File size: 1,645 Bytes
7def60a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 |
package main
// This is a wrapper to statisfy the GRPC service interface
// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
import (
"fmt"
"github.com/mudler/LocalAI/pkg/grpc/base"
pb "github.com/mudler/LocalAI/pkg/grpc/proto"
gpt4all "github.com/nomic-ai/gpt4all/gpt4all-bindings/golang"
)
type LLM struct {
base.SingleThread
gpt4all *gpt4all.Model
}
func (llm *LLM) Load(opts *pb.ModelOptions) error {
model, err := gpt4all.New(opts.ModelFile,
gpt4all.SetThreads(int(opts.Threads)),
gpt4all.SetLibrarySearchPath(opts.LibrarySearchPath))
llm.gpt4all = model
return err
}
func buildPredictOptions(opts *pb.PredictOptions) []gpt4all.PredictOption {
predictOptions := []gpt4all.PredictOption{
gpt4all.SetTemperature(float64(opts.Temperature)),
gpt4all.SetTopP(float64(opts.TopP)),
gpt4all.SetTopK(int(opts.TopK)),
gpt4all.SetTokens(int(opts.Tokens)),
}
if opts.Batch != 0 {
predictOptions = append(predictOptions, gpt4all.SetBatch(int(opts.Batch)))
}
return predictOptions
}
func (llm *LLM) Predict(opts *pb.PredictOptions) (string, error) {
return llm.gpt4all.Predict(opts.Prompt, buildPredictOptions(opts)...)
}
func (llm *LLM) PredictStream(opts *pb.PredictOptions, results chan string) error {
predictOptions := buildPredictOptions(opts)
go func() {
llm.gpt4all.SetTokenCallback(func(token string) bool {
results <- token
return true
})
_, err := llm.gpt4all.Predict(opts.Prompt, predictOptions...)
if err != nil {
fmt.Println("err: ", err)
}
llm.gpt4all.SetTokenCallback(nil)
close(results)
}()
return nil
}
|