os.path.exists(request.ModelFile):
+                    local = True
+                    modelFile = request.ModelFile
+            
+            fromSingleFile = request.Model.startswith("http") or request.Model.startswith("/") or local
+            
+            if request.IMG2IMG and request.PipelineType == "":
+                request.PipelineType == "StableDiffusionImg2ImgPipeline"
+                
             if request.PipelineType == "":
                 request.PipelineType == "StableDiffusionPipeline"
 
+            ## img2img
+            if request.PipelineType == "StableDiffusionImg2ImgPipeline":
+                if fromSingleFile:
+                    self.pipe = StableDiffusionImg2ImgPipeline.from_single_file(modelFile,
+                                torch_dtype=torchType,
+                                guidance_scale=cfg_scale)
+                else:
+                    self.pipe = StableDiffusionImg2ImgPipeline.from_pretrained(request.Model,
+                                torch_dtype=torchType,
+                                guidance_scale=cfg_scale)
+
+            if request.PipelineType == "StableDiffusionDepth2ImgPipeline":
+                self.pipe = StableDiffusionDepth2ImgPipeline.from_pretrained(request.Model,
+                            torch_dtype=torchType,
+                            guidance_scale=cfg_scale)
+            ## text2img
             if request.PipelineType == "StableDiffusionPipeline":
-                self.pipe = StableDiffusionPipeline.from_pretrained(request.Model,
-                                                               torch_dtype=torchType)
+                if fromSingleFile:
+                    self.pipe = StableDiffusionPipeline.from_single_file(modelFile,
+                                                        torch_dtype=torchType,
+                                                        guidance_scale=cfg_scale)
+                else:
+                    self.pipe = StableDiffusionPipeline.from_pretrained(request.Model,
+                                                        torch_dtype=torchType,
+                                                        guidance_scale=cfg_scale)
 
             if request.PipelineType == "DiffusionPipeline":
                 self.pipe = DiffusionPipeline.from_pretrained(request.Model,
-                                                               torch_dtype=torchType)
+                                                        torch_dtype=torchType,
+                                                        guidance_scale=cfg_scale)
 
             if request.PipelineType == "StableDiffusionXLPipeline":
-                self.pipe = StableDiffusionXLPipeline.from_pretrained(
-                    request.Model, 
-                    torch_dtype=torchType, 
-                    use_safetensors=True, 
-             #       variant="fp16"
-                    )
-          
+                if fromSingleFile:
+                    self.pipe = StableDiffusionXLPipeline.from_single_file(modelFile,
+                                                               torch_dtype=torchType, use_safetensors=True,
+                                                               guidance_scale=cfg_scale)
+                else:
+                    self.pipe = StableDiffusionXLPipeline.from_pretrained(
+                        request.Model, 
+                        torch_dtype=torchType, 
+                        use_safetensors=True, 
+                #       variant="fp16"
+                        guidance_scale=cfg_scale)
+            # https://github.com/huggingface/diffusers/issues/4446
+            # do not use text_encoder in the constructor since then
+            # https://github.com/huggingface/diffusers/issues/3212#issuecomment-1521841481
+            if CLIPSKIP and request.CLIPSkip != 0:
+                text_encoder = CLIPTextModel.from_pretrained(clipmodel, num_hidden_layers=request.CLIPSkip,  subfolder=clipsubfolder, torch_dtype=torchType)
+                self.pipe.text_encoder=text_encoder
             # torch_dtype needs to be customized. float16 for GPU, float32 for CPU
             # TODO: this needs to be customized
-            if request.SchedulerType == "EulerAncestralDiscreteScheduler":
-                self.pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(self.pipe.scheduler.config)
-            if request.SchedulerType == "DPMSolverMultistepScheduler":
-                self.pipe.scheduler = DPMSolverMultistepScheduler.from_config(self.pipe.scheduler.config)
-
+            if request.SchedulerType != "":
+                self.pipe.scheduler = get_scheduler(request.SchedulerType, self.pipe.scheduler.config)
+            self.compel = Compel(tokenizer=self.pipe.tokenizer, text_encoder=self.pipe.text_encoder)
             if request.CUDA:
                 self.pipe.to('cuda')
+            # Assume directory from request.ModelFile.
+            # Only if request.LoraAdapter it's not an absolute path
+            if request.LoraAdapter and request.ModelFile != "" and not os.path.isabs(request.LoraAdapter) and request.LoraAdapter:
+                # get base path of modelFile
+                modelFileBase = os.path.dirname(request.ModelFile)
+                # modify LoraAdapter to be relative to modelFileBase
+                request.LoraAdapter = os.path.join(modelFileBase, request.LoraAdapter)
+            device = "cpu" if not request.CUDA else "cuda"
+            self.device = device
+            if request.LoraAdapter:
+                # Check if its a local file and not a directory ( we load lora differently for a safetensor file )
+                if os.path.exists(request.LoraAdapter) and not os.path.isdir(request.LoraAdapter):
+                    self.load_lora_weights(request.LoraAdapter, 1, device, torchType)
+                else:
+                    self.pipe.unet.load_attn_procs(request.LoraAdapter)
+
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         # Implement your logic here for the LoadModel service
         # Replace this with your desired response
         return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    # https://github.com/huggingface/diffusers/issues/3064
+    def load_lora_weights(self, checkpoint_path, multiplier, device, dtype):
+        LORA_PREFIX_UNET = "lora_unet"
+        LORA_PREFIX_TEXT_ENCODER = "lora_te"
+        # load LoRA weight from .safetensors
+        state_dict = load_file(checkpoint_path, device=device)
+
+        updates = defaultdict(dict)
+        for key, value in state_dict.items():
+            # it is suggested to print out the key, it usually will be something like below
+            # "lora_te_text_model_encoder_layers_0_self_attn_k_proj.lora_down.weight"
+
+            layer, elem = key.split('.', 1)
+            updates[layer][elem] = value
+
+        # directly update weight in diffusers model
+        for layer, elems in updates.items():
+
+            if "text" in layer:
+                layer_infos = layer.split(LORA_PREFIX_TEXT_ENCODER + "_")[-1].split("_")
+                curr_layer = self.pipe.text_encoder
+            else:
+                layer_infos = layer.split(LORA_PREFIX_UNET + "_")[-1].split("_")
+                curr_layer = self.pipe.unet
+
+            # find the target layer
+            temp_name = layer_infos.pop(0)
+            while len(layer_infos) > -1:
+                try:
+                    curr_layer = curr_layer.__getattr__(temp_name)
+                    if len(layer_infos) > 0:
+                        temp_name = layer_infos.pop(0)
+                    elif len(layer_infos) == 0:
+                        break
+                except Exception:
+                    if len(temp_name) > 0:
+                        temp_name += "_" + layer_infos.pop(0)
+                    else:
+                        temp_name = layer_infos.pop(0)
+
+            # get elements for this layer
+            weight_up = elems['lora_up.weight'].to(dtype)
+            weight_down = elems['lora_down.weight'].to(dtype)
+            alpha = elems['alpha'] if 'alpha' in elems else None
+            if alpha:
+                alpha = alpha.item() / weight_up.shape[1]
+            else:
+                alpha = 1.0
+
+            # update weight
+            if len(weight_up.shape) == 4:
+                curr_layer.weight.data += multiplier * alpha * torch.mm(weight_up.squeeze(3).squeeze(2), weight_down.squeeze(3).squeeze(2)).unsqueeze(2).unsqueeze(3)
+            else:
+                curr_layer.weight.data += multiplier * alpha * torch.mm(weight_up, weight_down)
+
     def GenerateImage(self, request, context):
 
         prompt = request.positive_prompt
-        negative_prompt = request.negative_prompt
 
-        image = self.pipe(
-            prompt, 
-            negative_prompt=negative_prompt, 
-            width=request.width,
-            height=request.height,
-      #      guidance_scale=12,
-            target_size=(request.width,request.height),
-            original_size=(4096,4096),
-            num_inference_steps=request.step
-            ).images[0]
+        # create a dictionary of values for the parameters
+        options = {
+            "negative_prompt":     request.negative_prompt, 
+            "width":               request.width, 
+            "height":              request.height,
+            "num_inference_steps": request.step,
+        }
 
+        if request.src != "":
+            image = Image.open(request.src)
+            options["image"] = image
+
+        # Get the keys that we will build the args for our pipe for
+        keys = options.keys()
+
+        if request.EnableParameters != "":
+            keys = request.EnableParameters.split(",")
+
+        if request.EnableParameters == "none":
+            keys = []
+
+        # create a dictionary of parameters by using the keys from EnableParameters and the values from defaults
+        kwargs = {key: options[key] for key in keys}
+
+        # Set seed
+        if request.seed > 0:
+            kwargs["generator"] = torch.Generator(device=self.device).manual_seed(
+                request.seed
+            )
+
+        image = {}
+        if COMPEL:
+            conditioning = self.compel.build_conditioning_tensor(prompt)
+            kwargs["prompt_embeds"]= conditioning
+            # pass the kwargs dictionary to the self.pipe method
+            image = self.pipe( 
+                **kwargs
+                ).images[0] 
+        else:
+            # pass the kwargs dictionary to the self.pipe method
+            image = self.pipe(
+                prompt, 
+                **kwargs
+                ).images[0]
+
+        # save the result
         image.save(request.dst)
 
         return backend_pb2.Result(message="Model loaded successfully", success=True)
 
 def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
     server.add_insecure_port(address)
     server.start()

extra/grpc/diffusers/backend_pb2_grpc.py

@@ -54,6 +54,16 @@ class BackendStub(object):
                 request_serializer=backend__pb2.TTSRequest.SerializeToString,
                 response_deserializer=backend__pb2.Result.FromString,
                 )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
 
 
 class BackendServicer(object):
@@ -107,6 +117,18 @@ class BackendServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_BackendServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -150,6 +172,16 @@ def add_BackendServicer_to_server(servicer, server):
                     request_deserializer=backend__pb2.TTSRequest.FromString,
                     response_serializer=backend__pb2.Result.SerializeToString,
             ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'backend.Backend', rpc_method_handlers)
@@ -295,3 +327,37 @@ class Backend(object):
             backend__pb2.Result.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/exllama/backend_pb2_grpc.py

@@ -54,6 +54,16 @@ class BackendStub(object):
                 request_serializer=backend__pb2.TTSRequest.SerializeToString,
                 response_deserializer=backend__pb2.Result.FromString,
                 )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
 
 
 class BackendServicer(object):
@@ -107,6 +117,18 @@ class BackendServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_BackendServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -150,6 +172,16 @@ def add_BackendServicer_to_server(servicer, server):
                     request_deserializer=backend__pb2.TTSRequest.FromString,
                     response_serializer=backend__pb2.Result.SerializeToString,
             ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'backend.Backend', rpc_method_handlers)
@@ -295,3 +327,37 @@ class Backend(object):
             backend__pb2.Result.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/exllama/exllama.py

@@ -19,6 +19,9 @@ from exllama.tokenizer import ExLlamaTokenizer
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
     def generate(self,prompt, max_new_tokens):
@@ -110,7 +113,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
 
 
 def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
     server.add_insecure_port(address)
     server.start()

extra/grpc/huggingface/backend_pb2_grpc.py

@@ -54,6 +54,16 @@ class BackendStub(object):
                 request_serializer=backend__pb2.TTSRequest.SerializeToString,
                 response_deserializer=backend__pb2.Result.FromString,
                 )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
 
 
 class BackendServicer(object):
@@ -107,6 +117,18 @@ class BackendServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_BackendServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -150,6 +172,16 @@ def add_BackendServicer_to_server(servicer, server):
                     request_deserializer=backend__pb2.TTSRequest.FromString,
                     response_serializer=backend__pb2.Result.SerializeToString,
             ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'backend.Backend', rpc_method_handlers)
@@ -295,3 +327,37 @@ class Backend(object):
             backend__pb2.Result.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/huggingface/huggingface.py

@@ -12,6 +12,9 @@ from sentence_transformers import SentenceTransformer
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
     def Health(self, request, context):
@@ -34,7 +37,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
 
 
 def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
     server.add_insecure_port(address)
     server.start()

extra/grpc/vall-e-x/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
+"""Client and server classes corresponding to protobuf-defined services."""
+import grpc
+
+import backend_pb2 as backend__pb2
+
+
+class BackendStub(object):
+    """Missing associated documentation comment in .proto file."""
+
+    def __init__(self, channel):
+        """Constructor.
+
+        Args:
+            channel: A grpc.Channel.
+        """
+        self.Health = channel.unary_unary(
+                '/backend.Backend/Health',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.Predict = channel.unary_unary(
+                '/backend.Backend/Predict',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.LoadModel = channel.unary_unary(
+                '/backend.Backend/LoadModel',
+                request_serializer=backend__pb2.ModelOptions.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.PredictStream = channel.unary_stream(
+                '/backend.Backend/PredictStream',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.Embedding = channel.unary_unary(
+                '/backend.Backend/Embedding',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.EmbeddingResult.FromString,
+                )
+        self.GenerateImage = channel.unary_unary(
+                '/backend.Backend/GenerateImage',
+                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.AudioTranscription = channel.unary_unary(
+                '/backend.Backend/AudioTranscription',
+                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
+                response_deserializer=backend__pb2.TranscriptResult.FromString,
+                )
+        self.TTS = channel.unary_unary(
+                '/backend.Backend/TTS',
+                request_serializer=backend__pb2.TTSRequest.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
+
+
+class BackendServicer(object):
+    """Missing associated documentation comment in .proto file."""
+
+    def Health(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Predict(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def LoadModel(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def PredictStream(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Embedding(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def GenerateImage(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def AudioTranscription(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def TTS(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+
+def add_BackendServicer_to_server(servicer, server):
+    rpc_method_handlers = {
+            'Health': grpc.unary_unary_rpc_method_handler(
+                    servicer.Health,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'Predict': grpc.unary_unary_rpc_method_handler(
+                    servicer.Predict,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'LoadModel': grpc.unary_unary_rpc_method_handler(
+                    servicer.LoadModel,
+                    request_deserializer=backend__pb2.ModelOptions.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'PredictStream': grpc.unary_stream_rpc_method_handler(
+                    servicer.PredictStream,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'Embedding': grpc.unary_unary_rpc_method_handler(
+                    servicer.Embedding,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
+            ),
+            'GenerateImage': grpc.unary_unary_rpc_method_handler(
+                    servicer.GenerateImage,
+                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
+                    servicer.AudioTranscription,
+                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
+                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
+            ),
+            'TTS': grpc.unary_unary_rpc_method_handler(
+                    servicer.TTS,
+                    request_deserializer=backend__pb2.TTSRequest.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
+    }
+    generic_handler = grpc.method_handlers_generic_handler(
+            'backend.Backend', rpc_method_handlers)
+    server.add_generic_rpc_handlers((generic_handler,))
+
+
+ # This class is part of an EXPERIMENTAL API.
+class Backend(object):
+    """Missing associated documentation comment in .proto file."""
+
+    @staticmethod
+    def Health(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Predict(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def LoadModel(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
+            backend__pb2.ModelOptions.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def PredictStream(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Embedding(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.EmbeddingResult.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def GenerateImage(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
+            backend__pb2.GenerateImageRequest.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def AudioTranscription(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
+            backend__pb2.TranscriptRequest.SerializeToString,
+            backend__pb2.TranscriptResult.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TTS(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
+            backend__pb2.TTSRequest.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/vall-e-x/ttsvalle.py

@@ -0,0 +1,100 @@
+#!/usr/bin/env python3
+import grpc
+from concurrent import futures
+import time
+import backend_pb2
+import backend_pb2_grpc
+import argparse
+import signal
+import sys
+import os
+from pathlib import Path
+
+from utils.generation import SAMPLE_RATE, generate_audio, preload_models
+from scipy.io.wavfile import write as write_wav
+from utils.prompt_making import make_prompt
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    def Health(self, request, context):
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+    def LoadModel(self, request, context):
+        model_name = request.Model
+        try:
+            print("Preparing models, please wait", file=sys.stderr)
+            # download and load all models
+            preload_models()
+            # Assume directory from request.ModelFile.
+            # Only if request.LoraAdapter it's not an absolute path
+            if request.AudioPath and request.ModelFile != "" and not os.path.isabs(request.AudioPath):
+                # get base path of modelFile
+                modelFileBase = os.path.dirname(request.ModelFile)
+                # modify LoraAdapter to be relative to modelFileBase
+                request.AudioPath = os.path.join(modelFileBase, request.AudioPath)
+            if request.AudioPath != "":
+                print("Generating model", file=sys.stderr)
+                make_prompt(name=model_name, audio_prompt_path=request.AudioPath)
+                ### Use given transcript
+                ##make_prompt(name=model_name, audio_prompt_path="paimon_prompt.wav",
+                ##                transcript="Just, what was that? Paimon thought we were gonna get eaten.")
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        # Implement your logic here for the LoadModel service
+        # Replace this with your desired response
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def TTS(self, request, context):
+        model = request.model
+        print(request, file=sys.stderr)
+        try:
+            audio_array = None
+            if model != "":
+                audio_array = generate_audio(request.text, prompt=model)
+            else:
+                audio_array = generate_audio(request.text)
+            print("saving to", request.dst, file=sys.stderr)
+            # save audio to disk
+            write_wav(request.dst, SAMPLE_RATE, audio_array)
+            print("saved to", request.dst, file=sys.stderr)
+            print("tts for", file=sys.stderr)
+            print(request, file=sys.stderr)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        return backend_pb2.Result(success=True)
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+
+    serve(args.addr)

extra/grpc/vllm/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
+"""Client and server classes corresponding to protobuf-defined services."""
+import grpc
+
+import backend_pb2 as backend__pb2
+
+
+class BackendStub(object):
+    """Missing associated documentation comment in .proto file."""
+
+    def __init__(self, channel):
+        """Constructor.
+
+        Args:
+            channel: A grpc.Channel.
+        """
+        self.Health = channel.unary_unary(
+                '/backend.Backend/Health',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.Predict = channel.unary_unary(
+                '/backend.Backend/Predict',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.LoadModel = channel.unary_unary(
+                '/backend.Backend/LoadModel',
+                request_serializer=backend__pb2.ModelOptions.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.PredictStream = channel.unary_stream(
+                '/backend.Backend/PredictStream',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.Reply.FromString,
+                )
+        self.Embedding = channel.unary_unary(
+                '/backend.Backend/Embedding',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.EmbeddingResult.FromString,
+                )
+        self.GenerateImage = channel.unary_unary(
+                '/backend.Backend/GenerateImage',
+                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.AudioTranscription = channel.unary_unary(
+                '/backend.Backend/AudioTranscription',
+                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
+                response_deserializer=backend__pb2.TranscriptResult.FromString,
+                )
+        self.TTS = channel.unary_unary(
+                '/backend.Backend/TTS',
+                request_serializer=backend__pb2.TTSRequest.SerializeToString,
+                response_deserializer=backend__pb2.Result.FromString,
+                )
+        self.TokenizeString = channel.unary_unary(
+                '/backend.Backend/TokenizeString',
+                request_serializer=backend__pb2.PredictOptions.SerializeToString,
+                response_deserializer=backend__pb2.TokenizationResponse.FromString,
+                )
+        self.Status = channel.unary_unary(
+                '/backend.Backend/Status',
+                request_serializer=backend__pb2.HealthMessage.SerializeToString,
+                response_deserializer=backend__pb2.StatusResponse.FromString,
+                )
+
+
+class BackendServicer(object):
+    """Missing associated documentation comment in .proto file."""
+
+    def Health(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Predict(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def LoadModel(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def PredictStream(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Embedding(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def GenerateImage(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def AudioTranscription(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def TTS(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def TokenizeString(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def Status(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+
+def add_BackendServicer_to_server(servicer, server):
+    rpc_method_handlers = {
+            'Health': grpc.unary_unary_rpc_method_handler(
+                    servicer.Health,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'Predict': grpc.unary_unary_rpc_method_handler(
+                    servicer.Predict,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'LoadModel': grpc.unary_unary_rpc_method_handler(
+                    servicer.LoadModel,
+                    request_deserializer=backend__pb2.ModelOptions.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'PredictStream': grpc.unary_stream_rpc_method_handler(
+                    servicer.PredictStream,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.Reply.SerializeToString,
+            ),
+            'Embedding': grpc.unary_unary_rpc_method_handler(
+                    servicer.Embedding,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
+            ),
+            'GenerateImage': grpc.unary_unary_rpc_method_handler(
+                    servicer.GenerateImage,
+                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
+                    servicer.AudioTranscription,
+                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
+                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
+            ),
+            'TTS': grpc.unary_unary_rpc_method_handler(
+                    servicer.TTS,
+                    request_deserializer=backend__pb2.TTSRequest.FromString,
+                    response_serializer=backend__pb2.Result.SerializeToString,
+            ),
+            'TokenizeString': grpc.unary_unary_rpc_method_handler(
+                    servicer.TokenizeString,
+                    request_deserializer=backend__pb2.PredictOptions.FromString,
+                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
+            ),
+            'Status': grpc.unary_unary_rpc_method_handler(
+                    servicer.Status,
+                    request_deserializer=backend__pb2.HealthMessage.FromString,
+                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
+            ),
+    }
+    generic_handler = grpc.method_handlers_generic_handler(
+            'backend.Backend', rpc_method_handlers)
+    server.add_generic_rpc_handlers((generic_handler,))
+
+
+ # This class is part of an EXPERIMENTAL API.
+class Backend(object):
+    """Missing associated documentation comment in .proto file."""
+
+    @staticmethod
+    def Health(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Predict(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def LoadModel(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
+            backend__pb2.ModelOptions.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def PredictStream(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.Reply.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Embedding(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.EmbeddingResult.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def GenerateImage(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
+            backend__pb2.GenerateImageRequest.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def AudioTranscription(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
+            backend__pb2.TranscriptRequest.SerializeToString,
+            backend__pb2.TranscriptResult.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TTS(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
+            backend__pb2.TTSRequest.SerializeToString,
+            backend__pb2.Result.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def TokenizeString(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
+            backend__pb2.PredictOptions.SerializeToString,
+            backend__pb2.TokenizationResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def Status(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
+            backend__pb2.HealthMessage.SerializeToString,
+            backend__pb2.StatusResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

extra/grpc/vllm/backend_vllm.py

@@ -0,0 +1,107 @@
+#!/usr/bin/env python3
+import grpc
+from concurrent import futures
+import time
+import backend_pb2
+import backend_pb2_grpc
+import argparse
+import signal
+import sys
+import os, glob
+
+from pathlib import Path
+from vllm import LLM, SamplingParams
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    def generate(self,prompt, max_new_tokens):
+        self.generator.end_beam_search()
+
+        # Tokenizing the input
+        ids = self.generator.tokenizer.encode(prompt)
+
+        self.generator.gen_begin_reuse(ids)
+        initial_len = self.generator.sequence[0].shape[0]
+        has_leading_space = False
+        decoded_text = ''
+        for i in range(max_new_tokens):
+            token = self.generator.gen_single_token()
+            if i == 0 and self.generator.tokenizer.tokenizer.IdToPiece(int(token)).startswith('▁'):
+                has_leading_space = True
+
+            decoded_text = self.generator.tokenizer.decode(self.generator.sequence[0][initial_len:])
+            if has_leading_space:
+                decoded_text = ' ' + decoded_text
+
+            if token.item() == self.generator.tokenizer.eos_token_id:
+                break
+        return decoded_text
+    def Health(self, request, context):
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+    def LoadModel(self, request, context):
+        try:
+            if request.Quantization != "":
+                self.llm = LLM(model=request.Model, quantization=request.Quantization)
+            else:
+                self.llm = LLM(model=request.Model)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def Predict(self, request, context):
+        if request.TopP == 0:
+            request.TopP = 0.9
+
+        sampling_params = SamplingParams(temperature=request.Temperature, top_p=request.TopP)
+        outputs = self.llm.generate([request.Prompt], sampling_params)
+
+        generated_text = outputs[0].outputs[0].text
+        # Remove prompt from response if present
+        if request.Prompt in generated_text:
+            generated_text = generated_text.replace(request.Prompt, "")
+
+        return backend_pb2.Result(message=bytes(generated_text, encoding='utf-8'))
+
+    def PredictStream(self, request, context):
+        # Implement PredictStream RPC
+        #for reply in some_data_generator():
+        #    yield reply
+        # Not implemented yet
+        return self.Predict(request, context)
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+
+    serve(args.addr)

extra/requirements.txt

@@ -1,4 +1,7 @@
 sentence_transformers
 grpcio
 google
-protobuf
+protobuf
+six
+omegaconf
+compel

go.mod

@@ -1,6 +1,6 @@
 module github.com/go-skynet/LocalAI
 
-go 1.20
+go 1.21
 
 require (
 	github.com/donomii/go-rwkv.cpp v0.0.0-20230715075832-c898cd0f62df
@@ -9,33 +9,44 @@ require (
 	github.com/go-skynet/bloomz.cpp v0.0.0-20230529155654-1834e77b83fa
 	github.com/go-skynet/go-bert.cpp v0.0.0-20230716133540-6abe312cded1
 	github.com/go-skynet/go-ggml-transformers.cpp v0.0.0-20230714203132-ffb09d7dd71e
-	github.com/go-skynet/go-llama.cpp v0.0.0-20230802220037-50cee7712066
-	github.com/gofiber/fiber/v2 v2.48.0
-	github.com/google/uuid v1.3.0
+	github.com/go-skynet/go-llama.cpp v0.0.0-20230916071220-b471eb7d8c93
+	github.com/gofiber/fiber/v2 v2.49.2
+	github.com/google/uuid v1.3.1
 	github.com/hashicorp/go-multierror v1.1.1
 	github.com/hpcloud/tail v1.0.0
 	github.com/imdario/mergo v0.3.16
 	github.com/json-iterator/go v1.1.12
 	github.com/mholt/archiver/v3 v3.5.1
 	github.com/mudler/go-ggllm.cpp v0.0.0-20230709223052-862477d16eef
-	github.com/mudler/go-processmanager v0.0.0-20220724164624-c45b5c61312d
+	github.com/mudler/go-processmanager v0.0.0-20230818213616-f204007f963c
 	github.com/mudler/go-stable-diffusion v0.0.0-20230605122230-d89260f598af
-	github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230811181453-4d855afe973a
-	github.com/onsi/ginkgo/v2 v2.11.0
+	github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230916184256-e86c63750de2
+	github.com/onsi/ginkgo/v2 v2.12.1
 	github.com/onsi/gomega v1.27.10
-	github.com/otiai10/openaigo v1.5.2
+	github.com/otiai10/openaigo v1.6.0
 	github.com/phayes/freeport v0.0.0-20220201140144-74d24b5ae9f5
 	github.com/rs/zerolog v1.30.0
-	github.com/sashabaranov/go-openai v1.14.1
-	github.com/tmc/langchaingo v0.0.0-20230811231558-fd8b7f099537
+	github.com/sashabaranov/go-openai v1.15.3
+	github.com/tmc/langchaingo v0.0.0-20230923172951-2c309cf27956
 	github.com/urfave/cli/v2 v2.25.7
-	github.com/valyala/fasthttp v1.48.0
-	google.golang.org/grpc v1.57.0
+	github.com/valyala/fasthttp v1.50.0
+	google.golang.org/grpc v1.58.2
 	google.golang.org/protobuf v1.31.0
 	gopkg.in/yaml.v2 v2.4.0
 	gopkg.in/yaml.v3 v3.0.1
 )
 
+require (
+	github.com/go-ole/go-ole v1.2.6 // indirect
+	github.com/lufia/plan9stats v0.0.0-20211012122336-39d0f177ccd0 // indirect
+	github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c // indirect
+	github.com/shirou/gopsutil/v3 v3.23.8
+	github.com/shoenig/go-m1cpu v0.1.6 // indirect
+	github.com/tklauser/go-sysconf v0.3.12 // indirect
+	github.com/tklauser/numcpus v0.6.1 // indirect
+	github.com/yusufpapurcu/wmi v1.2.3 // indirect
+)
+
 require (
 	github.com/dlclark/regexp2 v1.8.1 // indirect
 	github.com/dsnet/compress v0.0.2-0.20210315054119-f66993602bf5 // indirect
@@ -50,8 +61,7 @@ require (
 	github.com/pkoukk/tiktoken-go v0.1.2 // indirect
 	github.com/ulikunitz/xz v0.5.9 // indirect
 	github.com/xi2/xz v0.0.0-20171230120015-48954b6210f8 // indirect
-	google.golang.org/genproto v0.0.0-20230526161137-0005af68ea54 // indirect
-	google.golang.org/genproto/googleapis/rpc v0.0.0-20230525234030-28d5490b6b19 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20230711160842-782d3b101e98 // indirect
 	gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c // indirect
 	gopkg.in/fsnotify.v1 v1.4.7 // indirect
 	gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 // indirect
@@ -67,18 +77,18 @@ require (
 	github.com/google/go-cmp v0.5.9 // indirect
 	github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38 // indirect
 	github.com/hashicorp/errwrap v1.0.0 // indirect
-	github.com/klauspost/compress v1.16.3 // indirect
+	github.com/klauspost/compress v1.16.7 // indirect
 	github.com/mattn/go-colorable v0.1.13 // indirect
 	github.com/mattn/go-isatty v0.0.19 // indirect
-	github.com/mattn/go-runewidth v0.0.14 // indirect
+	github.com/mattn/go-runewidth v0.0.15 // indirect
 	github.com/mudler/go-piper v0.0.0-20230621222733-56b8a81b4760
 	github.com/rivo/uniseg v0.2.0 // indirect
 	github.com/russross/blackfriday/v2 v2.1.0 // indirect
 	github.com/valyala/bytebufferpool v1.0.0 // indirect
 	github.com/valyala/tcplisten v1.0.0 // indirect
 	github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673 // indirect
-	golang.org/x/net v0.12.0 // indirect
-	golang.org/x/sys v0.10.0 // indirect
-	golang.org/x/text v0.11.0 // indirect
-	golang.org/x/tools v0.9.3 // indirect
+	golang.org/x/net v0.14.0 // indirect
+	golang.org/x/sys v0.12.0 // indirect
+	golang.org/x/text v0.12.0 // indirect
+	golang.org/x/tools v0.12.0 // indirect
 )

go.sum

@@ -13,8 +13,6 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/dlclark/regexp2 v1.8.1 h1:6Lcdwya6GjPUNsBct8Lg/yRPwMhABj269AAzdGSiR+0=
 github.com/dlclark/regexp2 v1.8.1/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
-github.com/donomii/go-rwkv.cpp v0.0.0-20230619005719-f5a8c4539674 h1:G70Yf/QOCEL1v24idWnGd6rJsbqiGkJAJnMaWaolzEg=
-github.com/donomii/go-rwkv.cpp v0.0.0-20230619005719-f5a8c4539674/go.mod h1:gWy7FIWioqYmYxkaoFyBnaKApeZVrUkHhv9EV9pz4dM=
 github.com/donomii/go-rwkv.cpp v0.0.0-20230715075832-c898cd0f62df h1:qVcBEZlvp5A1gGWNJj02xyDtbsUI2hohlQMSB1fgER4=
 github.com/donomii/go-rwkv.cpp v0.0.0-20230715075832-c898cd0f62df/go.mod h1:gWy7FIWioqYmYxkaoFyBnaKApeZVrUkHhv9EV9pz4dM=
 github.com/dsnet/compress v0.0.2-0.20210315054119-f66993602bf5 h1:iFaUwBSo5Svw6L7HYpRu/0lE3e0BaElwnNO1qkNQxBY=
@@ -33,28 +31,22 @@ github.com/go-audio/wav v1.1.0 h1:jQgLtbqBzY7G+BM8fXF7AHUk1uHUviWS4X39d5rsL2g=
 github.com/go-audio/wav v1.1.0/go.mod h1:mpe9qfwbScEbkd8uybLuIpTgHyrISw/OTuvjUW2iGtE=
 github.com/go-logr/logr v1.2.4 h1:g01GSCwiDw2xSZfjJ2/T9M+S6pFdcNtFYsp+Y43HYDQ=
 github.com/go-logr/logr v1.2.4/go.mod h1:jdQByPbusPIv2/zmleS9BjJVeZ6kBagPoEUsqbVz/1A=
+github.com/go-ole/go-ole v1.2.6 h1:/Fpf6oFPoeFik9ty7siob0G6Ke8QvQEuVcuChpwXzpY=
+github.com/go-ole/go-ole v1.2.6/go.mod h1:pprOEPIfldk/42T2oK7lQ4v4JSDwmV0As9GaiUsvbm0=
 github.com/go-skynet/bloomz.cpp v0.0.0-20230529155654-1834e77b83fa h1:gxr68r/6EWroay4iI81jxqGCDbKotY4+CiwdUkBz2NQ=
 github.com/go-skynet/bloomz.cpp v0.0.0-20230529155654-1834e77b83fa/go.mod h1:wc0fJ9V04yiYTfgKvE5RUUSRQ5Kzi0Bo4I+U3nNOUuA=
 github.com/go-skynet/go-bert.cpp v0.0.0-20230716133540-6abe312cded1 h1:yXvc7QfGtoZ51tUW/YVjoTwAfh8HG88XU7UOrbNlz5Y=
 github.com/go-skynet/go-bert.cpp v0.0.0-20230716133540-6abe312cded1/go.mod h1:fYjkCDRzC+oRLHSjQoajmYK6AmeJnmEanV27CClAcDc=
 github.com/go-skynet/go-ggml-transformers.cpp v0.0.0-20230714203132-ffb09d7dd71e h1:4reMY29i1eOZaRaSTMPNyXI7X8RMNxCTfDDBXYzrbr0=
 github.com/go-skynet/go-ggml-transformers.cpp v0.0.0-20230714203132-ffb09d7dd71e/go.mod h1:31j1odgFXP8hDSUVfH0zErKI5aYVP18ddYnPkwCso2A=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230709163512-6c97625cca76 h1:NRdxo2MKi8qhWZXxu6CIZOkdH+LBERFz1kk22U1FD3k=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230709163512-6c97625cca76/go.mod h1:tzi97YvT1bVQ+iTG39LvpDkKG1WbizgtljC+orSoM40=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230724222459-562d2b5a7119 h1:FeUSk5yMHT7J7jeCQKAOs4x5LRNSYH0SR6djM/i1jcc=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230724222459-562d2b5a7119/go.mod h1:fiJBto+Le1XLtD/cID5SAKs8cKE7wFXJKfTT3wvPQRA=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230727163958-6ba16de8e965 h1:2MO/rABKpkXnnKQ3Ar90aqhnlMEejE9gnKG6bafv+ow=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230727163958-6ba16de8e965/go.mod h1:fiJBto+Le1XLtD/cID5SAKs8cKE7wFXJKfTT3wvPQRA=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230729200103-8c51308e42d7 h1:1uBwholTaJ8Lva8ySJjT4jNaCDAh+MJXtsbZBbQq9lA=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230729200103-8c51308e42d7/go.mod h1:fiJBto+Le1XLtD/cID5SAKs8cKE7wFXJKfTT3wvPQRA=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230802220037-50cee7712066 h1:v4Js+yEdgY9IV7n35M+5MELLxlOMp3qC5whZm5YTLjI=
-github.com/go-skynet/go-llama.cpp v0.0.0-20230802220037-50cee7712066/go.mod h1:fiJBto+Le1XLtD/cID5SAKs8cKE7wFXJKfTT3wvPQRA=
+github.com/go-skynet/go-llama.cpp v0.0.0-20230916071220-b471eb7d8c93 h1:aJQn0Lb5TQtSXw5hK+kAUuaUjsXwDXyEItzI9U+GiOA=
+github.com/go-skynet/go-llama.cpp v0.0.0-20230916071220-b471eb7d8c93/go.mod h1:VGPpMv2UMxRgBiDIFrs7AGMzrX884K0Vy8tqYu3if/Q=
 github.com/go-task/slim-sprig v0.0.0-20210107165309-348f09dbbbc0/go.mod h1:fyg7847qk6SyHyPtNmDHnmrv/HOrqktSC+C9fM+CJOE=
 github.com/go-task/slim-sprig v0.0.0-20230315185526-52ccab3ef572 h1:tfuBGBXKqDEevZMzYi5KSi8KkcZtzBcTgAUUtapy0OI=
 github.com/go-task/slim-sprig v0.0.0-20230315185526-52ccab3ef572/go.mod h1:9Pwr4B2jHnOSGXyyzV8ROjYa2ojvAY6HCGYYfMoC3Ls=
 github.com/godbus/dbus/v5 v5.0.4/go.mod h1:xhWf0FNVPg57R7Z0UbKHbJfkEywrmjJnf7w5xrFpKfA=
-github.com/gofiber/fiber/v2 v2.48.0 h1:cRVMCb9aUJDsyHxGFLwz/sGzDggdailZZyptU9F9cU0=
-github.com/gofiber/fiber/v2 v2.48.0/go.mod h1:xqJgfqrc23FJuqGOW6DVgi3HyZEm2Mn9pRqUb2kHSX8=
+github.com/gofiber/fiber/v2 v2.49.2 h1:ONEN3/Vc+dUCxxDgZZwpqvhISgHqb+bu+isBiEyKEQs=
+github.com/gofiber/fiber/v2 v2.49.2/go.mod h1:gNsKnyrmfEWFpJxQAV0qvW6l70K1dZGno12oLtukcts=
 github.com/golang/protobuf v1.2.0/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
 github.com/golang/protobuf v1.4.0-rc.1/go.mod h1:ceaxUfeHdC40wWswd/P6IGgMaK3YpKi5j83Wpe3EHw8=
 github.com/golang/protobuf v1.4.0-rc.1.0.20200221234624-67d41d38c208/go.mod h1:xKAWHe0F5eneWXFV3EuXVDTCmh+JuBKY0li0aMyXATA=
@@ -72,13 +64,14 @@ github.com/google/go-cmp v0.3.0/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMyw
 github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
 github.com/google/go-cmp v0.4.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.5.6/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.9 h1:O2Tfq5qg4qc4AmwVlvv0oLiVAGB7enBSJ2x2DqQFi38=
 github.com/google/go-cmp v0.5.9/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38 h1:yAJXTCF9TqKcTiHJAE8dj7HMvPfh66eeA2JYW7eFpSE=
 github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38/go.mod h1:kpwsk12EmLew5upagYY7GY0pfYCcupk39gWOCRROcvE=
-github.com/google/uuid v1.3.0 h1:t6JiXgmwXMjEs8VusXIJk2BXHsn+wx8BZdTaoZ5fu7I=
-github.com/google/uuid v1.3.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/google/uuid v1.3.1 h1:KjJaJ9iWZ3jOFZIf1Lqf4laDRCasjl0BCmnEGxkdLb4=
+github.com/google/uuid v1.3.1/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/hashicorp/errwrap v1.0.0 h1:hLrqtEDnRye3+sgx6z4qVLNuviH3MR5aQ0ykNJa/UYA=
 github.com/hashicorp/errwrap v1.0.0/go.mod h1:YH+1FKiLXxHSkmPseP+kNlulaMuP3n2brvKWEqk/Jc4=
 github.com/hashicorp/go-multierror v1.1.1 h1:H5DkEtf6CXdFp0N0Em5UCwQpXMWke8IA0+lD48awMYo=
@@ -92,8 +85,8 @@ github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnr
 github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo=
 github.com/klauspost/compress v1.4.1/go.mod h1:RyIbtBH6LamlWaDj8nUwkbUhJ87Yi3uG0guNDohfE1A=
 github.com/klauspost/compress v1.11.4/go.mod h1:aoV0uJVorq1K+umq18yTdKaF57EivdYsUV+/s2qKfXs=
-github.com/klauspost/compress v1.16.3 h1:XuJt9zzcnaz6a16/OU53ZjWp/v7/42WcR5t2a0PcNQY=
-github.com/klauspost/compress v1.16.3/go.mod h1:ntbaceVETuRiXiv4DpjP66DpAtAGkEQskQzEyD//IeE=
+github.com/klauspost/compress v1.16.7 h1:2mk3MPGNzKyxErAw8YaohYh69+pa4sIQSC0fPGCFR9I=
+github.com/klauspost/compress v1.16.7/go.mod h1:ntbaceVETuRiXiv4DpjP66DpAtAGkEQskQzEyD//IeE=
 github.com/klauspost/cpuid v1.2.0/go.mod h1:Pj4uuM528wm8OyEC2QMXAi2YiTZ96dNQPGgoMS4s3ek=
 github.com/klauspost/pgzip v1.2.5 h1:qnWYvvKqedOF2ulHpMG72XQol4ILEJ8k2wwRl/Km8oE=
 github.com/klauspost/pgzip v1.2.5/go.mod h1:Ch1tH69qFZu15pkjo5kYi6mth2Zzwzt50oCQKQE9RUs=
@@ -103,6 +96,8 @@ github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
 github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
 github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
+github.com/lufia/plan9stats v0.0.0-20211012122336-39d0f177ccd0 h1:6E+4a0GO5zZEnZ81pIr0yLvtUWk2if982qA3F3QD6H4=
+github.com/lufia/plan9stats v0.0.0-20211012122336-39d0f177ccd0/go.mod h1:zJYVVT2jmtg6P3p1VtQj7WsuWi/y4VnjVBn7F8KPB3I=
 github.com/mattn/go-colorable v0.1.12/go.mod h1:u5H1YNBxpqRaxsYJYSkiCWKzEfiAb1Gb520KVy5xxl4=
 github.com/mattn/go-colorable v0.1.13 h1:fFA4WZxdEF4tXPZVKMLwD8oUnCTTo08duU7wxecdEvA=
 github.com/mattn/go-colorable v0.1.13/go.mod h1:7S9/ev0klgBDR4GtXTXX8a3vIGJpMovkB8vQcUbaXHg=
@@ -110,8 +105,8 @@ github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27k
 github.com/mattn/go-isatty v0.0.16/go.mod h1:kYGgaQfpe5nmfYZH+SKPsOc2e4SrIfOl2e/yFXSvRLM=
 github.com/mattn/go-isatty v0.0.19 h1:JITubQf0MOLdlGRuRq+jtsDlekdYPia9ZFsB8h/APPA=
 github.com/mattn/go-isatty v0.0.19/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
-github.com/mattn/go-runewidth v0.0.14 h1:+xnbZSEeDbOIg5/mE6JF0w6n9duR1l3/WmbinWVwUuU=
-github.com/mattn/go-runewidth v0.0.14/go.mod h1:Jdepj2loyihRzMpdS35Xk/zdY8IAYHsh153qUoGf23w=
+github.com/mattn/go-runewidth v0.0.15 h1:UNAjwbU9l54TA3KzvqLGxwWjHmMgBUVhBiTjelZgg3U=
+github.com/mattn/go-runewidth v0.0.15/go.mod h1:Jdepj2loyihRzMpdS35Xk/zdY8IAYHsh153qUoGf23w=
 github.com/mholt/archiver/v3 v3.5.1 h1:rDjOBX9JSF5BvoJGvjqK479aL70qh9DIpZCl+k7Clwo=
 github.com/mholt/archiver/v3 v3.5.1/go.mod h1:e3dqJ7H78uzsRSEACH1joayhuSyhnonssnDhppzS1L4=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421 h1:ZqeYNhU3OHLH3mGKHDcjJRFFRrJa6eAM5H+CtDdOsPc=
@@ -122,24 +117,12 @@ github.com/mudler/go-ggllm.cpp v0.0.0-20230709223052-862477d16eef h1:OJZtJ5vYhlk
 github.com/mudler/go-ggllm.cpp v0.0.0-20230709223052-862477d16eef/go.mod h1:00giAi/vwF8LX29JBjkPQhtASsivPnGNzB6sdmk8JGE=
 github.com/mudler/go-piper v0.0.0-20230621222733-56b8a81b4760 h1:OFVkSxR7CRSRSNm5dvpMRZwmSwWa8EMMnHbc84fW5tU=
 github.com/mudler/go-piper v0.0.0-20230621222733-56b8a81b4760/go.mod h1:O7SwdSWMilAWhBZMK9N9Y/oBDyMMzshE3ju8Xkexwig=
-github.com/mudler/go-processmanager v0.0.0-20220724164624-c45b5c61312d h1:/lAg9vPAAU+s35cDMCx1IyeMn+4OYfCBPqi08Q8vXDg=
-github.com/mudler/go-processmanager v0.0.0-20220724164624-c45b5c61312d/go.mod h1:HGGAOJhipApckwNV8ZTliRJqxctUv3xRY+zbQEwuytc=
+github.com/mudler/go-processmanager v0.0.0-20230818213616-f204007f963c h1:CI5uGwqBpN8N7BrSKC+nmdfw+9nPQIDyjHHlaIiitZI=
+github.com/mudler/go-processmanager v0.0.0-20230818213616-f204007f963c/go.mod h1:gY3wyrhkRySJtmtI/JPt4a2mKv48h/M9pEZIW+SjeC0=
 github.com/mudler/go-stable-diffusion v0.0.0-20230605122230-d89260f598af h1:XFq6OUqsWQam0OrEr05okXsJK/TQur3zoZTHbiZD3Ks=
 github.com/mudler/go-stable-diffusion v0.0.0-20230605122230-d89260f598af/go.mod h1:8ufRkpz/S/9ahkaxzZ5i4WMgO9w4InEhuRoT7vK5Rnw=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230714185456-cfd70b69fcf5 h1:bmQnxyKiqCu8i2y/N/Sf0coWoG2/Ed12YGQeb7lTnjo=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230714185456-cfd70b69fcf5/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230725212419-9100b2ef6fb9 h1:/oRwZhulKTU8LpPD2fXi2o2kdlTutQjYWDVMkrv14po=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230725212419-9100b2ef6fb9/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230727161923-39acbc837816 h1:hRi7hpDUuaO0dB4NZ8eyaeD2fRar6CPyNAARsO5DhzA=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230727161923-39acbc837816/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230731161838-cbdcde8b7586 h1:WVEMSZMyHFe68PN204c3Fdk5g2lZouPvbU9/2zkPpWc=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230731161838-cbdcde8b7586/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230802145814-c449b71b56de h1:E5EGczxEAcbaO8yqj074MQxU609QbtB6in3qTOW1EFo=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230802145814-c449b71b56de/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230807175413-0f2bb506a8ee h1:Y/j+GNytyncmDnAEuDZwzkYC9nzUPvXJPF+nntQG0VU=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230807175413-0f2bb506a8ee/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230811181453-4d855afe973a h1:bX26Zfwh72ug2aZTEwFISTMEJ56Wa/4KqboidD+g92A=
-github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230811181453-4d855afe973a/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
+github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230916184256-e86c63750de2 h1:nBbgd4ORk4pubuh/kugHw5Sdytt+SxtJLEQ+vJfK6Ko=
+github.com/nomic-ai/gpt4all/gpt4all-bindings/golang v0.0.0-20230916184256-e86c63750de2/go.mod h1:4T3CHXyrt+7FQHXaxULZfPjHbD8/99WuDDJa0YVZARI=
 github.com/nwaples/rardecode v1.1.0 h1:vSxaY8vQhOcVr4mm5e8XllHWTiM4JF507A0Katqw7MQ=
 github.com/nwaples/rardecode v1.1.0/go.mod h1:5DzqNKiOdpKKBH87u8VlvAnPZMXcGRhxWkRpHbbfGS0=
 github.com/nxadm/tail v1.4.4/go.mod h1:kenIhsEOeOJmVchQTgglprH7qJGnHDVpk1VPCcaMI8A=
@@ -149,18 +132,17 @@ github.com/onsi/ginkgo v1.6.0/go.mod h1:lLunBs/Ym6LB5Z9jYTR76FiuTmxDTDusOGeTQH+W
 github.com/onsi/ginkgo v1.12.1/go.mod h1:zj2OWP4+oCPe1qIXoGWkgMRwljMUYCdkwsT2108oapk=
 github.com/onsi/ginkgo v1.16.4 h1:29JGrr5oVBm5ulCWet69zQkzWipVXIol6ygQUe/EzNc=
 github.com/onsi/ginkgo v1.16.4/go.mod h1:dX+/inL/fNMqNlz0e9LfyB9TswhZpCVdJM/Z6Vvnwo0=
-github.com/onsi/ginkgo/v2 v2.11.0 h1:WgqUCUt/lT6yXoQ8Wef0fsNn5cAuMK7+KT9UFRz2tcU=
-github.com/onsi/ginkgo/v2 v2.11.0/go.mod h1:ZhrRA5XmEE3x3rhlzamx/JJvujdZoJ2uvgI7kR0iZvM=
+github.com/onsi/ginkgo/v2 v2.12.1 h1:uHNEO1RP2SpuZApSkel9nEh1/Mu+hmQe7Q+Pepg5OYA=
+github.com/onsi/ginkgo/v2 v2.12.1/go.mod h1:TE309ZR8s5FsKKpuB1YAQYBzCaAfUgatB/xlT/ETL/o=
 github.com/onsi/gomega v1.7.1/go.mod h1:XdKZgCCFLUoM/7CFJVPcG8C1xQ1AJ0vpAezJrB7JYyY=
 github.com/onsi/gomega v1.10.1/go.mod h1:iN09h71vgCQne3DLsj+A5owkum+a2tYe+TOCB1ybHNo=
 github.com/onsi/gomega v1.16.0/go.mod h1:HnhC7FXeEQY45zxNK3PPoIUhzk/80Xly9PcubAlGdZY=
-github.com/onsi/gomega v1.27.8 h1:gegWiwZjBsf2DgiSbf5hpokZ98JVDMcWkUiigk6/KXc=
-github.com/onsi/gomega v1.27.8/go.mod h1:2J8vzI/s+2shY9XHRApDkdgPo1TKT7P2u6fXeJKFnNQ=
 github.com/onsi/gomega v1.27.10 h1:naR28SdDFlqrG6kScpT8VWpu1xWY5nJRCF3XaYyBjhI=
 github.com/onsi/gomega v1.27.10/go.mod h1:RsS8tutOdbdgzbPtzzATp12yT7kM5I5aElG3evPbQ0M=
 github.com/otiai10/mint v1.6.1 h1:kgbTJmOpp/0ce7hk3H8jiSuR0MXmpwWRfqUdKww17qg=
-github.com/otiai10/openaigo v1.5.2 h1:YnNDisZmA4syArF3IxMCIrfgZOq30PLV219gPY7n2z8=
-github.com/otiai10/openaigo v1.5.2/go.mod h1:kIaXc3V+Xy5JLplcBxehVyGYDtufHp3PFPy04jOwOAI=
+github.com/otiai10/mint v1.6.1/go.mod h1:MJm72SBthJjz8qhefc4z1PYEieWmy8Bku7CjcAqyUSM=
+github.com/otiai10/openaigo v1.6.0 h1:YTQEbtDSvawETOB/Kmb/6JvuHdHH/eIpSQfHVufiwY8=
+github.com/otiai10/openaigo v1.6.0/go.mod h1:kIaXc3V+Xy5JLplcBxehVyGYDtufHp3PFPy04jOwOAI=
 github.com/phayes/freeport v0.0.0-20220201140144-74d24b5ae9f5 h1:Ii+DKncOVM8Cu1Hc+ETb5K+23HdAMvESYE3ZJ5b5cMI=
 github.com/phayes/freeport v0.0.0-20220201140144-74d24b5ae9f5/go.mod h1:iIss55rKnNBTvrwdmkUpLnDpZoAHvWaiq5+iMmen4AE=
 github.com/pierrec/lz4/v4 v4.1.2 h1:qvY3YFXRQE/XB8MlLzJH7mSzBs74eA2gg52YTk6jUPM=
@@ -170,37 +152,44 @@ github.com/pkoukk/tiktoken-go v0.1.2 h1:u7PCSBiWJ3nJYoTGShyM9iHXz4dNyYkurwwp+GHt
 github.com/pkoukk/tiktoken-go v0.1.2/go.mod h1:boMWvk9pQCOTx11pgu0DrIdrAKgQzzJKUP6vLXaz7Rw=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c h1:ncq/mPwQF4JjgDlrVEn3C11VoGHZN7m8qihwgMEtzYw=
+github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/rivo/uniseg v0.2.0 h1:S1pD9weZBuJdFmowNwbpi7BJ8TNftyUImj/0WQi72jY=
 github.com/rivo/uniseg v0.2.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=
-github.com/rs/xid v1.4.0/go.mod h1:trrq9SKmegXys3aeAKXMUTdJsYXVwGY3RLcfgqegfbg=
 github.com/rs/xid v1.5.0/go.mod h1:trrq9SKmegXys3aeAKXMUTdJsYXVwGY3RLcfgqegfbg=
-github.com/rs/zerolog v1.29.1 h1:cO+d60CHkknCbvzEWxP0S9K6KqyTjrCNUy1LdQLCGPc=
-github.com/rs/zerolog v1.29.1/go.mod h1:Le6ESbR7hc+DP6Lt1THiV8CQSdkkNrd3R0XbEgp3ZBU=
 github.com/rs/zerolog v1.30.0 h1:SymVODrcRsaRaSInD9yQtKbtWqwsfoPcRff/oRXLj4c=
 github.com/rs/zerolog v1.30.0/go.mod h1:/tk+P47gFdPXq4QYjvCmT5/Gsug2nagsFWBWhAiSi1w=
 github.com/russross/blackfriday/v2 v2.1.0 h1:JIOH55/0cWyOuilr9/qlrm0BSXldqnqwMsf35Ld67mk=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
-github.com/sashabaranov/go-openai v1.14.0 h1:D1yAB+DHElgbJFdYyjxfTWMFzhddn+PwZmkQ039L7mQ=
-github.com/sashabaranov/go-openai v1.14.0/go.mod h1:lj5b/K+zjTSFxVLijLSTDZuP7adOgerWeFyZLUhAKRg=
-github.com/sashabaranov/go-openai v1.14.1 h1:jqfkdj8XHnBF84oi2aNtT8Ktp3EJ0MfuVjvcMkfI0LA=
-github.com/sashabaranov/go-openai v1.14.1/go.mod h1:lj5b/K+zjTSFxVLijLSTDZuP7adOgerWeFyZLUhAKRg=
+github.com/sashabaranov/go-openai v1.15.3 h1:rzoNK9n+Cak+PM6OQ9puxDmFllxfnVea9StlmhglXqA=
+github.com/sashabaranov/go-openai v1.15.3/go.mod h1:lj5b/K+zjTSFxVLijLSTDZuP7adOgerWeFyZLUhAKRg=
+github.com/shirou/gopsutil/v3 v3.23.7/go.mod h1:c4gnmoRC0hQuaLqvxnx1//VXQ0Ms/X9UnJF8pddY5z4=
+github.com/shirou/gopsutil/v3 v3.23.8 h1:xnATPiybo6GgdRoC4YoGnxXZFRc3dqQTGi73oLvvBrE=
+github.com/shirou/gopsutil/v3 v3.23.8/go.mod h1:7hmCaBn+2ZwaZOr6jmPBZDfawwMGuo1id3C6aM8EDqQ=
+github.com/shoenig/go-m1cpu v0.1.6 h1:nxdKQNcEB6vzgA2E2bvzKIYRuNj7XNJ4S/aRSwKzFtM=
+github.com/shoenig/go-m1cpu v0.1.6/go.mod h1:1JJMcUBvfNwpq05QDQVAnx3gUHr9IYF7GNg9SUEw2VQ=
+github.com/shoenig/test v0.6.4 h1:kVTaSd7WLz5WZ2IaoM0RSzRsUD+m8wRR+5qvntpn4LU=
+github.com/shoenig/test v0.6.4/go.mod h1:byHiCGXqrVaflBLAMq/srcZIHynQPQgeyvkvXnjqq0k=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/objx v0.4.0/go.mod h1:YvHI0jy2hoMjB+UWwv71VJQ9isScKT/TqJzVSSt89Yw=
+github.com/stretchr/objx v0.5.0/go.mod h1:Yh+to48EsGEfYuaHDzXPcE3xhTkx73EhmCGUpEOglKo=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
 github.com/stretchr/testify v1.5.1/go.mod h1:5W2xD1RspED5o8YsWQXVCued0rvSQ+mT+I5cxcmMvtA=
 github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
-github.com/stretchr/testify v1.8.2 h1:+h33VjcLVPDHtOdpUCuF+7gSuG3yGIftsP1YvFihtJ8=
-github.com/tmc/langchaingo v0.0.0-20230713201705-dcf7ecdc8ac8 h1:wdJigYmmIRCuXhCkADDr53Oa1fp/WlxCPoVXR2r7GrU=
-github.com/tmc/langchaingo v0.0.0-20230713201705-dcf7ecdc8ac8/go.mod h1:mTzgQfAGwmBz2hhQELZfu2bwsbHwyKHA6IHOa+9LDFg=
-github.com/tmc/langchaingo v0.0.0-20230726025230-7d5f9fd5e90a h1:I/2JSuYXkWaVVLSZmrPfrgbvvvPR0IaulZcB0Iu8oVI=
-github.com/tmc/langchaingo v0.0.0-20230726025230-7d5f9fd5e90a/go.mod h1:8T+nNIGBr3nYQEYFmF/YaT8t8YTKLvFYZBuVZOAYn5E=
-github.com/tmc/langchaingo v0.0.0-20230729232647-7df4fe5fb8fe h1:+XVrCjh3rPibfISkUFG2Ck5NLKODQ9cFdmraFye1bGA=
-github.com/tmc/langchaingo v0.0.0-20230729232647-7df4fe5fb8fe/go.mod h1:8T+nNIGBr3nYQEYFmF/YaT8t8YTKLvFYZBuVZOAYn5E=
-github.com/tmc/langchaingo v0.0.0-20230731024823-8f101609f600 h1:SABuIthjhIXEsxnokuA16CZOxxdW9XohIHQqd/go8Nc=
-github.com/tmc/langchaingo v0.0.0-20230731024823-8f101609f600/go.mod h1:8T+nNIGBr3nYQEYFmF/YaT8t8YTKLvFYZBuVZOAYn5E=
-github.com/tmc/langchaingo v0.0.0-20230802030916-271e9bd7e7c5 h1:js7vYDJGzUGVSt0YlIusUc5BXYVECu3LUI/asby5Ggo=
-github.com/tmc/langchaingo v0.0.0-20230802030916-271e9bd7e7c5/go.mod h1:8T+nNIGBr3nYQEYFmF/YaT8t8YTKLvFYZBuVZOAYn5E=
-github.com/tmc/langchaingo v0.0.0-20230811231558-fd8b7f099537 h1:vkeNjlW+0Xiw2XizMHoQuLG8pg6AN1hU8zJuMV9GQBc=
-github.com/tmc/langchaingo v0.0.0-20230811231558-fd8b7f099537/go.mod h1:8T+nNIGBr3nYQEYFmF/YaT8t8YTKLvFYZBuVZOAYn5E=
+github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
+github.com/stretchr/testify v1.8.4 h1:CcVxjf3Q8PM0mHUKJCdn+eZZtm5yQwehR5yeSVQQcUk=
+github.com/stretchr/testify v1.8.4/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo=
+github.com/tklauser/go-sysconf v0.3.11/go.mod h1:GqXfhXY3kiPa0nAXPDIQIWzJbMCB7AmcWpGR8lSZfqI=
+github.com/tklauser/go-sysconf v0.3.12 h1:0QaGUFOdQaIVdPgfITYzaTegZvdCjmYO52cSFAEVmqU=
+github.com/tklauser/go-sysconf v0.3.12/go.mod h1:Ho14jnntGE1fpdOqQEEaiKRpvIavV0hSfmBq8nJbHYI=
+github.com/tklauser/numcpus v0.6.0/go.mod h1:FEZLMke0lhOUG6w2JadTzp0a+Nl8PF/GFkQ5UVIcaL4=
+github.com/tklauser/numcpus v0.6.1 h1:ng9scYS7az0Bk4OZLvrNXNSAO2Pxr1XXRAPyjhIx+Fk=
+github.com/tklauser/numcpus v0.6.1/go.mod h1:1XfjsgE2zo8GVw7POkMbHENHzVg3GzmoZ9fESEdAacY=
+github.com/tmc/langchaingo v0.0.0-20230910230029-9c8845b2b019 h1:hKNvPanCqhpVszNBKYBzwknU2iJhhE29P04wpdpu/uY=
+github.com/tmc/langchaingo v0.0.0-20230910230029-9c8845b2b019/go.mod h1:vCdA1t5qnS5YPkDsznowOziBHFn0Ul11ZqfJ2GOAi0s=
+github.com/tmc/langchaingo v0.0.0-20230923172951-2c309cf27956 h1:N6H+Jd6Hf9vL57N25s0nHYYouDj4wUpLDd6rMNDfU+Y=
+github.com/tmc/langchaingo v0.0.0-20230923172951-2c309cf27956/go.mod h1:vCdA1t5qnS5YPkDsznowOziBHFn0Ul11ZqfJ2GOAi0s=
 github.com/ulikunitz/xz v0.5.8/go.mod h1:nbz6k7qbPmH4IRqmfOplQw/tblSgqTqBwxkY0oWt/14=
 github.com/ulikunitz/xz v0.5.9 h1:RsKRIA2MO8x56wkkcd3LbtcE/uMszhb6DpRf+3uwa3I=
 github.com/ulikunitz/xz v0.5.9/go.mod h1:nbz6k7qbPmH4IRqmfOplQw/tblSgqTqBwxkY0oWt/14=
@@ -208,8 +197,8 @@ github.com/urfave/cli/v2 v2.25.7 h1:VAzn5oq403l5pHjc4OhD54+XGO9cdKVL/7lDjF+iKUs=
 github.com/urfave/cli/v2 v2.25.7/go.mod h1:8qnjx1vcq5s2/wpsqoZFndg2CE5tNFyrTvS6SinrnYQ=
 github.com/valyala/bytebufferpool v1.0.0 h1:GqA5TC/0021Y/b9FG4Oi9Mr3q7XYx6KllzawFIhcdPw=
 github.com/valyala/bytebufferpool v1.0.0/go.mod h1:6bBcMArwyJ5K/AmCkWv1jt77kVWyCJ6HpOuEn7z0Csc=
-github.com/valyala/fasthttp v1.48.0 h1:oJWvHb9BIZToTQS3MuQ2R3bJZiNSa2KiNdeI8A+79Tc=
-github.com/valyala/fasthttp v1.48.0/go.mod h1:k2zXd82h/7UZc3VOdJ2WaUqt1uZ/XpXAfE9i+HBC3lA=
+github.com/valyala/fasthttp v1.50.0 h1:H7fweIlBm0rXLs2q0XbalvJ6r0CUPFWK3/bB4N13e9M=
+github.com/valyala/fasthttp v1.50.0/go.mod h1:k2zXd82h/7UZc3VOdJ2WaUqt1uZ/XpXAfE9i+HBC3lA=
 github.com/valyala/tcplisten v1.0.0 h1:rBHj/Xf+E1tRGZyWIWwJDiRY0zc1Js+CV5DqwacVSA8=
 github.com/valyala/tcplisten v1.0.0/go.mod h1:T0xQ8SeCZGxckz9qRXTfG43PvQ/mcWh7FwZEA7Ioqkc=
 github.com/xi2/xz v0.0.0-20171230120015-48954b6210f8 h1:nIPpBwaJSVYIxUFsDv3M8ofmx9yWTog9BfvIu0q41lo=
@@ -217,21 +206,22 @@ github.com/xi2/xz v0.0.0-20171230120015-48954b6210f8/go.mod h1:HUYIGzjTL3rfEspMx
 github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673 h1:bAn7/zixMGCfxrRTfdpNzjtPYqr8smhKouy9mxVdGPU=
 github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673/go.mod h1:N3UwUGtsrSj3ccvlPHLoLsHnpR27oXr4ZE984MbSER8=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
+github.com/yusufpapurcu/wmi v1.2.3 h1:E1ctvB7uKFMOJw3fdOW32DwGE9I7t++CRUEMKvFoFiw=
+github.com/yusufpapurcu/wmi v1.2.3/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQmPyzfmi0=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/mod v0.3.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
-golang.org/x/mod v0.10.0 h1:lFO9qtOdlre5W1jxS3r/4szv2/6iXxScdzjoBMXNhYk=
+golang.org/x/mod v0.12.0 h1:rmsUpXtvNzj340zd98LZ4KntptpfRHwpFOHG188oHXc=
+golang.org/x/mod v0.12.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
 golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20200520004742-59133d7f0dd7/go.mod h1:qpuaurCH72eLCgpAm/N6yyVIVM9cpaDIP3A8BGJEC5A=
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20210428140749-89ef3d95e781/go.mod h1:OJAsFXCWl8Ukc7SiCT/9KSuxbyM7479/AVlXFRxuMCk=
-golang.org/x/net v0.10.0 h1:X2//UzNDwYmtCLn7To6G58Wr6f5ahEAQgKNzv9Y951M=
-golang.org/x/net v0.10.0/go.mod h1:0qNGK6F8kojg2nk9dLZ2mShWaEBan6FAoqfSigmmuDg=
-golang.org/x/net v0.12.0 h1:cfawfvKITfUsFCeJIHJrbSxpeu/E81khclypR0GVT50=
-golang.org/x/net v0.12.0/go.mod h1:zEVYFnQC7m/vmpQFELhcD1EWkZlX69l4oqgmer6hfKA=
+golang.org/x/net v0.14.0 h1:BONx9s002vGdD9umnlX1Po8vOZmrgH34qlHcD1MfK14=
+golang.org/x/net v0.14.0/go.mod h1:PpSgVXXLK0OxS0F31C1/tv6XNguvCrnXIDrFMspZIUI=
 golang.org/x/sync v0.0.0-20180314180146-1d60e4601c6f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
@@ -239,47 +229,47 @@ golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5h
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190904154756-749cb33beabd/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20191005200804-aed5e4c7ecf9/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20191120155948-bd437916bb0e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200323222414-85ca7c5b95cd/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20201204225414-ed752295db88/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210112080510-489259a85091/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210423082822-04245dca01da/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210927094055-39ccf1dd6fa6/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220811171246-fbc7d0a398ab/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.10.0 h1:SqMFp9UcQJZa+pmYuAKjd9xq1f0j5rLcDIk0mj4qAsA=
+golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.12.0 h1:CM0HF96J0hcLAwsHPJZjfdNzs0gftsLfgKt57wWHJ0o=
+golang.org/x/sys v0.12.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.9.0 h1:2sjJmO8cDvYveuX97RDLsxlyUxLl+GHoLxBiRdHllBE=
-golang.org/x/text v0.9.0/go.mod h1:e1OnstbJyHTd6l/uOt8jFFHp6TRDWZR/bV3emEE/zU8=
-golang.org/x/text v0.11.0 h1:LAntKIrcmeSKERyiOh0XMV39LXS8IE9UL2yP7+f5ij4=
-golang.org/x/text v0.11.0/go.mod h1:TvPlkZtksWOMsz7fbANvkp4WM8x/WCo/om8BMLbz+aE=
+golang.org/x/text v0.12.0 h1:k+n5B8goJNdU7hSvEtMUz3d1Q6D/XW4COJSJR6fN0mc=
+golang.org/x/text v0.12.0/go.mod h1:TvPlkZtksWOMsz7fbANvkp4WM8x/WCo/om8BMLbz+aE=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20201224043029-2b0845dc783e/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
-golang.org/x/tools v0.9.3 h1:Gn1I8+64MsuTb/HpH+LmQtNas23LhUVr3rYZ0eKuaMM=
-golang.org/x/tools v0.9.3/go.mod h1:owI94Op576fPu3cIGQeHs3joujW/2Oc6MtlxbF5dfNc=
+golang.org/x/tools v0.12.0 h1:YW6HUoUmYBpwSgyaGaZq1fHjrBjX1rlpZ54T6mu2kss=
+golang.org/x/tools v0.12.0/go.mod h1:Sc0INKfu04TlqNoRA1hgpFZbhYXHPr4V5DzpSBTPqQM=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-google.golang.org/genproto v0.0.0-20230410155749-daa745c078e1 h1:KpwkzHKEF7B9Zxg18WzOa7djJ+Ha5DzthMyZYQfEn2A=
-google.golang.org/genproto v0.0.0-20230410155749-daa745c078e1/go.mod h1:nKE/iIaLqn2bQwXBg8f1g2Ylh6r5MN5CmZvuzZCgsCU=
-google.golang.org/genproto v0.0.0-20230526161137-0005af68ea54 h1:9NWlQfY2ePejTmfwUH1OWwmznFa+0kKcHGPDvcPza9M=
-google.golang.org/genproto v0.0.0-20230526161137-0005af68ea54/go.mod h1:zqTuNwFlFRsw5zIts5VnzLQxSRqh+CGOTVMlYbY0Eyk=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20230525234030-28d5490b6b19 h1:0nDDozoAU19Qb2HwhXadU8OcsiO/09cnTqhUtq2MEOM=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20230525234030-28d5490b6b19/go.mod h1:66JfowdXAEgad5O9NnYcsNPLCPZJD++2L9X0PCMODrA=
-google.golang.org/grpc v1.56.2 h1:fVRFRnXvU+x6C4IlHZewvJOVHoOv1TUuQyoRsYnB4bI=
-google.golang.org/grpc v1.56.2/go.mod h1:I9bI3vqKfayGqPUAwGdOSu7kt6oIJLixfffKrpXqQ9s=
-google.golang.org/grpc v1.57.0 h1:kfzNeI/klCGD2YPMUlaGNT3pxvYfga7smW3Vth8Zsiw=
-google.golang.org/grpc v1.57.0/go.mod h1:Sd+9RMTACXwmub0zcNY2c4arhtrbBYD1AUHI/dt16Mo=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20230711160842-782d3b101e98 h1:bVf09lpb+OJbByTj913DRJioFFAjf/ZGxEz7MajTp2U=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20230711160842-782d3b101e98/go.mod h1:TUfxEVdsvPg18p6AslUXFoLdpED4oBnGwyqk3dV1XzM=
+google.golang.org/grpc v1.58.1 h1:OL+Vz23DTtrrldqHK49FUOPHyY75rvFqJfXC84NYW58=
+google.golang.org/grpc v1.58.1/go.mod h1:tgX3ZQDlNJGU96V6yHh1T/JeoBQ2TXdr43YbYSsCJk0=
+google.golang.org/grpc v1.58.2 h1:SXUpjxeVF3FKrTYQI4f4KvbGD5u2xccdYdurwowix5I=
+google.golang.org/grpc v1.58.2/go.mod h1:tgX3ZQDlNJGU96V6yHh1T/JeoBQ2TXdr43YbYSsCJk0=
 google.golang.org/protobuf v0.0.0-20200109180630-ec00e32a8dfd/go.mod h1:DFci5gLYBciE7Vtevhsrf46CRTquxDuWsQurQQe4oz8=
 google.golang.org/protobuf v0.0.0-20200221191635-4d8936d0db64/go.mod h1:kwYJMbMJ01Woi6D6+Kah6886xMZcty6N08ah7+eCXa0=
 google.golang.org/protobuf v0.0.0-20200228230310-ab0ca4ff8a60/go.mod h1:cfTl7dwQJ+fmap5saPgwCLgHXTUD7jkjRqWcaiX5VyM=

main.go

@@ -49,6 +49,11 @@ func main() {
 				Name:    "debug",
 				EnvVars: []string{"DEBUG"},
 			},
+			&cli.BoolFlag{
+				Name:    "single-active-backend",
+				EnvVars: []string{"SINGLE_ACTIVE_BACKEND"},
+				Usage:   "Allow only one backend to be running.",
+			},
 			&cli.BoolFlag{
 				Name:    "cors",
 				EnvVars: []string{"CORS"},
@@ -135,6 +140,12 @@ func main() {
 				Usage:   "List of API Keys to enable API authentication. When this is set, all the requests must be authenticated with one of these API keys.",
 				EnvVars: []string{"API_KEY"},
 			},
+			&cli.BoolFlag{
+				Name:    "preload-backend-only",
+				Usage:   "If set, the api is NOT launched, and only the preloaded models / backends are started. This is intended for multi-node setups.",
+				EnvVars: []string{"PRELOAD_BACKEND_ONLY"},
+				Value:   false,
+			},
 		},
 		Description: `
 LocalAI is a drop-in replacement OpenAI API which runs inference locally.
@@ -175,6 +186,10 @@ For a list of compatible model, check out: https://localai.io/model-compatibilit
 				options.WithApiKeys(ctx.StringSlice("api-keys")),
 			}
 
+			if ctx.Bool("single-active-backend") {
+				opts = append(opts, options.EnableSingleBackend)
+			}
+
 			externalgRPC := ctx.StringSlice("external-grpc-backends")
 			// split ":" to get backend name and the uri
 			for _, v := range externalgRPC {
@@ -187,6 +202,11 @@ For a list of compatible model, check out: https://localai.io/model-compatibilit
 				opts = append(opts, options.EnableGalleriesAutoload)
 			}
 
+			if ctx.Bool("preload-backend-only") {
+				_, _, err := api.Startup(opts...)
+				return err
+			}
+
 			app, err := api.App(opts...)
 			if err != nil {
 				return err

pkg/backend/image/stablediffusion.go

@@ -9,7 +9,7 @@ import (
 )
 
 type StableDiffusion struct {
-	base.Base
+	base.SingleThread
 	stablediffusion *stablediffusion.StableDiffusion
 }
 

pkg/backend/llm/bert/bert.go

@@ -10,7 +10,7 @@ import (
 )
 
 type Embeddings struct {
-	base.Base
+	base.SingleThread
 	bert *bert.Bert
 }
 
@@ -21,6 +21,7 @@ func (llm *Embeddings) Load(opts *pb.ModelOptions) error {
 }
 
 func (llm *Embeddings) Embeddings(opts *pb.PredictOptions) ([]float32, error) {
+
 	if len(opts.EmbeddingTokens) > 0 {
 		tokens := []int{}
 		for _, t := range opts.EmbeddingTokens {