2023-04-27 04:18:18 +00:00
|
|
|
package api
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
2023-04-29 07:22:09 +00:00
|
|
|
"regexp"
|
|
|
|
"strings"
|
2023-04-27 04:18:18 +00:00
|
|
|
"sync"
|
|
|
|
|
2023-05-03 09:45:22 +00:00
|
|
|
"github.com/donomii/go-rwkv.cpp"
|
2023-04-27 04:18:18 +00:00
|
|
|
model "github.com/go-skynet/LocalAI/pkg/model"
|
|
|
|
gpt2 "github.com/go-skynet/go-gpt2.cpp"
|
|
|
|
gptj "github.com/go-skynet/go-gpt4all-j.cpp"
|
|
|
|
llama "github.com/go-skynet/go-llama.cpp"
|
|
|
|
)
|
|
|
|
|
|
|
|
// mutex still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
|
|
|
|
var mutexMap sync.Mutex
|
|
|
|
var mutexes map[string]*sync.Mutex = make(map[string]*sync.Mutex)
|
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
func defaultLLamaOpts(c Config) []llama.ModelOption {
|
|
|
|
llamaOpts := []llama.ModelOption{}
|
|
|
|
if c.ContextSize != 0 {
|
|
|
|
llamaOpts = append(llamaOpts, llama.SetContext(c.ContextSize))
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
2023-05-05 09:20:06 +00:00
|
|
|
if c.F16 {
|
|
|
|
llamaOpts = append(llamaOpts, llama.EnableF16Memory)
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
2023-05-05 09:20:06 +00:00
|
|
|
if c.Embeddings {
|
|
|
|
llamaOpts = append(llamaOpts, llama.EnableEmbeddings)
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
2023-05-05 13:54:59 +00:00
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
return llamaOpts
|
|
|
|
}
|
2023-05-02 22:31:28 +00:00
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
func ModelEmbedding(s string, loader *model.ModelLoader, c Config) (func() ([]float32, error), error) {
|
|
|
|
if !c.Embeddings {
|
|
|
|
return nil, fmt.Errorf("endpoint disabled for this model by API configuration")
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
modelFile := c.Model
|
2023-05-02 22:31:28 +00:00
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
llamaOpts := defaultLLamaOpts(c)
|
2023-05-02 22:31:28 +00:00
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
var inferenceModel interface{}
|
|
|
|
var err error
|
|
|
|
if c.Backend == "" {
|
|
|
|
inferenceModel, err = loader.GreedyLoader(modelFile, llamaOpts, uint32(c.Threads))
|
2023-05-02 22:31:28 +00:00
|
|
|
} else {
|
2023-05-05 09:20:06 +00:00
|
|
|
inferenceModel, err = loader.BackendLoader(c.Backend, modelFile, llamaOpts, uint32(c.Threads))
|
|
|
|
}
|
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
var fn func() ([]float32, error)
|
|
|
|
switch model := inferenceModel.(type) {
|
|
|
|
case *llama.LLama:
|
|
|
|
fn = func() ([]float32, error) {
|
2023-05-05 13:56:02 +00:00
|
|
|
predictOptions := buildLLamaPredictOptions(c)
|
|
|
|
return model.Embeddings(s, predictOptions...)
|
2023-05-05 09:20:06 +00:00
|
|
|
}
|
|
|
|
default:
|
|
|
|
fn = func() ([]float32, error) {
|
|
|
|
return nil, fmt.Errorf("embeddings not supported by the backend")
|
|
|
|
}
|
2023-05-03 09:45:22 +00:00
|
|
|
}
|
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
return func() ([]float32, error) {
|
|
|
|
// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
|
|
|
|
mutexMap.Lock()
|
|
|
|
l, ok := mutexes[modelFile]
|
|
|
|
if !ok {
|
|
|
|
m := &sync.Mutex{}
|
|
|
|
mutexes[modelFile] = m
|
|
|
|
l = m
|
|
|
|
}
|
|
|
|
mutexMap.Unlock()
|
|
|
|
l.Lock()
|
|
|
|
defer l.Unlock()
|
|
|
|
|
|
|
|
return fn()
|
|
|
|
}, nil
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
|
|
|
|
2023-05-05 13:56:02 +00:00
|
|
|
func buildLLamaPredictOptions(c Config) []llama.PredictOption {
|
|
|
|
// Generate the prediction using the language model
|
|
|
|
predictOptions := []llama.PredictOption{
|
|
|
|
llama.SetTemperature(c.Temperature),
|
|
|
|
llama.SetTopP(c.TopP),
|
|
|
|
llama.SetTopK(c.TopK),
|
|
|
|
llama.SetTokens(c.Maxtokens),
|
|
|
|
llama.SetThreads(c.Threads),
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Mirostat != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetMirostat(c.Mirostat))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.MirostatETA != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetMirostatETA(c.MirostatETA))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.MirostatTAU != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetMirostatTAU(c.MirostatTAU))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Debug {
|
|
|
|
predictOptions = append(predictOptions, llama.Debug)
|
|
|
|
}
|
|
|
|
|
|
|
|
predictOptions = append(predictOptions, llama.SetStopWords(c.StopWords...))
|
|
|
|
|
|
|
|
if c.RepeatPenalty != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetPenalty(c.RepeatPenalty))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Keep != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetNKeep(c.Keep))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Batch != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetBatch(c.Batch))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.F16 {
|
|
|
|
predictOptions = append(predictOptions, llama.EnableF16KV)
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.IgnoreEOS {
|
|
|
|
predictOptions = append(predictOptions, llama.IgnoreEOS)
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Seed != 0 {
|
|
|
|
predictOptions = append(predictOptions, llama.SetSeed(c.Seed))
|
|
|
|
}
|
|
|
|
|
|
|
|
return predictOptions
|
|
|
|
}
|
|
|
|
|
2023-05-02 18:03:35 +00:00
|
|
|
func ModelInference(s string, loader *model.ModelLoader, c Config, tokenCallback func(string) bool) (func() (string, error), error) {
|
|
|
|
supportStreams := false
|
2023-04-27 04:18:18 +00:00
|
|
|
modelFile := c.Model
|
|
|
|
|
2023-05-05 09:20:06 +00:00
|
|
|
llamaOpts := defaultLLamaOpts(c)
|
2023-04-27 04:18:18 +00:00
|
|
|
|
2023-05-02 22:31:28 +00:00
|
|
|
var inferenceModel interface{}
|
|
|
|
var err error
|
|
|
|
if c.Backend == "" {
|
2023-05-05 09:20:06 +00:00
|
|
|
inferenceModel, err = loader.GreedyLoader(modelFile, llamaOpts, uint32(c.Threads))
|
2023-05-02 22:31:28 +00:00
|
|
|
} else {
|
2023-05-05 09:20:06 +00:00
|
|
|
inferenceModel, err = loader.BackendLoader(c.Backend, modelFile, llamaOpts, uint32(c.Threads))
|
2023-05-02 22:31:28 +00:00
|
|
|
}
|
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
2023-04-27 04:18:18 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
var fn func() (string, error)
|
|
|
|
|
2023-05-02 22:31:28 +00:00
|
|
|
switch model := inferenceModel.(type) {
|
2023-05-03 09:45:22 +00:00
|
|
|
case *rwkv.RwkvState:
|
|
|
|
supportStreams = true
|
|
|
|
|
|
|
|
fn = func() (string, error) {
|
|
|
|
stopWord := "\n"
|
|
|
|
if len(c.StopWords) > 0 {
|
|
|
|
stopWord = c.StopWords[0]
|
|
|
|
}
|
|
|
|
|
2023-05-04 15:32:23 +00:00
|
|
|
if err := model.ProcessInput(s); err != nil {
|
|
|
|
return "", err
|
|
|
|
}
|
|
|
|
|
2023-05-03 09:45:22 +00:00
|
|
|
response := model.GenerateResponse(c.Maxtokens, stopWord, float32(c.Temperature), float32(c.TopP), tokenCallback)
|
|
|
|
|
|
|
|
return response, nil
|
|
|
|
}
|
2023-05-02 22:31:28 +00:00
|
|
|
case *gpt2.StableLM:
|
2023-04-27 04:18:18 +00:00
|
|
|
fn = func() (string, error) {
|
|
|
|
// Generate the prediction using the language model
|
|
|
|
predictOptions := []gpt2.PredictOption{
|
|
|
|
gpt2.SetTemperature(c.Temperature),
|
|
|
|
gpt2.SetTopP(c.TopP),
|
|
|
|
gpt2.SetTopK(c.TopK),
|
|
|
|
gpt2.SetTokens(c.Maxtokens),
|
|
|
|
gpt2.SetThreads(c.Threads),
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Batch != 0 {
|
|
|
|
predictOptions = append(predictOptions, gpt2.SetBatch(c.Batch))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Seed != 0 {
|
|
|
|
predictOptions = append(predictOptions, gpt2.SetSeed(c.Seed))
|
|
|
|
}
|
|
|
|
|
2023-05-02 22:31:28 +00:00
|
|
|
return model.Predict(
|
2023-04-27 04:18:18 +00:00
|
|
|
s,
|
|
|
|
predictOptions...,
|
|
|
|
)
|
|
|
|
}
|
2023-05-02 22:31:28 +00:00
|
|
|
case *gpt2.GPT2:
|
2023-04-27 04:18:18 +00:00
|
|
|
fn = func() (string, error) {
|
|
|
|
// Generate the prediction using the language model
|
|
|
|
predictOptions := []gpt2.PredictOption{
|
|
|
|
gpt2.SetTemperature(c.Temperature),
|
|
|
|
gpt2.SetTopP(c.TopP),
|
|
|
|
gpt2.SetTopK(c.TopK),
|
|
|
|
gpt2.SetTokens(c.Maxtokens),
|
|
|
|
gpt2.SetThreads(c.Threads),
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Batch != 0 {
|
|
|
|
predictOptions = append(predictOptions, gpt2.SetBatch(c.Batch))
|
|
|
|
}
|
|
|
|
|
|
|
|
if c.Seed != 0 {
|
|
|
|
predictOptions = append(predictOptions, gpt2.SetSeed(c.Seed))
|
|
|
|
}
|
|
|
|
|
2023-05-02 22:31:28 +00:00
|
|
|
return model.Predict(
|
2023-04-27 04:18:18 +00:00
|
|
|
s,
|
|
|
|
predictOptions...,
|
|
|
|
)
|
|
|
|
}
|
2023-05-02 22:31:28 +00:00
|
|
|
case *gptj.GPTJ:
|
2023-04-27 04:18:18 +00:00
|
|
|
fn = func() (string, error) {
|
|
|
|
// Generate the prediction using the language model
|
|
|
|
predictOptions := []gptj.PredictOption{
|
|
|
|
gptj.SetTemperature(c.Temperature),
|
|
|
|
gptj.SetTopP(c.TopP),
|
|
|
|
gptj.SetTopK(c.TopK),
|
|
|
|
gptj.SetTokens(c.Maxtokens),
|
2023-05-02 14:07:18 +00:00
|
|
|
gptj.SetThreads(c.Threads),
|
2023-04-27 04:18:18 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if c.Batch != 0 {
|
|
|
|
predictOptions = append(predictOptions, gptj.SetBatch(c.Batch))
|
|
|
|
}
|
|
|
|
|
2023-05-02 14:07:18 +00:00
|
|
|
if c.Seed != 0 {
|
|
|
|
predictOptions = append(predictOptions, gptj.SetSeed(c.Seed))
|
|
|
|
}
|
|
|
|
|
2023-05-02 22:31:28 +00:00
|
|
|
return model.Predict(
|
2023-04-27 04:18:18 +00:00
|
|
|
s,
|
|
|
|
predictOptions...,
|
|
|
|
)
|
|
|
|
}
|
2023-05-02 22:31:28 +00:00
|
|
|
case *llama.LLama:
|
2023-05-02 18:03:35 +00:00
|
|
|
supportStreams = true
|
2023-04-27 04:18:18 +00:00
|
|
|
fn = func() (string, error) {
|
2023-05-02 18:03:35 +00:00
|
|
|
|
|
|
|
if tokenCallback != nil {
|
|
|
|
model.SetTokenCallback(tokenCallback)
|
|
|
|
}
|
|
|
|
|
2023-05-05 13:56:02 +00:00
|
|
|
predictOptions := buildLLamaPredictOptions(c)
|
2023-04-27 04:18:18 +00:00
|
|
|
|
2023-05-04 17:49:43 +00:00
|
|
|
str, er := model.Predict(
|
2023-04-27 04:18:18 +00:00
|
|
|
s,
|
|
|
|
predictOptions...,
|
|
|
|
)
|
2023-05-04 17:49:43 +00:00
|
|
|
// Seems that if we don't free the callback explicitly we leave functions registered (that might try to send on closed channels)
|
|
|
|
// For instance otherwise the API returns: {"error":{"code":500,"message":"send on closed channel","type":""}}
|
|
|
|
// after a stream event has occurred
|
|
|
|
model.SetTokenCallback(nil)
|
|
|
|
return str, er
|
2023-04-27 04:18:18 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return func() (string, error) {
|
|
|
|
// This is still needed, see: https://github.com/ggerganov/llama.cpp/discussions/784
|
|
|
|
mutexMap.Lock()
|
|
|
|
l, ok := mutexes[modelFile]
|
|
|
|
if !ok {
|
|
|
|
m := &sync.Mutex{}
|
|
|
|
mutexes[modelFile] = m
|
|
|
|
l = m
|
|
|
|
}
|
|
|
|
mutexMap.Unlock()
|
|
|
|
l.Lock()
|
|
|
|
defer l.Unlock()
|
|
|
|
|
2023-05-02 18:03:35 +00:00
|
|
|
res, err := fn()
|
|
|
|
if tokenCallback != nil && !supportStreams {
|
|
|
|
tokenCallback(res)
|
|
|
|
}
|
|
|
|
return res, err
|
2023-04-27 04:18:18 +00:00
|
|
|
}, nil
|
|
|
|
}
|
2023-04-29 07:22:09 +00:00
|
|
|
|
2023-05-02 18:03:35 +00:00
|
|
|
func ComputeChoices(predInput string, input *OpenAIRequest, config *Config, loader *model.ModelLoader, cb func(string, *[]Choice), tokenCallback func(string) bool) ([]Choice, error) {
|
2023-04-29 07:22:09 +00:00
|
|
|
result := []Choice{}
|
|
|
|
|
|
|
|
n := input.N
|
|
|
|
|
|
|
|
if input.N == 0 {
|
|
|
|
n = 1
|
|
|
|
}
|
|
|
|
|
|
|
|
// get the model function to call for the result
|
2023-05-02 18:03:35 +00:00
|
|
|
predFunc, err := ModelInference(predInput, loader, *config, tokenCallback)
|
2023-04-29 07:22:09 +00:00
|
|
|
if err != nil {
|
|
|
|
return result, err
|
|
|
|
}
|
|
|
|
|
|
|
|
for i := 0; i < n; i++ {
|
|
|
|
prediction, err := predFunc()
|
|
|
|
if err != nil {
|
|
|
|
return result, err
|
|
|
|
}
|
|
|
|
|
|
|
|
prediction = Finetune(*config, predInput, prediction)
|
|
|
|
cb(prediction, &result)
|
|
|
|
|
|
|
|
//result = append(result, Choice{Text: prediction})
|
|
|
|
|
|
|
|
}
|
|
|
|
return result, err
|
|
|
|
}
|
|
|
|
|
|
|
|
var cutstrings map[string]*regexp.Regexp = make(map[string]*regexp.Regexp)
|
|
|
|
var mu sync.Mutex = sync.Mutex{}
|
|
|
|
|
|
|
|
func Finetune(config Config, input, prediction string) string {
|
|
|
|
if config.Echo {
|
|
|
|
prediction = input + prediction
|
|
|
|
}
|
|
|
|
|
|
|
|
for _, c := range config.Cutstrings {
|
|
|
|
mu.Lock()
|
|
|
|
reg, ok := cutstrings[c]
|
|
|
|
if !ok {
|
|
|
|
cutstrings[c] = regexp.MustCompile(c)
|
|
|
|
reg = cutstrings[c]
|
|
|
|
}
|
|
|
|
mu.Unlock()
|
|
|
|
prediction = reg.ReplaceAllString(prediction, "")
|
|
|
|
}
|
|
|
|
|
|
|
|
for _, c := range config.TrimSpace {
|
|
|
|
prediction = strings.TrimSpace(strings.TrimPrefix(prediction, c))
|
|
|
|
}
|
|
|
|
return prediction
|
|
|
|
|
|
|
|
}
|