Files
LocalAI/pkg/model/loader.go
Ettore Di Giacinto 2d64269763
Some checks failed
Explorer deployment / build-linux (push) Has been cancelled
GPU tests / ubuntu-latest (1.21.x) (push) Has been cancelled
generate and publish intel docker caches / generate_caches (intel/oneapi-basekit:2025.1.0-0-devel-ubuntu22.04, linux/amd64, ubuntu-latest) (push) Has been cancelled
build container images / hipblas-jobs (-aio-gpu-hipblas, rocm/dev-ubuntu-22.04:6.1, hipblas, true, ubuntu:22.04, core, latest-gpu-hipblas, latest-aio-gpu-hipblas, --jobs=3 --output-sync=target, linux/amd64, arc-runner-set, false, -hipblas) (push) Has been cancelled
build container images / core-image-build (-aio-cpu, ubuntu:22.04, , true, core, latest-cpu, latest-aio-cpu, --jobs=4 --output-sync=target, linux/amd64,linux/arm64, arc-runner-set, false, auto, ) (push) Has been cancelled
build container images / core-image-build (-aio-gpu-intel-f16, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, true, ubuntu:22.04, core, latest-gpu-intel-f16, latest-aio-gpu-intel-f16, --jobs=3 --output-sync=target, linux/amd64, arc-runner-set, false, -sycl-f16) (push) Has been cancelled
build container images / core-image-build (-aio-gpu-intel-f32, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, true, ubuntu:22.04, core, latest-gpu-intel-f32, latest-aio-gpu-intel-f32, --jobs=3 --output-sync=target, linux/amd64, arc-runner-set, false, -sycl-f32) (push) Has been cancelled
build container images / core-image-build (-aio-gpu-nvidia-cuda-11, ubuntu:22.04, cublas, 11, 7, true, core, latest-gpu-nvidia-cuda-11, latest-aio-gpu-nvidia-cuda-11, --jobs=4 --output-sync=target, linux/amd64, arc-runner-set, false, false, -cublas-cuda11) (push) Has been cancelled
build container images / core-image-build (-aio-gpu-nvidia-cuda-12, ubuntu:22.04, cublas, 12, 0, true, core, latest-gpu-nvidia-cuda-12, latest-aio-gpu-nvidia-cuda-12, --jobs=4 --output-sync=target, linux/amd64, arc-runner-set, false, false, -cublas-cuda12) (push) Has been cancelled
build container images / core-image-build (-aio-gpu-vulkan, ubuntu:22.04, vulkan, true, core, latest-gpu-vulkan, latest-aio-gpu-vulkan, --jobs=4 --output-sync=target, linux/amd64, arc-runner-set, false, false, -vulkan) (push) Has been cancelled
build container images / gh-runner (nvcr.io/nvidia/l4t-jetpack:r36.4.0, cublas, 12, 0, true, core, latest-nvidia-l4t-arm64, --jobs=4 --output-sync=target, linux/arm64, ubuntu-24.04-arm, true, false, -nvidia-l4t-arm64) (push) Has been cancelled
build python backend container images / backend-jobs (bark, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-bark, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-bark) (push) Has been cancelled
build python backend container images / backend-jobs (bark, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-bark, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-bark) (push) Has been cancelled
build python backend container images / backend-jobs (bark, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-bark, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-bark) (push) Has been cancelled
build python backend container images / backend-jobs (bark, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-bark, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-bark) (push) Has been cancelled
build python backend container images / backend-jobs (bark, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-bark, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-bark) (push) Has been cancelled
build python backend container images / backend-jobs (chatterbox, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-chatterbox, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-chatterbox) (push) Has been cancelled
build python backend container images / backend-jobs (chatterbox, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-chatterbox, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-chatterbox) (push) Has been cancelled
build python backend container images / backend-jobs (coqui, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-coqui, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-coqui) (push) Has been cancelled
build python backend container images / backend-jobs (coqui, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-coqui, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-coqui) (push) Has been cancelled
build python backend container images / backend-jobs (coqui, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-coqui, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-coqui) (push) Has been cancelled
build python backend container images / backend-jobs (coqui, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-coqui, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-coqui) (push) Has been cancelled
build python backend container images / backend-jobs (coqui, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-coqui, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-coqui) (push) Has been cancelled
build python backend container images / backend-jobs (diffusers, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-diffusers, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-diffusers) (push) Has been cancelled
build python backend container images / backend-jobs (diffusers, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-diffusers, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-diffusers) (push) Has been cancelled
build python backend container images / backend-jobs (diffusers, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-diffusers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-diffusers) (push) Has been cancelled
build python backend container images / backend-jobs (diffusers, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-diffusers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-diffusers) (push) Has been cancelled
build python backend container images / backend-jobs (faster-whisper, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-faster-whisper, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-faster-whisper) (push) Has been cancelled
build python backend container images / backend-jobs (faster-whisper, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-faster-whisper, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-faster-whisper) (push) Has been cancelled
build python backend container images / backend-jobs (faster-whisper, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-faster-whisper, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-faster-whisper) (push) Has been cancelled
build python backend container images / backend-jobs (faster-whisper, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-faster-whisper, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-faster-whisper) (push) Has been cancelled
build python backend container images / backend-jobs (faster-whisper, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-faster-whisper, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-faster-whisper) (push) Has been cancelled
build python backend container images / backend-jobs (kokoro, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-kokoro, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-kokoro) (push) Has been cancelled
build python backend container images / backend-jobs (kokoro, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-kokoro, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-kokoro) (push) Has been cancelled
build python backend container images / backend-jobs (kokoro, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-kokoro, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-kokoro) (push) Has been cancelled
build python backend container images / backend-jobs (kokoro, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-kokoro, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-kokoro) (push) Has been cancelled
build python backend container images / backend-jobs (kokoro, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-kokoro, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-kokoro) (push) Has been cancelled
build python backend container images / backend-jobs (rerankers, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-rerankers, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-rerankers) (push) Has been cancelled
build python backend container images / backend-jobs (rerankers, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-rerankers, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-rerankers) (push) Has been cancelled
build python backend container images / backend-jobs (rerankers, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-rerankers, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-rerankers) (push) Has been cancelled
build python backend container images / backend-jobs (rerankers, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-rerankers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-rerankers) (push) Has been cancelled
build python backend container images / backend-jobs (rerankers, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-rerankers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-rerankers) (push) Has been cancelled
build python backend container images / backend-jobs (transformers, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-transformers, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-transformers) (push) Has been cancelled
build python backend container images / backend-jobs (transformers, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-transformers, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-transformers) (push) Has been cancelled
build python backend container images / backend-jobs (transformers, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-transformers, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-transformers) (push) Has been cancelled
build python backend container images / backend-jobs (transformers, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-transformers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-transformers) (push) Has been cancelled
build python backend container images / backend-jobs (transformers, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-transformers, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-transformers) (push) Has been cancelled
build python backend container images / backend-jobs (vllm, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f16, , , latest-gpu-intel-sycl-f16-vllm, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f16-vllm) (push) Has been cancelled
build python backend container images / backend-jobs (vllm, quay.io/go-skynet/intel-oneapi-base:latest, sycl_f32, , , latest-gpu-intel-sycl-f32-vllm, linux/amd64, arc-runner-set, true, -gpu-intel-sycl-f32-vllm) (push) Has been cancelled
build python backend container images / backend-jobs (vllm, rocm/dev-ubuntu-22.04:6.1, hipblas, , , latest-gpu-rocm-hipblas-vllm, linux/amd64, arc-runner-set, true, -gpu-rocm-hipblas-vllm) (push) Has been cancelled
build python backend container images / backend-jobs (vllm, ubuntu:22.04, cublas, 11, 7, latest-gpu-nvidia-cuda-11-vllm, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-11-vllm) (push) Has been cancelled
build python backend container images / backend-jobs (vllm, ubuntu:22.04, cublas, 12, 0, latest-gpu-nvidia-cuda-12-vllm, linux/amd64, arc-runner-set, true, -gpu-nvidia-cuda-12-vllm) (push) Has been cancelled
Security Scan / tests (push) Has been cancelled
Tests extras backends / tests-transformers (push) Has been cancelled
Tests extras backends / tests-rerankers (push) Has been cancelled
Tests extras backends / tests-diffusers (push) Has been cancelled
Tests extras backends / tests-coqui (push) Has been cancelled
tests / tests-linux (1.21.x) (push) Has been cancelled
tests / tests-aio-container (push) Has been cancelled
tests / tests-apple (1.21.x) (push) Has been cancelled
Update swagger / swagger (push) Has been cancelled
Check if checksums are up-to-date / checksum_check (push) Has been cancelled
Bump dependencies / bump (mudler/LocalAI) (push) Has been cancelled
Bump dependencies / bump (main, PABannier/bark.cpp, BARKCPP_VERSION) (push) Has been cancelled
Bump dependencies / bump (master, ggml-org/llama.cpp, CPPLLAMA_VERSION) (push) Has been cancelled
Bump dependencies / bump (master, ggml-org/whisper.cpp, WHISPER_CPP_VERSION) (push) Has been cancelled
Bump dependencies / bump (master, leejet/stable-diffusion.cpp, STABLEDIFFUSION_GGML_VERSION) (push) Has been cancelled
Bump dependencies / bump (master, mudler/go-piper, PIPER_VERSION) (push) Has been cancelled
Bump dependencies / bump (master, mudler/go-stable-diffusion, STABLEDIFFUSION_VERSION) (push) Has been cancelled
feat: Add backend gallery (#5607)
* feat: Add backend gallery

This PR add support to manage backends as similar to models. There is
now available a backend gallery which can be used to install and remove
extra backends.
The backend gallery can be configured similarly as a model gallery, and
API calls allows to install and remove new backends in runtime, and as
well during the startup phase of LocalAI.

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

* Add backends docs

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

* wip: Backend Dockerfile for python backends

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

* feat: drop extras images, build python backends separately

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

* fixup on all backends

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

* test CI

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

* Tweaks

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

* Drop old backends leftovers

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

* Fixup CI

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

* Move dockerfile upper

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

* Fix proto

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

* Feature dropped for consistency - we prefer model galleries

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

* Add missing packages in the build image

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

* exllama is ponly available on cublas

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

* pin torch on chatterbox

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

* Fixups to index

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

* CI

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

* Debug CI

* Install accellerators deps

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

* Add target arch

* Add cuda minor version

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

* Use self-hosted runners

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

* ci: use quay for test images

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

* fixups for vllm and chatterbox

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

* Small fixups on CI

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

* chatterbox is only available for nvidia

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

* Simplify CI builds

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

* Adapt test, use qwen3

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

* chore(model gallery): add jina-reranker-v1-tiny-en-gguf

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

* fix(gguf-parser): recover from potential panics that can happen while reading ggufs with gguf-parser

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

* Use reranker from llama.cpp in AIO images

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

* Limit concurrent jobs

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

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
2025-06-15 14:56:52 +02:00

219 lines
4.8 KiB
Go

package model
import (
"context"
"fmt"
"maps"
"os"
"path/filepath"
"strings"
"sync"
"time"
"github.com/mudler/LocalAI/pkg/utils"
"github.com/rs/zerolog/log"
)
// new idea: what if we declare a struct of these here, and use a loop to check?
// TODO: Split ModelLoader and TemplateLoader? Just to keep things more organized. Left together to share a mutex until I look into that. Would split if we separate directories for .bin/.yaml and .tmpl
type ModelLoader struct {
ModelPath string
mu sync.Mutex
singletonLock sync.Mutex
singletonMode bool
models map[string]*Model
wd *WatchDog
externalBackends map[string]string
}
func NewModelLoader(modelPath string, singleActiveBackend bool) *ModelLoader {
nml := &ModelLoader{
ModelPath: modelPath,
models: make(map[string]*Model),
singletonMode: singleActiveBackend,
externalBackends: make(map[string]string),
}
return nml
}
func (ml *ModelLoader) SetWatchDog(wd *WatchDog) {
ml.wd = wd
}
func (ml *ModelLoader) ExistsInModelPath(s string) bool {
return utils.ExistsInPath(ml.ModelPath, s)
}
func (ml *ModelLoader) SetExternalBackend(name, uri string) {
ml.mu.Lock()
defer ml.mu.Unlock()
ml.externalBackends[name] = uri
}
func (ml *ModelLoader) DeleteExternalBackend(name string) {
ml.mu.Lock()
defer ml.mu.Unlock()
delete(ml.externalBackends, name)
}
func (ml *ModelLoader) GetExternalBackend(name string) string {
ml.mu.Lock()
defer ml.mu.Unlock()
return ml.externalBackends[name]
}
func (ml *ModelLoader) GetAllExternalBackends(o *Options) map[string]string {
backends := make(map[string]string)
maps.Copy(backends, ml.externalBackends)
if o != nil {
maps.Copy(backends, o.externalBackends)
}
return backends
}
var knownFilesToSkip []string = []string{
"MODEL_CARD",
"README",
"README.md",
}
var knownModelsNameSuffixToSkip []string = []string{
".tmpl",
".keep",
".yaml",
".yml",
".json",
".txt",
".pt",
".onnx",
".md",
".MD",
".DS_Store",
".",
".safetensors",
".partial",
".tar.gz",
}
const retryTimeout = time.Duration(2 * time.Minute)
func (ml *ModelLoader) ListFilesInModelPath() ([]string, error) {
files, err := os.ReadDir(ml.ModelPath)
if err != nil {
return []string{}, err
}
models := []string{}
FILE:
for _, file := range files {
for _, skip := range knownFilesToSkip {
if strings.EqualFold(file.Name(), skip) {
continue FILE
}
}
// Skip templates, YAML, .keep, .json, and .DS_Store files
for _, skip := range knownModelsNameSuffixToSkip {
if strings.HasSuffix(file.Name(), skip) {
continue FILE
}
}
// Skip directories
if file.IsDir() {
continue
}
models = append(models, file.Name())
}
return models, nil
}
func (ml *ModelLoader) ListModels() []*Model {
ml.mu.Lock()
defer ml.mu.Unlock()
models := []*Model{}
for _, model := range ml.models {
models = append(models, model)
}
return models
}
func (ml *ModelLoader) LoadModel(modelID, modelName string, loader func(string, string, string) (*Model, error)) (*Model, error) {
// Check if we already have a loaded model
if model := ml.CheckIsLoaded(modelID); model != nil {
return model, nil
}
// Load the model and keep it in memory for later use
modelFile := filepath.Join(ml.ModelPath, modelName)
log.Debug().Msgf("Loading model in memory from file: %s", modelFile)
ml.mu.Lock()
defer ml.mu.Unlock()
model, err := loader(modelID, modelName, modelFile)
if err != nil {
return nil, fmt.Errorf("failed to load model with internal loader: %s", err)
}
if model == nil {
return nil, fmt.Errorf("loader didn't return a model")
}
ml.models[modelID] = model
return model, nil
}
func (ml *ModelLoader) ShutdownModel(modelName string) error {
ml.mu.Lock()
defer ml.mu.Unlock()
return ml.deleteProcess(modelName)
}
func (ml *ModelLoader) CheckIsLoaded(s string) *Model {
ml.mu.Lock()
defer ml.mu.Unlock()
m, ok := ml.models[s]
if !ok {
return nil
}
log.Debug().Msgf("Model already loaded in memory: %s", s)
client := m.GRPC(false, ml.wd)
log.Debug().Msgf("Checking model availability (%s)", s)
cTimeout, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
defer cancel()
alive, err := client.HealthCheck(cTimeout)
if !alive {
log.Warn().Msgf("GRPC Model not responding: %s", err.Error())
log.Warn().Msgf("Deleting the process in order to recreate it")
process := m.Process()
if process == nil {
log.Error().Msgf("Process not found for '%s' and the model is not responding anymore !", s)
return m
}
if !process.IsAlive() {
log.Debug().Msgf("GRPC Process is not responding: %s", s)
// stop and delete the process, this forces to re-load the model and re-create again the service
err := ml.deleteProcess(s)
if err != nil {
log.Error().Err(err).Str("process", s).Msg("error stopping process")
}
return nil
}
}
return m
}