* fix(backend): resolve relative draft_model paths against the models dir
The main model file and mmproj are joined with the configured models
directory before reaching the backend, but draft_model was sent
verbatim. With a relative draft_model in the YAML config, llama.cpp
opens the path from the backend process's CWD and fails with "No such
file or directory", forcing users to hard-code an absolute path.
Mirror the existing mmproj resolution: if draft_model is relative,
join it with modelPath. Absolute paths are passed through unchanged.
Adds an e2e regression test against the mock backend that asserts the
main model file, mmproj, and draft_model all arrive at the backend
resolved to absolute paths.
Closes#9675
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Assisted-by: Claude:claude-opus-4-7-1m [Read] [Edit] [Bash] [Write]
* fix(backend): always join draft_model with models dir (drop IsAbs shortcut)
The previous commit kept absolute draft_model paths intact via an
IsAbs check. That left a path-traversal vector open: a user-supplied
YAML config could set draft_model to /etc/passwd (or any other host
file the backend process can read) and the path would be sent through
unchanged.
filepath.Join cleans the leading slash from absolute components, so
joining unconditionally — the way mmproj already does — keeps the
result rooted at the configured models directory regardless of input.
Adds a second e2e spec that feeds an absolute draft_model into the
mock backend and asserts the path is clamped under modelsPath.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Assisted-by: Claude:claude-opus-4-7-1m [Read] [Edit] [Bash]
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
* feat(vllm): expose AsyncEngineArgs via generic engine_args YAML map
LocalAI's vLLM backend wraps a small typed subset of vLLM's
AsyncEngineArgs (quantization, tensor_parallel_size, dtype, etc.).
Anything outside that subset -- pipeline/data/expert parallelism,
speculative_config, kv_transfer_config, all2all_backend, prefix
caching, chunked prefill, etc. -- requires a new protobuf field, a
Go struct field, an options.go line, and a backend.py mapping per
feature. That cadence is the bottleneck on shipping vLLM's
production feature set.
Add a generic `engine_args:` map on the model YAML that is
JSON-serialised into a new ModelOptions.EngineArgs proto field and
applied verbatim to AsyncEngineArgs at LoadModel time. Validation
is done by the Python backend via dataclasses.fields(); unknown
keys fail with the closest valid name as a hint.
dataclasses.replace() is used so vLLM's __post_init__ re-runs and
auto-converts dict values into nested config dataclasses
(CompilationConfig, AttentionConfig, ...). speculative_config and
kv_transfer_config flow through as dicts; vLLM converts them at
engine init.
Operators can now write:
engine_args:
data_parallel_size: 8
enable_expert_parallel: true
all2all_backend: deepep_low_latency
speculative_config:
method: deepseek_mtp
num_speculative_tokens: 3
kv_cache_dtype: fp8
without further proto/Go/Python plumbing per field.
Production defaults seeded by hooks_vllm.go: enable_prefix_caching
and enable_chunked_prefill default to true unless explicitly set.
Existing typed YAML fields (gpu_memory_utilization,
tensor_parallel_size, etc.) remain for back-compat; engine_args
overrides them when both are set.
Assisted-by: Claude:claude-opus-4-7 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* chore(vllm): pin cublas13 to vLLM 0.20.0 cu130 wheel
vLLM's PyPI wheel is built against CUDA 12 (libcudart.so.12) and won't
load on a cu130 host. Switch the cublas13 build to vLLM's per-tag cu130
simple-index (https://wheels.vllm.ai/0.20.0/cu130/) and pin
vllm==0.20.0. The cu130-flavoured wheel ships libcudart.so.13 and
includes the DFlash speculative-decoding method that landed in 0.20.0.
cublas13 install gets --index-strategy=unsafe-best-match so uv consults
both the cu130 index and PyPI when resolving — PyPI also publishes
vllm==0.20.0, but with cu12 binaries that error at import time.
Verified: Qwen3.5-4B + z-lab/Qwen3.5-4B-DFlash loads and serves chat
completions on RTX 5070 Ti (sm_120, cu130).
Assisted-by: Claude:claude-opus-4-7 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* ci(vllm): bot job to bump cublas13 vLLM wheel pin
vLLM's cu130 wheel index URL is itself version-locked
(wheels.vllm.ai/<TAG>/cu130/, no /latest/ alias upstream), so a vLLM
bump means rewriting two values atomically — the URL segment and the
version constraint. bump_deps.sh handles git-sha-in-Makefile only;
add a sibling bump_vllm_wheel.sh and a matching workflow job that
mirrors the existing matrix's PR-creation pattern.
The bumper queries /releases/latest (which excludes prereleases),
strips the leading 'v', and seds both lines unconditionally. When the
file is already on the latest tag the rewrite is a no-op and
peter-evans/create-pull-request opens no PR.
Assisted-by: Claude:claude-opus-4-7 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* docs(vllm): document engine_args and speculative decoding
The new engine_args: map plumbs arbitrary AsyncEngineArgs through to
vLLM, but the public docs only covered the basic typed fields. Add a
short subsection in the vLLM section explaining the typed/generic
split and showing a worked DFlash speculative-decoding config, with
pointers to vLLM's SpeculativeConfig reference and z-lab's drafter
collection.
Assisted-by: Claude:claude-opus-4-7 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
---------
Signed-off-by: Richard Palethorpe <io@richiejp.com>
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
* always enable parallel requests
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: add node reconciler, allow to schedule to group of nodes, min/max autoscaler
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: move tests to ginkgo
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(smart router): order by available vram
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: add distributed mode (experimental)
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix data races, mutexes, transactions
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactorings
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fixups
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix events and tool stream in agent chat
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* use ginkgo
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(cron): compute correctly time boundaries avoiding re-triggering
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* enhancements, refactorings
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* do not flood of healthy checks
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* do not list obvious backends as text backends
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* tests fixups
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* refactoring and consolidation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Drop redundant healthcheck
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* enhancements, refactorings
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: wire min_p
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: inferencing defaults
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(refactor): re-use iterative parser
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: generate automatically inference defaults from unsloth
Instead of trying to re-invent the wheel and maintain here the inference
defaults, prefer to consume unsloth ones, and contribute there as
necessary.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: apply defaults also to models installed via gallery
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: be consistent and apply fallback to all endpoint
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat(gallery): Switch to expandable box instead of pop-over and display model files
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* feat(ui, backends): Add individual backend logging
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* fix(ui): Set the context settings from the model config
Signed-off-by: Richard Palethorpe <io@richiejp.com>
---------
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* fix: Automatically disable mmap for Intel SYCL backends
Fixes issue #9012 where Qwen3.5 models fail to load on Intel Arc GPU
with RPC EOF error.
The Intel SYCL backend has a known issue where mmap enabled causes
the backend to hang. This change automatically disables mmap when
detecting Intel or SYCL backends.
References:
- https://github.com/mudler/LocalAI/issues/9012
- Documentation mentions: SYCL hangs when mmap: true is set
* feat: Add logging for mmap auto-disable on Intel SYCL backends
As requested in PR review, add xlog.Info call to log when mmap
is automatically disabled for Intel SYCL backends. This helps
with debugging and confirms the auto-disable logic is working.
---------
Co-authored-by: localai-bot <localai-bot@users.noreply.github.com>
* feat: add support to logprobs in results
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: add support to logitbias
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(llama.cpp): correctly set grammar triggers
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Do not enable lazy by default
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
- Add a system backend path
- Refactor and consolidate system information in system state
- Use system state in all the components to figure out the system paths
to used whenever needed
- Refactor BackendConfig -> ModelConfig. This was otherway misleading as
now we do have a backend configuration which is not the model config.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: split remaining backends and drop embedded backends
- Drop silero-vad, huggingface, and stores backend from embedded
binaries
- Refactor Makefile and Dockerfile to avoid building grpc backends
- Drop golang code that was used to embed backends
- Simplify building by using goreleaser
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(gallery): be specific with llama-cpp backend templates
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(docs): update
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(ci): minor fixes
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: drop all ffmpeg references
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix: run protogen-go
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Always enable p2p mode
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Update gorelease file
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(stores): do not always load
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Fix linting issues
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Simplify
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Mac OS fixup
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(refactor): track internally started models by ID
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Just extend options, no need to copy
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Improve debugging for rerankers failures
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Simplify model loading with rerankers
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Be more consistent when generating model options
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Uncommitted code
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Make deleteProcess more idiomatic
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Adapt CLI for sound generation
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Fixup threads definition
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Handle corner case where c.Seed is nil
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Consistently use ModelOptions
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Adapt new code to refactoring
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Dave <dave@gray101.com>
* feat(llama.cpp): add embeddings
Also enable embeddings by default for llama.cpp models
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(Makefile): prepare llama.cpp sources only once
Otherwise we keep cloning llama.cpp for each of the variants
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* do not set embeddings to false
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* docs: add embeddings to the YAML config reference
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(seed): generate random seed per-request if -1 is set
Also update ci with new workflows and allow the aio tests to run with an
api key
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* docs(openvino): Add OpenVINO example
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(defaults): set better defaults for inferencing
This changeset aim to have better defaults and to properly detect when
no inference settings are provided with the model.
If not specified, we defaults to mirostat sampling, and offload all the
GPU layers (if a GPU is detected).
Related to https://github.com/mudler/LocalAI/issues/1373 and https://github.com/mudler/LocalAI/issues/1723
* Adapt tests
* Also pre-initialize default seed
* feat(intel): add diffusers support
* try to consume upstream container image
* Debug
* Manually install deps
* Map transformers/hf cache dir to modelpath if not specified
* fix(compel): update initialization, pass by all gRPC options
* fix: add dependencies, implement transformers for xpu
* base it from the oneapi image
* Add pillow
* set threads if specified when launching the API
* Skip conda install if intel
* defaults to non-intel
* ci: add to pipelines
* prepare compel only if enabled
* Skip conda install if intel
* fix cleanup
* Disable compel by default
* Install torch 2.1.0 with Intel
* Skip conda on some setups
* Detect python
* Quiet output
* Do not override system python with conda
* Prefer python3
* Fixups
* exllama2: do not install without conda (overrides pytorch version)
* exllama/exllama2: do not install if not using cuda
* Add missing dataset dependency
* Small fixups, symlink to python, add requirements
* Add neural_speed to the deps
* correctly handle model offloading
* fix: device_map == xpu
* go back at calling python, fixed at dockerfile level
* Exllama2 restricted to only nvidia gpus
* Tokenizer to xpu
* core 1
* api/openai/files fix
* core 2 - core/config
* move over core api.go and tests to the start of core/http
* move over localai specific endpoints to core/http, begin the service/endpoint split there
* refactor big chunk on the plane
* refactor chunk 2 on plane, next step: port and modify changes to request.go
* easy fixes for request.go, major changes not done yet
* lintfix
* json tag lintfix?
* gitignore and .keep files
* strange fix attempt: rename the config dir?
This PR specifically introduces a `core` folder and moves the following packages over, without any other changes:
- `api/backend`
- `api/config`
- `api/options`
- `api/schema`
Once this is merged and we confirm there's no regressions, I can migrate over the remaining changes piece by piece to split up application startup, backend services, http, and mqtt as was the goal of the earlier PRs!
