Files
LocalAI/core/config/model_config_test.go
Richard Palethorpe 4916f8c880 feat(vllm): expose AsyncEngineArgs via generic engine_args YAML map (#9563)
* 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>
2026-04-29 00:49:28 +02:00

268 lines
7.4 KiB
Go

package config
import (
"io"
"net/http"
"os"
. "github.com/onsi/ginkgo/v2"
. "github.com/onsi/gomega"
)
var _ = Describe("Test cases for config related functions", func() {
Context("Test Read configuration functions", func() {
It("Test Validate", func() {
tmp, err := os.CreateTemp("", "config.yaml")
Expect(err).To(BeNil())
defer os.Remove(tmp.Name())
_, err = tmp.WriteString(
`backend: "../foo-bar"
name: "foo"
parameters:
model: "foo-bar"
known_usecases:
- chat
- COMPLETION
`)
Expect(err).ToNot(HaveOccurred())
configs, err := readModelConfigsFromFile(tmp.Name())
config := configs[0]
Expect(err).To(BeNil())
Expect(config).ToNot(BeNil())
valid, err := config.Validate()
Expect(err).To(HaveOccurred())
Expect(valid).To(BeFalse())
Expect(config.KnownUsecases).ToNot(BeNil())
})
It("Test Validate", func() {
tmp, err := os.CreateTemp("", "config.yaml")
Expect(err).To(BeNil())
defer os.Remove(tmp.Name())
_, err = tmp.WriteString(
`name: bar-baz
backend: "foo-bar"
parameters:
model: "foo-bar"`)
Expect(err).ToNot(HaveOccurred())
configs, err := readModelConfigsFromFile(tmp.Name())
config := configs[0]
Expect(err).To(BeNil())
Expect(config).ToNot(BeNil())
// two configs in config.yaml
Expect(config.Name).To(Equal("bar-baz"))
valid, err := config.Validate()
Expect(err).To(BeNil())
Expect(valid).To(BeTrue())
// download https://raw.githubusercontent.com/mudler/LocalAI/v2.25.0/embedded/models/hermes-2-pro-mistral.yaml
httpClient := http.Client{}
resp, err := httpClient.Get("https://raw.githubusercontent.com/mudler/LocalAI/v2.25.0/embedded/models/hermes-2-pro-mistral.yaml")
Expect(err).To(BeNil())
defer resp.Body.Close()
tmp, err = os.CreateTemp("", "config.yaml")
Expect(err).To(BeNil())
defer os.Remove(tmp.Name())
_, err = io.Copy(tmp, resp.Body)
Expect(err).To(BeNil())
configs, err = readModelConfigsFromFile(tmp.Name())
config = configs[0]
Expect(err).To(BeNil())
Expect(config).ToNot(BeNil())
// two configs in config.yaml
Expect(config.Name).To(Equal("hermes-2-pro-mistral"))
valid, err = config.Validate()
Expect(err).To(BeNil())
Expect(valid).To(BeTrue())
})
})
It("Properly handles backend usecase matching", func() {
a := ModelConfig{
Name: "a",
}
Expect(a.HasUsecases(FLAG_ANY)).To(BeTrue()) // FLAG_ANY just means the config _exists_ essentially.
b := ModelConfig{
Name: "b",
Backend: "stablediffusion",
}
Expect(b.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(b.HasUsecases(FLAG_IMAGE)).To(BeTrue())
Expect(b.HasUsecases(FLAG_CHAT)).To(BeFalse())
c := ModelConfig{
Name: "c",
Backend: "llama-cpp",
TemplateConfig: TemplateConfig{
Chat: "chat",
},
}
Expect(c.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(c.HasUsecases(FLAG_IMAGE)).To(BeFalse())
Expect(c.HasUsecases(FLAG_COMPLETION)).To(BeFalse())
Expect(c.HasUsecases(FLAG_CHAT)).To(BeTrue())
d := ModelConfig{
Name: "d",
Backend: "llama-cpp",
TemplateConfig: TemplateConfig{
Chat: "chat",
Completion: "completion",
},
}
Expect(d.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(d.HasUsecases(FLAG_IMAGE)).To(BeFalse())
Expect(d.HasUsecases(FLAG_COMPLETION)).To(BeTrue())
Expect(d.HasUsecases(FLAG_CHAT)).To(BeTrue())
trueValue := true
e := ModelConfig{
Name: "e",
Backend: "llama-cpp",
TemplateConfig: TemplateConfig{
Completion: "completion",
},
Embeddings: &trueValue,
}
Expect(e.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(e.HasUsecases(FLAG_IMAGE)).To(BeFalse())
Expect(e.HasUsecases(FLAG_COMPLETION)).To(BeTrue())
Expect(e.HasUsecases(FLAG_CHAT)).To(BeFalse())
Expect(e.HasUsecases(FLAG_EMBEDDINGS)).To(BeTrue())
f := ModelConfig{
Name: "f",
Backend: "piper",
}
Expect(f.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(f.HasUsecases(FLAG_TTS)).To(BeTrue())
Expect(f.HasUsecases(FLAG_CHAT)).To(BeFalse())
g := ModelConfig{
Name: "g",
Backend: "whisper",
}
Expect(g.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(g.HasUsecases(FLAG_TRANSCRIPT)).To(BeTrue())
Expect(g.HasUsecases(FLAG_TTS)).To(BeFalse())
h := ModelConfig{
Name: "h",
Backend: "transformers-musicgen",
}
Expect(h.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(h.HasUsecases(FLAG_TRANSCRIPT)).To(BeFalse())
Expect(h.HasUsecases(FLAG_TTS)).To(BeTrue())
Expect(h.HasUsecases(FLAG_SOUND_GENERATION)).To(BeTrue())
knownUsecases := FLAG_CHAT | FLAG_COMPLETION
i := ModelConfig{
Name: "i",
Backend: "whisper",
// Earlier test checks parsing, this just needs to set final values
KnownUsecases: &knownUsecases,
}
Expect(i.HasUsecases(FLAG_ANY)).To(BeTrue())
Expect(i.HasUsecases(FLAG_TRANSCRIPT)).To(BeTrue())
Expect(i.HasUsecases(FLAG_TTS)).To(BeFalse())
Expect(i.HasUsecases(FLAG_COMPLETION)).To(BeTrue())
Expect(i.HasUsecases(FLAG_CHAT)).To(BeTrue())
})
It("Test Validate with invalid MCP config", func() {
tmp, err := os.CreateTemp("", "config.yaml")
Expect(err).To(BeNil())
defer os.Remove(tmp.Name())
_, err = tmp.WriteString(
`name: test-mcp
backend: "llama-cpp"
mcp:
stdio: |
{
"mcpServers": {
"ddg": {
"command": "/docker/docker",
"args": ["run", "-i"]
}
"weather": {
"command": "/docker/docker",
"args": ["run", "-i"]
}
}
}`)
Expect(err).ToNot(HaveOccurred())
configs, err := readModelConfigsFromFile(tmp.Name())
config := configs[0]
Expect(err).To(BeNil())
Expect(config).ToNot(BeNil())
valid, err := config.Validate()
Expect(err).To(HaveOccurred())
Expect(valid).To(BeFalse())
Expect(err.Error()).To(ContainSubstring("invalid MCP configuration"))
})
It("Test Validate with valid MCP config", func() {
tmp, err := os.CreateTemp("", "config.yaml")
Expect(err).To(BeNil())
defer os.Remove(tmp.Name())
_, err = tmp.WriteString(
`name: test-mcp-valid
backend: "llama-cpp"
mcp:
stdio: |
{
"mcpServers": {
"ddg": {
"command": "/docker/docker",
"args": ["run", "-i"]
},
"weather": {
"command": "/docker/docker",
"args": ["run", "-i"]
}
}
}`)
Expect(err).ToNot(HaveOccurred())
configs, err := readModelConfigsFromFile(tmp.Name())
config := configs[0]
Expect(err).To(BeNil())
Expect(config).ToNot(BeNil())
valid, err := config.Validate()
Expect(err).To(BeNil())
Expect(valid).To(BeTrue())
})
It("Test Validate rejects unmarshalable engine_args", func() {
// chan values cannot be JSON-marshalled. A valid YAML config could
// not produce one, but a Go caller stuffing a bad value would, and
// silently dropping it would change runtime behaviour.
cfg := &ModelConfig{
Backend: "vllm",
LLMConfig: LLMConfig{
EngineArgs: map[string]any{
"speculative_config": make(chan int),
},
},
}
valid, err := cfg.Validate()
Expect(valid).To(BeFalse())
Expect(err).ToNot(BeNil())
Expect(err.Error()).To(ContainSubstring("engine_args is not JSON-serialisable"))
})
It("Test Validate accepts well-formed engine_args", func() {
cfg := &ModelConfig{
Backend: "vllm",
LLMConfig: LLMConfig{
EngineArgs: map[string]any{
"data_parallel_size": 8,
"speculative_config": map[string]any{
"method": "ngram",
"num_speculative_tokens": 4,
},
},
},
}
valid, err := cfg.Validate()
Expect(err).To(BeNil())
Expect(valid).To(BeTrue())
})
})