mirror of
https://github.com/mudler/LocalAI.git
synced 2026-07-30 09:57:57 -04:00
7d8e0bac18
* fix(model): deterministic, file-type-filtered backend auto-detect (#9287) When a model config declares no explicit `backend:`, Load() fell into a trial loop built by ranging the external-backends Go map (random order) with no filtering, returning the first backend whose gRPC LoadModel succeeded. An unrelated installed backend - e.g. the "opus" audio codec - could therefore win a GGUF/LLM model load, so a model that should run on llama.cpp wrongly tried to use opus. Extract the candidate selection into a pure, testable function SelectAutoLoadBackends that: - sorts the candidate list deterministically (no more map-order nondeterminism), and - for a `.gguf` model, filters to LLM-capable backends (via core/config.BackendCapabilities) and puts llama-cpp first, so an incompatible audio/codec/image backend can never win the trial loop. If filtering would leave zero candidates, the full sorted set is returned unchanged, so a previously-loadable model is never made unloadable. Signed-off-by: Ettore Di Giacinto <mudler@localai.io> Assisted-by: claude:claude-opus-4-8 [Claude Code] Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * fix(model): break core/config <-> pkg/model import cycle in backend auto-detect The #9287 auto-detect change made pkg/model/autoload.go import core/config for the backend capability table. core/config already imports pkg/model (runtime_settings_registry.go uses model.DefaultWatchdogInterval), so this closed a core/config -> pkg/model -> core/config import cycle and broke the build and golangci-lint. Invert the dependency so the lower-level pkg/model no longer imports the higher-level core/config. pkg/model exposes RegisterLLMCapableBackendFunc and uses the registered predicate; core/config (which owns the capability table) registers it from an init(). The deterministic, GGUF-type-filtered selection behaviour is unchanged. When the predicate is unwired the GGUF filter is skipped, preserving the existing zero-candidate fallback. The unit test now injects a fake capability predicate so SelectAutoLoadBackends is exercised independently of the core/config table. Signed-off-by: Ettore Di Giacinto <mudler@localai.io> Assisted-by: Claude:opus-4.8 [Claude Code] --------- Signed-off-by: Ettore Di Giacinto <mudler@localai.io> Co-authored-by: Ettore Di Giacinto <mudler@localai.io> Co-authored-by: localai-org-maint-bot <bot-opensource@localaisrl.com>
56 lines
2.1 KiB
Go
56 lines
2.1 KiB
Go
package model_test
|
|
|
|
import (
|
|
"github.com/mudler/LocalAI/pkg/model"
|
|
|
|
. "github.com/onsi/ginkgo/v2"
|
|
. "github.com/onsi/gomega"
|
|
)
|
|
|
|
func init() {
|
|
// The real LLM-capability table lives in core/config, which pkg/model must
|
|
// not import (cycle). Register a fake predicate so the selection algorithm
|
|
// is exercised here independent of the capability table (#9287).
|
|
model.RegisterLLMCapableBackendFunc(func(name string) bool {
|
|
return name == "llama-cpp" || name == "vllm"
|
|
})
|
|
}
|
|
|
|
var _ = Describe("SelectAutoLoadBackends (#9287)", func() {
|
|
Describe("GGUF model auto-detection", func() {
|
|
It("excludes incompatible audio/codec backends (e.g. opus) for a .gguf model", func() {
|
|
// Regression for #9287: installing an unrelated audio backend like
|
|
// "opus" must never win the GGUF auto-detect trial loop.
|
|
got := model.SelectAutoLoadBackends([]string{"opus", "llama-cpp"}, "Qwen3.5-9b.gguf")
|
|
Expect(got).NotTo(ContainElement("opus"))
|
|
Expect(got).To(ContainElement("llama-cpp"))
|
|
})
|
|
|
|
It("places llama-cpp first for a .gguf model", func() {
|
|
got := model.SelectAutoLoadBackends([]string{"vllm", "opus", "llama-cpp"}, "model.gguf")
|
|
Expect(got).NotTo(BeEmpty())
|
|
Expect(got[0]).To(Equal("llama-cpp"))
|
|
})
|
|
|
|
It("is deterministic regardless of input ordering", func() {
|
|
a := model.SelectAutoLoadBackends([]string{"opus", "vllm", "llama-cpp", "whisper"}, "m.gguf")
|
|
b := model.SelectAutoLoadBackends([]string{"whisper", "llama-cpp", "vllm", "opus"}, "m.gguf")
|
|
Expect(a).To(Equal(b))
|
|
})
|
|
|
|
It("falls back to the full sorted set when filtering leaves no candidate", func() {
|
|
// No LLM-capable backend installed: never make a previously-loadable
|
|
// model unloadable, return the original set (sorted).
|
|
got := model.SelectAutoLoadBackends([]string{"opus"}, "model.gguf")
|
|
Expect(got).To(Equal([]string{"opus"}))
|
|
})
|
|
})
|
|
|
|
Describe("non-GGUF model auto-detection", func() {
|
|
It("returns a deterministic (sorted) set without filtering", func() {
|
|
got := model.SelectAutoLoadBackends([]string{"opus", "llama-cpp", "diffusers"}, "model-dir")
|
|
Expect(got).To(Equal([]string{"diffusers", "llama-cpp", "opus"}))
|
|
})
|
|
})
|
|
})
|