Skip to content

Commit

Permalink
fix(config): set better defaults for inferencing (#1822)
Browse files Browse the repository at this point in the history 
* 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 #1373 and #1723

* Adapt tests

* Also pre-initialize default seed
  • Loading branch information
@mudler
mudler committed Mar 13, 2024
1 parent bc8f648 commit f895d06
Show file tree
Hide file tree
Showing 12 changed files with 233 additions and 131 deletions.
2 changes: 1 addition & 1 deletion core/backend/embeddings.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@ func ModelEmbedding(s string, tokens []int, loader *model.ModelLoader, backendCo

opts := modelOpts(backendConfig, appConfig, []model.Option{
model.WithLoadGRPCLoadModelOpts(grpcOpts),
model.WithThreads(uint32(backendConfig.Threads)),
model.WithThreads(uint32(*backendConfig.Threads)),
model.WithAssetDir(appConfig.AssetsDestination),
model.WithModel(modelFile),
model.WithContext(appConfig.Context),
Expand Down
6 changes: 3 additions & 3 deletions core/backend/image.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,14 +9,14 @@ import (

func ImageGeneration(height, width, mode, step, seed int, positive_prompt, negative_prompt, src, dst string, loader *model.ModelLoader, backendConfig config.BackendConfig, appConfig *config.ApplicationConfig) (func() error, error) {
threads := backendConfig.Threads
if threads == 0 && appConfig.Threads != 0 {
threads = appConfig.Threads
if *threads == 0 && appConfig.Threads != 0 {
threads = &appConfig.Threads
}
gRPCOpts := gRPCModelOpts(backendConfig)
opts := modelOpts(backendConfig, appConfig, []model.Option{
model.WithBackendString(backendConfig.Backend),
model.WithAssetDir(appConfig.AssetsDestination),
model.WithThreads(uint32(threads)),
model.WithThreads(uint32(*threads)),
model.WithContext(appConfig.Context),
model.WithModel(backendConfig.Model),
model.WithLoadGRPCLoadModelOpts(gRPCOpts),
Expand Down
6 changes: 3 additions & 3 deletions core/backend/llm.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,8 +29,8 @@ type TokenUsage struct {
func ModelInference(ctx context.Context, s string, images []string, loader *model.ModelLoader, c config.BackendConfig, o *config.ApplicationConfig, tokenCallback func(string, TokenUsage) bool) (func() (LLMResponse, error), error) {
modelFile := c.Model
threads := c.Threads
if threads == 0 && o.Threads != 0 {
threads = o.Threads
if *threads == 0 && o.Threads != 0 {
threads = &o.Threads
}
grpcOpts := gRPCModelOpts(c)

Expand All @@ -39,7 +39,7 @@ func ModelInference(ctx context.Context, s string, images []string, loader *mode

opts := modelOpts(c, o, []model.Option{
model.WithLoadGRPCLoadModelOpts(grpcOpts),
model.WithThreads(uint32(threads)), // some models uses this to allocate threads during startup
model.WithThreads(uint32(*threads)), // some models uses this to allocate threads during startup
model.WithAssetDir(o.AssetsDestination),
model.WithModel(modelFile),
model.WithContext(o.Context),
Expand Down
45 changes: 23 additions & 22 deletions core/backend/options.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -46,15 +46,15 @@ func gRPCModelOpts(c config.BackendConfig) *pb.ModelOptions {
CFGScale: c.Diffusers.CFGScale,
LoraAdapter: c.LoraAdapter,
LoraScale: c.LoraScale,
F16Memory: c.F16,
F16Memory: *c.F16,
LoraBase: c.LoraBase,
IMG2IMG: c.Diffusers.IMG2IMG,
CLIPModel: c.Diffusers.ClipModel,
CLIPSubfolder: c.Diffusers.ClipSubFolder,
CLIPSkip: int32(c.Diffusers.ClipSkip),
ControlNet: c.Diffusers.ControlNet,
ContextSize: int32(c.ContextSize),
Seed: int32(c.Seed),
ContextSize: int32(*c.ContextSize),
Seed: int32(*c.Seed),
NBatch: int32(b),
NoMulMatQ: c.NoMulMatQ,
DraftModel: c.DraftModel,
Expand All @@ -72,18 +72,18 @@ func gRPCModelOpts(c config.BackendConfig) *pb.ModelOptions {
YarnBetaSlow: c.YarnBetaSlow,
NGQA: c.NGQA,
RMSNormEps: c.RMSNormEps,
MLock: c.MMlock,
MLock: *c.MMlock,
RopeFreqBase: c.RopeFreqBase,
RopeScaling: c.RopeScaling,
Type: c.ModelType,
RopeFreqScale: c.RopeFreqScale,
NUMA: c.NUMA,
Embeddings: c.Embeddings,
LowVRAM: c.LowVRAM,
NGPULayers: int32(c.NGPULayers),
MMap: c.MMap,
LowVRAM: *c.LowVRAM,
NGPULayers: int32(*c.NGPULayers),
MMap: *c.MMap,
MainGPU: c.MainGPU,
Threads: int32(c.Threads),
Threads: int32(*c.Threads),
TensorSplit: c.TensorSplit,
// AutoGPTQ
ModelBaseName: c.AutoGPTQ.ModelBaseName,
Expand All @@ -102,36 +102,37 @@ func gRPCPredictOpts(c config.BackendConfig, modelPath string) *pb.PredictOption
os.MkdirAll(filepath.Dir(p), 0755)
promptCachePath = p
}

return &pb.PredictOptions{
Temperature: float32(c.Temperature),
TopP: float32(c.TopP),
Temperature: float32(*c.Temperature),
TopP: float32(*c.TopP),
NDraft: c.NDraft,
TopK: int32(c.TopK),
Tokens: int32(c.Maxtokens),
Threads: int32(c.Threads),
TopK: int32(*c.TopK),
Tokens: int32(*c.Maxtokens),
Threads: int32(*c.Threads),
PromptCacheAll: c.PromptCacheAll,
PromptCacheRO: c.PromptCacheRO,
PromptCachePath: promptCachePath,
F16KV: c.F16,
DebugMode: c.Debug,
F16KV: *c.F16,
DebugMode: *c.Debug,
Grammar: c.Grammar,
NegativePromptScale: c.NegativePromptScale,
RopeFreqBase: c.RopeFreqBase,
RopeFreqScale: c.RopeFreqScale,
NegativePrompt: c.NegativePrompt,
Mirostat: int32(c.LLMConfig.Mirostat),
MirostatETA: float32(c.LLMConfig.MirostatETA),
MirostatTAU: float32(c.LLMConfig.MirostatTAU),
Debug: c.Debug,
Mirostat: int32(*c.LLMConfig.Mirostat),
MirostatETA: float32(*c.LLMConfig.MirostatETA),
MirostatTAU: float32(*c.LLMConfig.MirostatTAU),
Debug: *c.Debug,
StopPrompts: c.StopWords,
Repeat: int32(c.RepeatPenalty),
NKeep: int32(c.Keep),
Batch: int32(c.Batch),
IgnoreEOS: c.IgnoreEOS,
Seed: int32(c.Seed),
Seed: int32(*c.Seed),
FrequencyPenalty: float32(c.FrequencyPenalty),
MLock: c.MMlock,
MMap: c.MMap,
MLock: *c.MMlock,
MMap: *c.MMap,
MainGPU: c.MainGPU,
TensorSplit: c.TensorSplit,
TailFreeSamplingZ: float32(c.TFZ),
Expand Down
4 changes: 2 additions & 2 deletions core/backend/transcript.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ func ModelTranscription(audio, language string, ml *model.ModelLoader, backendCo
model.WithBackendString(model.WhisperBackend),
model.WithModel(backendConfig.Model),
model.WithContext(appConfig.Context),
model.WithThreads(uint32(backendConfig.Threads)),
model.WithThreads(uint32(*backendConfig.Threads)),
model.WithAssetDir(appConfig.AssetsDestination),
})

Expand All @@ -33,6 +33,6 @@ func ModelTranscription(audio, language string, ml *model.ModelLoader, backendCo
return whisperModel.AudioTranscription(context.Background(), &proto.TranscriptRequest{
Dst: audio,
Language: language,
Threads: uint32(backendConfig.Threads),
Threads: uint32(*backendConfig.Threads),
})
}
Loading

0 comments on commit f895d06

Please sign in to comment.