chore(deps): bump llama.cpp to '10f2e81809bbb69ecfe64fc8b4686285f84b0c07'

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

Makefile

@@ -6,7 +6,7 @@ BINARY_NAME=local-ai
 DETECT_LIBS?=true
 
 # llama.cpp versions
-CPPLLAMA_VERSION?=1e2f78a00450593e2dfa458796fcdd9987300dfc
+CPPLLAMA_VERSION?=10f2e81809bbb69ecfe64fc8b4686285f84b0c07
 
 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggerganov/whisper.cpp

backend/cpp/llama/patches/01-llava.patch

@@ -1,13 +1,13 @@
 diff --git a/examples/llava/clip.cpp b/examples/llava/clip.cpp
-index 3cd0d2fa..6c5e811a 100644
+index 7f892beb..0517e529 100644
 --- a/examples/llava/clip.cpp
 +++ b/examples/llava/clip.cpp
-@@ -2608,7 +2608,7 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
-                 struct ggml_tensor * patches = ggml_graph_get_tensor(gf, "patches");
+@@ -2766,7 +2766,7 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
+                 int patch_offset = ctx->has_class_embedding ? 1 : 0;
                  int* patches_data = (int*)malloc(ggml_nbytes(patches));
                  for (int i = 0; i < num_patches; i++) {
--                    patches_data[i] = i + 1;
-+                    patches_data[i] = i;
+-                    patches_data[i] = i + patch_offset;
++                    patches_data[i] = i + 1;
                  }
                  ggml_backend_tensor_set(patches, patches_data, 0, ggml_nbytes(patches));
                  free(patches_data);

backend/cpp/llama/prepare.sh

@@ -1,5 +1,7 @@
 #!/bin/bash
 
+set -e
+
 ## Patches
 ## Apply patches from the `patches` directory
 for patch in $(ls patches); do

backend/python/autogptq/requirements.txt

@@ -1,6 +1,6 @@
 accelerate
 auto-gptq==0.7.1
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi
 transformers

backend/python/bark/requirements.txt

@@ -1,4 +1,4 @@
 bark==0.1.5
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi

backend/python/common/template/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 grpcio-tools

backend/python/coqui/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi
 packaging==24.1

backend/python/diffusers/requirements.txt

@@ -1,5 +1,5 @@
 setuptools
-grpcio==1.70.0
+grpcio==1.71.0
 pillow
 protobuf
 certifi

backend/python/exllama2/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi
 wheel

backend/python/faster-whisper/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 grpcio-tools

backend/python/kokoro/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 phonemizer
 scipy

backend/python/rerankers/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi

backend/python/transformers/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi
 setuptools

backend/python/vllm/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.70.0
+grpcio==1.71.0
 protobuf
 certifi
 setuptools

docs/themes/hugo-theme-relearn

@@ -0,0 +1 @@
+9a020e7eadb7d8203f5b01b18756c72d94773ec9

gallery/index.yaml

@@ -4499,6 +4499,77 @@
     - filename: Tower-Babel_Babel-9B-Chat-Q4_K_M.gguf
       sha256: cf024c81b9c5e31dd9b4fe89f7bed01be8a6a704722780fe8d240b1ecb7942eb
       uri: huggingface://bartowski/Tower-Babel_Babel-9B-Chat-GGUF/Tower-Babel_Babel-9B-Chat-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "openpipe_deductive-reasoning-qwen-14b"
+  urls:
+    - https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-14B
+    - https://huggingface.co/bartowski/OpenPipe_Deductive-Reasoning-Qwen-14B-GGUF
+  description: |
+    Deductive Reasoning Qwen 14B is a reinforcement fine-tune of Qwen 2.5 14B Instruct to solve challenging deduction problems from the Temporal Clue dataset, trained by OpenPipe!
+  overrides:
+    parameters:
+      model: OpenPipe_Deductive-Reasoning-Qwen-14B-Q4_K_M.gguf
+  files:
+    - filename: OpenPipe_Deductive-Reasoning-Qwen-14B-Q4_K_M.gguf
+      sha256: 23474b114e1e14f5f63829369e9af14d3f8e6b437b7974e1d3ac0c842b4cc3f5
+      uri: huggingface://bartowski/OpenPipe_Deductive-Reasoning-Qwen-14B-GGUF/OpenPipe_Deductive-Reasoning-Qwen-14B-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "openpipe_deductive-reasoning-qwen-32b"
+  urls:
+    - https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-32B
+    - https://huggingface.co/bartowski/OpenPipe_Deductive-Reasoning-Qwen-32B-GGUF
+  description: |
+    Deductive Reasoning Qwen 32B is a reinforcement fine-tune of Qwen 2.5 32B Instruct to solve challenging deduction problems from the Temporal Clue dataset, trained by OpenPipe!
+  overrides:
+    parameters:
+      model: OpenPipe_Deductive-Reasoning-Qwen-32B-Q4_K_M.gguf
+  files:
+    - filename: OpenPipe_Deductive-Reasoning-Qwen-32B-Q4_K_M.gguf
+      sha256: 53a8314e572c60c867da897721d366f183dc6d2193c83a41ff8ad46a2a0692c8
+      uri: huggingface://bartowski/OpenPipe_Deductive-Reasoning-Qwen-32B-GGUF/OpenPipe_Deductive-Reasoning-Qwen-32B-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "open-r1_olympiccoder-32b"
+  urls:
+    - https://huggingface.co/open-r1/OlympicCoder-32B
+    - https://huggingface.co/bartowski/open-r1_OlympicCoder-32B-GGUF
+  description: |
+    OlympicCoder-32B is a code mode that achieves very strong performance on competitive coding benchmarks such as LiveCodeBench andthe 2024 International Olympiad in Informatics.
+  overrides:
+    parameters:
+      model: open-r1_OlympicCoder-32B-Q4_K_M.gguf
+  files:
+    - filename: open-r1_OlympicCoder-32B-Q4_K_M.gguf
+      sha256: bb82e4aa2219f655d37c7efad8985582cf3c32de0e0299ecd2f304d32ac39f12
+      uri: huggingface://bartowski/open-r1_OlympicCoder-32B-GGUF/open-r1_OlympicCoder-32B-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "open-r1_olympiccoder-7b"
+  urls:
+    - https://huggingface.co/open-r1/OlympicCoder-7B
+    - https://huggingface.co/bartowski/open-r1_OlympicCoder-7B-GGUF
+  description: |
+    OlympicCoder-7B is a code model that achieves strong performance on competitive coding benchmarks such as LiveCodeBench and the 2024 International Olympiad in Informatics.
+  overrides:
+    parameters:
+      model: open-r1_OlympicCoder-7B-Q4_K_M.gguf
+  files:
+    - filename: open-r1_OlympicCoder-7B-Q4_K_M.gguf
+      sha256: 21e18e7fd1fb244455a67d4dee538a4d86dc96d507c39a4ad16ef335fb9e6e2f
+      uri: huggingface://bartowski/open-r1_OlympicCoder-7B-GGUF/open-r1_OlympicCoder-7B-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "trashpanda-org_qwq-32b-snowdrop-v0"
+  icon: https://cdn-uploads.huggingface.co/production/uploads/675a77cf99ca23af9daacccc/Tdn0PJBFnG3J6UcjO9G94.png
+  urls:
+    - https://huggingface.co/trashpanda-org/QwQ-32B-Snowdrop-v0
+    - https://huggingface.co/bartowski/trashpanda-org_QwQ-32B-Snowdrop-v0-GGUF
+  description: |
+    R1 at home for RP, literally. Able to handle my cards with gimmicks and subtle tricks in them. With a good reasoning starter+prompt, I'm getting consistently-structured responses that have a good amount of variation across them still while rerolling. Char/scenario portrayal is good despite my focus on writing style, lorebooks are properly referenced at times. Slop doesn't seem to be too much of an issue with thinking enabled. Some user impersonation is rarely observed. Prose is refreshing if you take advantage of what I did (writing style fixation). I know I said Marigold would be my daily driver, but this one is that now, it's that good.
+  overrides:
+    parameters:
+      model: trashpanda-org_QwQ-32B-Snowdrop-v0-Q4_K_M.gguf
+  files:
+    - filename: trashpanda-org_QwQ-32B-Snowdrop-v0-Q4_K_M.gguf
+      sha256: 584d2f14f2f08ce499665c332bef30245b605ed2278e9075766237835f564c5f
+      uri: huggingface://bartowski/trashpanda-org_QwQ-32B-Snowdrop-v0-GGUF/trashpanda-org_QwQ-32B-Snowdrop-v0-Q4_K_M.gguf
 - &llama31
   url: "github:mudler/LocalAI/gallery/llama3.1-instruct.yaml@master" ## LLama3.1
   icon: https://avatars.githubusercontent.com/u/153379578
@@ -9028,6 +9099,21 @@
     - filename: TheDrummer_Gemmasutra-Pro-27B-v1.1-Q4_K_M.gguf
       sha256: 218a14f0bf8266f9e77d16b8b4f5cc1dc76e97eb582a2c97cca5a3a2c35de86b
       uri: huggingface://bartowski/TheDrummer_Gemmasutra-Pro-27B-v1.1-GGUF/TheDrummer_Gemmasutra-Pro-27B-v1.1-Q4_K_M.gguf
+- !!merge <<: *gemma
+  name: "thedrummer_gemmasutra-small-4b-v1"
+  icon: https://cdn-uploads.huggingface.co/production/uploads/65f2fd1c25b848bd061b5c2e/Cekk7d2UAKu7LPsw8SxV7.png
+  urls:
+    - https://huggingface.co/TheDrummer/Gemmasutra-Small-4B-v1
+    - https://huggingface.co/bartowski/TheDrummer_Gemmasutra-Small-4B-v1-GGUF
+  description: |
+    An upscaled Gemma 2B tune with modern techniques. Au Revoir, Gemma!
+  overrides:
+    parameters:
+      model: TheDrummer_Gemmasutra-Small-4B-v1-Q4_K_M.gguf
+  files:
+    - filename: TheDrummer_Gemmasutra-Small-4B-v1-Q4_K_M.gguf
+      sha256: 81dd2e2d9546f5dc2150c45c62acabc112068b801ca50b79feceabb1bd4d6f1a
+      uri: huggingface://bartowski/TheDrummer_Gemmasutra-Small-4B-v1-GGUF/TheDrummer_Gemmasutra-Small-4B-v1-Q4_K_M.gguf
 - &llama3
   url: "github:mudler/LocalAI/gallery/llama3-instruct.yaml@master"
   icon: https://avatars.githubusercontent.com/u/153379578