forked from mudler/LocalAI
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.go
92 lines (83 loc) · 2.25 KB
/
main.go
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
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
package main
import (
"os"
api "github.com/go-skynet/LocalAI/api"
model "github.com/go-skynet/LocalAI/pkg/model"
"github.com/jaypipes/ghw"
"github.com/rs/zerolog"
"github.com/rs/zerolog/log"
"github.com/urfave/cli/v2"
)
func main() {
log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stderr})
path, err := os.Getwd()
if err != nil {
log.Error().Msgf("error: %s", err.Error())
os.Exit(1)
}
threads := 4
cpu, err := ghw.CPU()
if err == nil {
threads = int(cpu.TotalCores)
}
app := &cli.App{
Name: "LocalAI",
Usage: "OpenAI compatible API for running LLaMA/GPT models locally on CPU with consumer grade hardware.",
Flags: []cli.Flag{
&cli.BoolFlag{
Name: "f16",
EnvVars: []string{"F16"},
},
&cli.BoolFlag{
Name: "debug",
EnvVars: []string{"DEBUG"},
},
&cli.IntFlag{
Name: "threads",
DefaultText: "Number of threads used for parallel computation. Usage of the number of physical cores in the system is suggested.",
EnvVars: []string{"THREADS"},
Value: threads,
},
&cli.StringFlag{
Name: "models-path",
DefaultText: "Path containing models used for inferencing",
EnvVars: []string{"MODELS_PATH"},
Value: path,
},
&cli.StringFlag{
Name: "address",
DefaultText: "Bind address for the API server.",
EnvVars: []string{"ADDRESS"},
Value: ":8080",
},
&cli.IntFlag{
Name: "context-size",
DefaultText: "Default context size of the model",
EnvVars: []string{"CONTEXT_SIZE"},
Value: 512,
},
},
Description: `
LocalAI is a drop-in replacement OpenAI API which runs inference locally.
Some of the models compatible are:
- Vicuna
- Koala
- GPT4ALL
- GPT4ALL-J
- Cerebras
- Alpaca
- StableLM (ggml quantized)
It uses llama.cpp, ggml and gpt4all as backend with golang c bindings.
`,
UsageText: `local-ai [options]`,
Copyright: "go-skynet authors",
Action: func(ctx *cli.Context) error {
return api.App(model.NewModelLoader(ctx.String("models-path")), ctx.Int("threads"), ctx.Int("context-size"), ctx.Bool("f16"), ctx.Bool("debug"), false).Listen(ctx.String("address"))
},
}
err = app.Run(os.Args)
if err != nil {
log.Error().Msgf("error: %s", err.Error())
os.Exit(1)
}
}