ollama/cmd/cmd.go

1413 lines
33 KiB
Go
Raw Normal View History

package cmd
import (
"archive/zip"
"bufio"
2023-11-14 22:07:40 +00:00
"bytes"
"context"
"crypto/ed25519"
"crypto/rand"
2023-11-14 22:07:40 +00:00
"crypto/sha256"
"encoding/pem"
2023-07-06 22:43:04 +00:00
"errors"
2023-07-06 16:24:49 +00:00
"fmt"
2023-07-18 21:01:19 +00:00
"io"
"log"
"math"
"net"
"net/http"
"os"
"os/signal"
"path/filepath"
2024-04-24 21:27:12 +00:00
"regexp"
"runtime"
2024-05-22 04:30:52 +00:00
"slices"
"strconv"
2023-07-07 17:12:58 +00:00
"strings"
"sync/atomic"
"syscall"
2023-07-06 22:43:04 +00:00
"time"
"github.com/containerd/console"
"github.com/mattn/go-runewidth"
2023-07-18 16:09:45 +00:00
"github.com/olekukonko/tablewriter"
2023-07-06 20:49:31 +00:00
"github.com/spf13/cobra"
"golang.org/x/crypto/ssh"
"golang.org/x/term"
2023-07-06 20:49:31 +00:00
"github.com/ollama/ollama/api"
"github.com/ollama/ollama/auth"
"github.com/ollama/ollama/envconfig"
"github.com/ollama/ollama/format"
"github.com/ollama/ollama/parser"
"github.com/ollama/ollama/progress"
"github.com/ollama/ollama/server"
"github.com/ollama/ollama/types/errtypes"
"github.com/ollama/ollama/types/model"
"github.com/ollama/ollama/version"
)
2023-07-20 23:09:23 +00:00
func CreateHandler(cmd *cobra.Command, args []string) error {
filename, _ := cmd.Flags().GetString("file")
filename, err := filepath.Abs(filename)
if err != nil {
return err
}
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-11-15 00:33:24 +00:00
p := progress.NewProgress(os.Stderr)
defer p.Stop()
2024-04-30 17:55:19 +00:00
f, err := os.Open(filename)
2023-11-14 22:07:40 +00:00
if err != nil {
return err
}
2024-04-30 17:55:19 +00:00
defer f.Close()
2023-11-14 22:07:40 +00:00
modelfile, err := parser.ParseFile(f)
2023-11-14 22:07:40 +00:00
if err != nil {
return err
}
home, err := os.UserHomeDir()
if err != nil {
return err
}
status := "transferring model data"
spinner := progress.NewSpinner(status)
2023-11-16 00:59:49 +00:00
p.Add(status, spinner)
defer p.Stop()
2023-11-16 00:59:49 +00:00
2024-04-30 17:55:19 +00:00
for i := range modelfile.Commands {
switch modelfile.Commands[i].Name {
2023-11-14 22:07:40 +00:00
case "model", "adapter":
2024-04-30 17:55:19 +00:00
path := modelfile.Commands[i].Args
2023-11-14 22:07:40 +00:00
if path == "~" {
path = home
} else if strings.HasPrefix(path, "~/") {
path = filepath.Join(home, path[2:])
}
if !filepath.IsAbs(path) {
path = filepath.Join(filepath.Dir(filename), path)
}
fi, err := os.Stat(path)
2024-04-30 17:55:19 +00:00
if errors.Is(err, os.ErrNotExist) && modelfile.Commands[i].Name == "model" {
2023-11-15 18:57:09 +00:00
continue
2023-11-14 22:07:40 +00:00
} else if err != nil {
return err
}
if fi.IsDir() {
2024-04-22 18:02:25 +00:00
// this is likely a safetensors or pytorch directory
// TODO make this work w/ adapters
tempfile, err := tempZipFiles(path)
if err != nil {
return err
}
2024-04-22 18:02:25 +00:00
defer os.RemoveAll(tempfile)
2024-04-22 18:02:25 +00:00
path = tempfile
2023-11-14 22:07:40 +00:00
}
digest, err := createBlob(cmd, client, path, spinner)
if err != nil {
2023-11-14 22:07:40 +00:00
return err
}
2024-04-30 17:55:19 +00:00
modelfile.Commands[i].Args = "@" + digest
2023-11-14 22:07:40 +00:00
}
}
2023-07-17 21:14:41 +00:00
2024-04-22 18:02:25 +00:00
bars := make(map[string]*progress.Bar)
fn := func(resp api.ProgressResponse) error {
2023-11-15 00:33:24 +00:00
if resp.Digest != "" {
spinner.Stop()
bar, ok := bars[resp.Digest]
if !ok {
bar = progress.NewBar(fmt.Sprintf("pulling %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
2023-11-15 00:33:24 +00:00
bars[resp.Digest] = bar
p.Add(resp.Digest, bar)
}
bar.Set(resp.Completed)
} else if status != resp.Status {
spinner.Stop()
status = resp.Status
spinner = progress.NewSpinner(status)
p.Add(status, spinner)
}
return nil
}
quantize, _ := cmd.Flags().GetString("quantize")
2024-04-05 15:49:04 +00:00
request := api.CreateRequest{Name: args[0], Modelfile: modelfile.String(), Quantize: quantize}
2023-10-12 22:56:40 +00:00
if err := client.Create(cmd.Context(), &request, fn); err != nil {
return err
}
return nil
}
2024-04-22 18:02:25 +00:00
func tempZipFiles(path string) (string, error) {
tempfile, err := os.CreateTemp("", "ollama-tf")
if err != nil {
return "", err
}
defer tempfile.Close()
2024-04-25 21:41:30 +00:00
detectContentType := func(path string) (string, error) {
f, err := os.Open(path)
2024-04-22 18:02:25 +00:00
if err != nil {
return "", err
}
2024-04-25 21:41:30 +00:00
defer f.Close()
2024-04-22 18:02:25 +00:00
2024-04-25 21:41:30 +00:00
var b bytes.Buffer
b.Grow(512)
2024-04-22 18:02:25 +00:00
2024-04-25 21:41:30 +00:00
if _, err := io.CopyN(&b, f, 512); err != nil && !errors.Is(err, io.EOF) {
return "", err
}
contentType, _, _ := strings.Cut(http.DetectContentType(b.Bytes()), ";")
return contentType, nil
2024-04-22 18:02:25 +00:00
}
2024-04-25 21:41:30 +00:00
glob := func(pattern, contentType string) ([]string, error) {
matches, err := filepath.Glob(pattern)
if err != nil {
return nil, err
}
for _, safetensor := range matches {
if ct, err := detectContentType(safetensor); err != nil {
return nil, err
} else if ct != contentType {
return nil, fmt.Errorf("invalid content type: expected %s for %s", ct, safetensor)
2024-04-22 18:02:25 +00:00
}
2024-04-25 21:41:30 +00:00
}
return matches, nil
}
var files []string
if st, _ := glob(filepath.Join(path, "model*.safetensors"), "application/octet-stream"); len(st) > 0 {
// safetensors files might be unresolved git lfs references; skip if they are
// covers model-x-of-y.safetensors, model.fp32-x-of-y.safetensors, model.safetensors
files = append(files, st...)
} else if st, _ := glob(filepath.Join(path, "adapters.safetensors"), "application/octet-stream"); len(st) > 0 {
// covers adapters.safetensors
files = append(files, st...)
} else if st, _ := glob(filepath.Join(path, "adapter_model.safetensors"), "application/octet-stream"); len(st) > 0 {
// covers adapter_model.safetensors
files = append(files, st...)
2024-04-25 21:41:30 +00:00
} else if pt, _ := glob(filepath.Join(path, "pytorch_model*.bin"), "application/zip"); len(pt) > 0 {
// pytorch files might also be unresolved git lfs references; skip if they are
// covers pytorch_model-x-of-y.bin, pytorch_model.fp32-x-of-y.bin, pytorch_model.bin
files = append(files, pt...)
2024-05-08 23:07:46 +00:00
} else if pt, _ := glob(filepath.Join(path, "consolidated*.pth"), "application/zip"); len(pt) > 0 {
2024-04-25 21:41:30 +00:00
// pytorch files might also be unresolved git lfs references; skip if they are
// covers consolidated.x.pth, consolidated.pth
files = append(files, pt...)
} else {
return "", errors.New("no safetensors or torch files found")
}
// add configuration files, json files are detected as text/plain
js, err := glob(filepath.Join(path, "*.json"), "text/plain")
if err != nil {
return "", err
}
files = append(files, js...)
2024-06-07 21:55:56 +00:00
// bert models require a nested config.json
// TODO(mxyng): merge this with the glob above
js, err = glob(filepath.Join(path, "**/*.json"), "text/plain")
if err != nil {
return "", err
}
files = append(files, js...)
2024-04-25 21:41:30 +00:00
if tks, _ := glob(filepath.Join(path, "tokenizer.model"), "application/octet-stream"); len(tks) > 0 {
// add tokenizer.model if it exists, tokenizer.json is automatically picked up by the previous glob
// tokenizer.model might be a unresolved git lfs reference; error if it is
files = append(files, tks...)
} else if tks, _ := glob(filepath.Join(path, "**/tokenizer.model"), "text/plain"); len(tks) > 0 {
// some times tokenizer.model is in a subdirectory (e.g. meta-llama/Meta-Llama-3-8B)
files = append(files, tks...)
}
zipfile := zip.NewWriter(tempfile)
defer zipfile.Close()
2024-04-25 21:41:30 +00:00
for _, file := range files {
f, err := os.Open(file)
if err != nil {
2024-04-22 18:02:25 +00:00
return "", err
}
2024-04-25 21:41:30 +00:00
defer f.Close()
2024-04-22 18:02:25 +00:00
fi, err := f.Stat()
if err != nil {
return "", err
}
2024-04-25 21:41:30 +00:00
zfi, err := zip.FileInfoHeader(fi)
2024-04-22 18:02:25 +00:00
if err != nil {
return "", err
}
2024-06-07 21:55:56 +00:00
zfi.Name, err = filepath.Rel(path, file)
if err != nil {
return "", err
}
2024-04-25 21:41:30 +00:00
zf, err := zipfile.CreateHeader(zfi)
2024-04-22 18:02:25 +00:00
if err != nil {
return "", err
}
2024-04-25 21:41:30 +00:00
if _, err := io.Copy(zf, f); err != nil {
2024-04-22 18:02:25 +00:00
return "", err
}
}
return tempfile.Name(), nil
}
func createBlob(cmd *cobra.Command, client *api.Client, path string, spinner *progress.Spinner) (string, error) {
bin, err := os.Open(path)
if err != nil {
return "", err
}
defer bin.Close()
// Get file info to retrieve the size
fileInfo, err := bin.Stat()
if err != nil {
return "", err
}
fileSize := fileInfo.Size()
hash := sha256.New()
if _, err := io.Copy(hash, bin); err != nil {
return "", err
}
if _, err := bin.Seek(0, io.SeekStart); err != nil {
return "", err
}
var pw progressWriter
status := "transferring model data 0%"
spinner.SetMessage(status)
done := make(chan struct{})
defer close(done)
go func() {
ticker := time.NewTicker(60 * time.Millisecond)
defer ticker.Stop()
for {
select {
case <-ticker.C:
spinner.SetMessage(fmt.Sprintf("transferring model data %d%%", int(100*pw.n.Load()/fileSize)))
case <-done:
spinner.SetMessage("transferring model data 100%")
return
}
}
}()
digest := fmt.Sprintf("sha256:%x", hash.Sum(nil))
if err = client.CreateBlob(cmd.Context(), digest, io.TeeReader(bin, &pw)); err != nil {
return "", err
}
return digest, nil
}
type progressWriter struct {
n atomic.Int64
}
func (w *progressWriter) Write(p []byte) (n int, err error) {
w.n.Add(int64(len(p)))
return len(p), nil
}
func RunHandler(cmd *cobra.Command, args []string) error {
interactive := true
opts := runOptions{
cmd: defer stating model info until necessary (#5248) This commit changes the 'ollama run' command to defer fetching model information until it really needs it. That is, when in interactive mode. It also removes one such case where the model information is fetch in duplicate, just before calling generateInteractive and then again, first thing, in generateInteractive. This positively impacts the performance of the command: ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.168 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.220 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.217 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 4% cpu 0.652 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 5% cpu 0.498 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 3% cpu 0.479 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total
2024-06-25 03:14:03 +00:00
Model: args[0],
WordWrap: os.Getenv("TERM") == "xterm-256color",
Options: map[string]interface{}{},
}
format, err := cmd.Flags().GetString("format")
if err != nil {
return err
}
opts.Format = format
keepAlive, err := cmd.Flags().GetString("keepalive")
if err != nil {
return err
}
if keepAlive != "" {
d, err := time.ParseDuration(keepAlive)
if err != nil {
return err
}
opts.KeepAlive = &api.Duration{Duration: d}
}
prompts := args[1:]
// prepend stdin to the prompt if provided
if !term.IsTerminal(int(os.Stdin.Fd())) {
in, err := io.ReadAll(os.Stdin)
if err != nil {
return err
}
prompts = append([]string{string(in)}, prompts...)
opts.WordWrap = false
interactive = false
}
opts.Prompt = strings.Join(prompts, " ")
if len(prompts) > 0 {
interactive = false
}
nowrap, err := cmd.Flags().GetBool("nowordwrap")
if err != nil {
return err
}
opts.WordWrap = !nowrap
cmd: defer stating model info until necessary (#5248) This commit changes the 'ollama run' command to defer fetching model information until it really needs it. That is, when in interactive mode. It also removes one such case where the model information is fetch in duplicate, just before calling generateInteractive and then again, first thing, in generateInteractive. This positively impacts the performance of the command: ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.168 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.220 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.217 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 4% cpu 0.652 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 5% cpu 0.498 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 3% cpu 0.479 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total
2024-06-25 03:14:03 +00:00
// Fill out the rest of the options based on information about the
// model.
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
name := args[0]
info, err := func() (*api.ShowResponse, error) {
showReq := &api.ShowRequest{Name: name}
info, err := client.Show(cmd.Context(), showReq)
var se api.StatusError
if errors.As(err, &se) && se.StatusCode == http.StatusNotFound {
if err := PullHandler(cmd, []string{name}); err != nil {
return nil, err
}
return client.Show(cmd.Context(), &api.ShowRequest{Name: name})
}
return info, err
}()
if err != nil {
return err
}
cmd: defer stating model info until necessary (#5248) This commit changes the 'ollama run' command to defer fetching model information until it really needs it. That is, when in interactive mode. It also removes one such case where the model information is fetch in duplicate, just before calling generateInteractive and then again, first thing, in generateInteractive. This positively impacts the performance of the command: ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.168 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.220 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.217 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 4% cpu 0.652 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 5% cpu 0.498 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 3% cpu 0.479 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total
2024-06-25 03:14:03 +00:00
opts.MultiModal = slices.Contains(info.Details.Families, "clip")
opts.ParentModel = info.Details.ParentModel
if interactive {
2024-07-26 20:39:38 +00:00
if err := loadModel(cmd, &opts); err != nil {
return err
}
for _, msg := range info.Messages {
switch msg.Role {
case "user":
fmt.Printf(">>> %s\n", msg.Content)
case "assistant":
state := &displayResponseState{}
displayResponse(msg.Content, opts.WordWrap, state)
fmt.Println()
fmt.Println()
}
}
cmd: defer stating model info until necessary (#5248) This commit changes the 'ollama run' command to defer fetching model information until it really needs it. That is, when in interactive mode. It also removes one such case where the model information is fetch in duplicate, just before calling generateInteractive and then again, first thing, in generateInteractive. This positively impacts the performance of the command: ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.168 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.220 total ; time ./before run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./before run llama3 'hi' 0.02s user 0.01s system 2% cpu 1.217 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 4% cpu 0.652 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 5% cpu 0.498 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with or would you like to chat? ./after run llama3 'hi' 0.01s user 0.01s system 3% cpu 0.479 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total ; time ./after run llama3 'hi' Hi! It's nice to meet you. Is there something I can help you with, or would you like to chat? ./after run llama3 'hi' 0.02s user 0.01s system 5% cpu 0.507 total
2024-06-25 03:14:03 +00:00
return generateInteractive(cmd, opts)
}
return generate(cmd, opts)
2023-07-06 18:18:40 +00:00
}
func errFromUnknownKey(unknownKeyErr error) error {
// find SSH public key in the error message
sshKeyPattern := `ssh-\w+ [^\s"]+`
re := regexp.MustCompile(sshKeyPattern)
matches := re.FindStringSubmatch(unknownKeyErr.Error())
if len(matches) > 0 {
serverPubKey := matches[0]
localPubKey, err := auth.GetPublicKey()
if err != nil {
return unknownKeyErr
}
if runtime.GOOS == "linux" && serverPubKey != localPubKey {
// try the ollama service public key
svcPubKey, err := os.ReadFile("/usr/share/ollama/.ollama/id_ed25519.pub")
if err != nil {
return unknownKeyErr
}
localPubKey = strings.TrimSpace(string(svcPubKey))
}
// check if the returned public key matches the local public key, this prevents adding a remote key to the user's account
if serverPubKey != localPubKey {
return unknownKeyErr
}
var msg strings.Builder
msg.WriteString(unknownKeyErr.Error())
msg.WriteString("\n\nYour ollama key is:\n")
msg.WriteString(localPubKey)
msg.WriteString("\nAdd your key at:\n")
msg.WriteString("https://ollama.com/settings/keys")
return errors.New(msg.String())
}
return unknownKeyErr
}
2023-07-20 23:09:23 +00:00
func PushHandler(cmd *cobra.Command, args []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
insecure, err := cmd.Flags().GetBool("insecure")
if err != nil {
return err
}
2023-11-15 00:33:24 +00:00
p := progress.NewProgress(os.Stderr)
defer p.Stop()
bars := make(map[string]*progress.Bar)
var status string
var spinner *progress.Spinner
2023-11-15 00:33:24 +00:00
2023-07-19 01:51:30 +00:00
fn := func(resp api.ProgressResponse) error {
2023-11-15 00:33:24 +00:00
if resp.Digest != "" {
if spinner != nil {
spinner.Stop()
}
2023-11-15 00:33:24 +00:00
bar, ok := bars[resp.Digest]
if !ok {
bar = progress.NewBar(fmt.Sprintf("pushing %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
2023-11-15 00:33:24 +00:00
bars[resp.Digest] = bar
p.Add(resp.Digest, bar)
}
bar.Set(resp.Completed)
} else if status != resp.Status {
if spinner != nil {
spinner.Stop()
}
2023-11-15 00:33:24 +00:00
status = resp.Status
spinner = progress.NewSpinner(status)
p.Add(status, spinner)
}
return nil
}
2023-11-15 00:33:24 +00:00
request := api.PushRequest{Name: args[0], Insecure: insecure}
2023-10-12 22:56:40 +00:00
if err := client.Push(cmd.Context(), &request, fn); err != nil {
if spinner != nil {
spinner.Stop()
}
if strings.Contains(err.Error(), "access denied") {
return errors.New("you are not authorized to push to this namespace, create the model under a namespace you own")
}
host := model.ParseName(args[0]).Host
isOllamaHost := strings.HasSuffix(host, ".ollama.ai") || strings.HasSuffix(host, ".ollama.com")
if strings.Contains(err.Error(), errtypes.UnknownOllamaKeyErrMsg) && isOllamaHost {
// the user has not added their ollama key to ollama.com
// re-throw an error with a more user-friendly message
return errFromUnknownKey(err)
}
2023-11-15 00:33:24 +00:00
return err
}
2023-11-16 00:59:49 +00:00
spinner.Stop()
2023-11-15 00:33:24 +00:00
return nil
}
2023-07-20 23:09:23 +00:00
func ListHandler(cmd *cobra.Command, args []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-07-18 16:09:45 +00:00
2023-10-12 22:56:40 +00:00
models, err := client.List(cmd.Context())
2023-07-18 16:09:45 +00:00
if err != nil {
return err
}
var data [][]string
for _, m := range models.Models {
2023-07-18 21:01:19 +00:00
if len(args) == 0 || strings.HasPrefix(m.Name, args[0]) {
data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), format.HumanTime(m.ModifiedAt, "Never")})
2023-07-18 21:01:19 +00:00
}
2023-07-18 16:09:45 +00:00
}
table := tablewriter.NewWriter(os.Stdout)
2023-08-29 03:50:24 +00:00
table.SetHeader([]string{"NAME", "ID", "SIZE", "MODIFIED"})
2023-07-18 16:09:45 +00:00
table.SetHeaderAlignment(tablewriter.ALIGN_LEFT)
table.SetAlignment(tablewriter.ALIGN_LEFT)
table.SetHeaderLine(false)
table.SetBorder(false)
table.SetNoWhiteSpace(true)
table.SetTablePadding(" ")
2023-07-18 16:09:45 +00:00
table.AppendBulk(data)
table.Render()
return nil
}
func ListRunningHandler(cmd *cobra.Command, args []string) error {
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
models, err := client.ListRunning(cmd.Context())
if err != nil {
return err
}
var data [][]string
for _, m := range models.Models {
if len(args) == 0 || strings.HasPrefix(m.Name, args[0]) {
var procStr string
switch {
case m.SizeVRAM == 0:
procStr = "100% CPU"
case m.SizeVRAM == m.Size:
procStr = "100% GPU"
case m.SizeVRAM > m.Size || m.Size == 0:
procStr = "Unknown"
default:
sizeCPU := m.Size - m.SizeVRAM
cpuPercent := math.Round(float64(sizeCPU) / float64(m.Size) * 100)
procStr = fmt.Sprintf("%d%%/%d%% CPU/GPU", int(cpuPercent), int(100-cpuPercent))
}
data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), procStr, format.HumanTime(m.ExpiresAt, "Never")})
}
}
table := tablewriter.NewWriter(os.Stdout)
table.SetHeader([]string{"NAME", "ID", "SIZE", "PROCESSOR", "UNTIL"})
table.SetHeaderAlignment(tablewriter.ALIGN_LEFT)
table.SetAlignment(tablewriter.ALIGN_LEFT)
table.SetHeaderLine(false)
table.SetBorder(false)
table.SetNoWhiteSpace(true)
table.SetTablePadding(" ")
table.AppendBulk(data)
table.Render()
return nil
}
2023-07-20 23:09:23 +00:00
func DeleteHandler(cmd *cobra.Command, args []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-07-20 23:09:23 +00:00
for _, name := range args {
req := api.DeleteRequest{Name: name}
2023-10-12 22:56:40 +00:00
if err := client.Delete(cmd.Context(), &req); err != nil {
return err
}
fmt.Printf("deleted '%s'\n", name)
2023-07-20 23:09:23 +00:00
}
return nil
}
2023-09-06 18:04:17 +00:00
func ShowHandler(cmd *cobra.Command, args []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
2023-09-06 18:04:17 +00:00
if err != nil {
return err
}
license, errLicense := cmd.Flags().GetBool("license")
modelfile, errModelfile := cmd.Flags().GetBool("modelfile")
parameters, errParams := cmd.Flags().GetBool("parameters")
system, errSystem := cmd.Flags().GetBool("system")
template, errTemplate := cmd.Flags().GetBool("template")
for _, boolErr := range []error{errLicense, errModelfile, errParams, errSystem, errTemplate} {
if boolErr != nil {
return errors.New("error retrieving flags")
}
}
flagsSet := 0
showType := ""
if license {
flagsSet++
showType = "license"
}
if modelfile {
flagsSet++
showType = "modelfile"
}
if parameters {
flagsSet++
showType = "parameters"
}
if system {
flagsSet++
showType = "system"
}
if template {
flagsSet++
showType = "template"
}
if flagsSet > 1 {
return errors.New("only one of '--license', '--modelfile', '--parameters', '--system', or '--template' can be specified")
}
req := api.ShowRequest{Name: args[0]}
resp, err := client.Show(cmd.Context(), &req)
if err != nil {
return err
}
if flagsSet == 1 {
switch showType {
case "license":
fmt.Println(resp.License)
case "modelfile":
fmt.Println(resp.Modelfile)
case "parameters":
fmt.Println(resp.Parameters)
case "system":
fmt.Println(resp.System)
case "template":
fmt.Println(resp.Template)
}
return nil
2023-09-06 18:04:17 +00:00
}
return showInfo(resp, os.Stdout)
}
func showInfo(resp *api.ShowResponse, w io.Writer) error {
tableRender := func(header string, rows func() [][]string) {
fmt.Fprintln(w, " ", header)
table := tablewriter.NewWriter(w)
table.SetAlignment(tablewriter.ALIGN_LEFT)
table.SetBorder(false)
table.SetNoWhiteSpace(true)
table.SetTablePadding(" ")
switch header {
case "Template", "System", "License":
table.SetColWidth(100)
}
table.AppendBulk(rows())
table.Render()
fmt.Fprintln(w)
2023-09-06 18:04:17 +00:00
}
tableRender("Model", func() (rows [][]string) {
if resp.ModelInfo != nil {
arch := resp.ModelInfo["general.architecture"].(string)
rows = append(rows, []string{"", "architecture", arch})
rows = append(rows, []string{"", "parameters", format.HumanNumber(uint64(resp.ModelInfo["general.parameter_count"].(float64)))})
rows = append(rows, []string{"", "context length", strconv.FormatFloat(resp.ModelInfo[fmt.Sprintf("%s.context_length", arch)].(float64), 'f', -1, 64)})
rows = append(rows, []string{"", "embedding length", strconv.FormatFloat(resp.ModelInfo[fmt.Sprintf("%s.embedding_length", arch)].(float64), 'f', -1, 64)})
} else {
rows = append(rows, []string{"", "architecture", resp.Details.Family})
rows = append(rows, []string{"", "parameters", resp.Details.ParameterSize})
}
rows = append(rows, []string{"", "quantization", resp.Details.QuantizationLevel})
return
})
if resp.ProjectorInfo != nil {
tableRender("Projector", func() (rows [][]string) {
arch := resp.ProjectorInfo["general.architecture"].(string)
rows = append(rows, []string{"", "architecture", arch})
rows = append(rows, []string{"", "parameters", format.HumanNumber(uint64(resp.ProjectorInfo["general.parameter_count"].(float64)))})
rows = append(rows, []string{"", "embedding length", strconv.FormatFloat(resp.ProjectorInfo[fmt.Sprintf("%s.vision.embedding_length", arch)].(float64), 'f', -1, 64)})
rows = append(rows, []string{"", "dimensions", strconv.FormatFloat(resp.ProjectorInfo[fmt.Sprintf("%s.vision.projection_dim", arch)].(float64), 'f', -1, 64)})
return
})
}
if resp.Parameters != "" {
tableRender("Parameters", func() (rows [][]string) {
scanner := bufio.NewScanner(strings.NewReader(resp.Parameters))
for scanner.Scan() {
if text := scanner.Text(); text != "" {
rows = append(rows, append([]string{""}, strings.Fields(text)...))
}
}
return
})
}
head := func(s string, n int) (rows [][]string) {
scanner := bufio.NewScanner(strings.NewReader(s))
for scanner.Scan() && (len(rows) < n || n < 0) {
if text := scanner.Text(); text != "" {
rows = append(rows, []string{"", strings.TrimSpace(text)})
}
}
return
}
if resp.System != "" {
tableRender("System", func() [][]string {
return head(resp.System, 2)
})
}
if resp.License != "" {
tableRender("License", func() [][]string {
return head(resp.License, 2)
})
}
return nil
}
2023-07-24 15:27:28 +00:00
func CopyHandler(cmd *cobra.Command, args []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-07-24 15:27:28 +00:00
req := api.CopyRequest{Source: args[0], Destination: args[1]}
2023-10-12 22:56:40 +00:00
if err := client.Copy(cmd.Context(), &req); err != nil {
2023-07-24 15:27:28 +00:00
return err
}
fmt.Printf("copied '%s' to '%s'\n", args[0], args[1])
return nil
}
2023-07-20 23:09:23 +00:00
func PullHandler(cmd *cobra.Command, args []string) error {
insecure, err := cmd.Flags().GetBool("insecure")
if err != nil {
return err
}
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-11-15 00:33:24 +00:00
p := progress.NewProgress(os.Stderr)
defer p.Stop()
bars := make(map[string]*progress.Bar)
var status string
var spinner *progress.Spinner
2023-11-15 00:33:24 +00:00
2023-07-19 01:51:30 +00:00
fn := func(resp api.ProgressResponse) error {
2023-11-15 00:33:24 +00:00
if resp.Digest != "" {
if spinner != nil {
spinner.Stop()
}
2023-11-15 00:33:24 +00:00
bar, ok := bars[resp.Digest]
if !ok {
bar = progress.NewBar(fmt.Sprintf("pulling %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
2023-11-15 00:33:24 +00:00
bars[resp.Digest] = bar
p.Add(resp.Digest, bar)
}
bar.Set(resp.Completed)
} else if status != resp.Status {
if spinner != nil {
spinner.Stop()
}
2023-11-15 00:33:24 +00:00
status = resp.Status
spinner = progress.NewSpinner(status)
p.Add(status, spinner)
}
return nil
}
2023-07-07 14:22:37 +00:00
2023-11-15 00:33:24 +00:00
request := api.PullRequest{Name: args[0], Insecure: insecure}
2023-10-12 22:56:40 +00:00
if err := client.Pull(cmd.Context(), &request, fn); err != nil {
2023-11-15 00:33:24 +00:00
return err
}
return nil
2023-07-06 20:49:31 +00:00
}
type generateContextKey string
type runOptions struct {
2024-01-25 20:12:36 +00:00
Model string
ParentModel string
Prompt string
Messages []api.Message
WordWrap bool
Format string
System string
Images []api.ImageData
Options map[string]interface{}
MultiModal bool
KeepAlive *api.Duration
}
type displayResponseState struct {
lineLength int
wordBuffer string
}
func displayResponse(content string, wordWrap bool, state *displayResponseState) {
termWidth, _, _ := term.GetSize(int(os.Stdout.Fd()))
if wordWrap && termWidth >= 10 {
for _, ch := range content {
2024-05-16 00:24:17 +00:00
if state.lineLength+1 > termWidth-5 {
if runewidth.StringWidth(state.wordBuffer) > termWidth-10 {
fmt.Printf("%s%c", state.wordBuffer, ch)
state.wordBuffer = ""
state.lineLength = 0
continue
}
// backtrack the length of the last word and clear to the end of the line
a := runewidth.StringWidth(state.wordBuffer)
if a > 0 {
2024-05-30 17:38:07 +00:00
fmt.Printf("\x1b[%dD", a)
}
fmt.Printf("\x1b[K\n")
fmt.Printf("%s%c", state.wordBuffer, ch)
chWidth := runewidth.RuneWidth(ch)
state.lineLength = runewidth.StringWidth(state.wordBuffer) + chWidth
} else {
fmt.Print(string(ch))
state.lineLength += runewidth.RuneWidth(ch)
if runewidth.RuneWidth(ch) >= 2 {
state.wordBuffer = ""
continue
2024-05-16 00:24:17 +00:00
}
switch ch {
case ' ':
state.wordBuffer = ""
case '\n':
state.lineLength = 0
default:
state.wordBuffer += string(ch)
}
}
}
} else {
fmt.Printf("%s%s", state.wordBuffer, content)
if len(state.wordBuffer) > 0 {
state.wordBuffer = ""
}
}
}
func chat(cmd *cobra.Command, opts runOptions) (*api.Message, error) {
client, err := api.ClientFromEnvironment()
if err != nil {
return nil, err
}
p := progress.NewProgress(os.Stderr)
defer p.StopAndClear()
spinner := progress.NewSpinner("")
p.Add("", spinner)
cancelCtx, cancel := context.WithCancel(cmd.Context())
defer cancel()
sigChan := make(chan os.Signal, 1)
signal.Notify(sigChan, syscall.SIGINT)
go func() {
<-sigChan
cancel()
}()
var state *displayResponseState = &displayResponseState{}
var latest api.ChatResponse
var fullResponse strings.Builder
var role string
fn := func(response api.ChatResponse) error {
p.StopAndClear()
latest = response
role = response.Message.Role
content := response.Message.Content
fullResponse.WriteString(content)
displayResponse(content, opts.WordWrap, state)
return nil
}
req := &api.ChatRequest{
Model: opts.Model,
Messages: opts.Messages,
Format: opts.Format,
Options: opts.Options,
}
if opts.KeepAlive != nil {
req.KeepAlive = opts.KeepAlive
}
if err := client.Chat(cancelCtx, req, fn); err != nil {
if errors.Is(err, context.Canceled) {
return nil, nil
}
return nil, err
}
if len(opts.Messages) > 0 {
fmt.Println()
fmt.Println()
}
verbose, err := cmd.Flags().GetBool("verbose")
if err != nil {
return nil, err
}
if verbose {
latest.Summary()
}
return &api.Message{Role: role, Content: fullResponse.String()}, nil
}
func generate(cmd *cobra.Command, opts runOptions) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
2023-09-18 19:26:56 +00:00
if err != nil {
return err
2023-09-18 19:26:56 +00:00
}
2023-07-07 17:12:58 +00:00
2023-11-15 00:58:51 +00:00
p := progress.NewProgress(os.Stderr)
defer p.StopAndClear()
2023-11-15 00:58:51 +00:00
spinner := progress.NewSpinner("")
p.Add("", spinner)
var latest api.GenerateResponse
generateContext, ok := cmd.Context().Value(generateContextKey("context")).([]int)
if !ok {
generateContext = []int{}
}
2023-10-12 22:56:40 +00:00
ctx, cancel := context.WithCancel(cmd.Context())
defer cancel()
sigChan := make(chan os.Signal, 1)
signal.Notify(sigChan, syscall.SIGINT)
go func() {
<-sigChan
cancel()
}()
var state *displayResponseState = &displayResponseState{}
fn := func(response api.GenerateResponse) error {
2023-11-15 00:58:51 +00:00
p.StopAndClear()
2023-09-18 19:26:56 +00:00
latest = response
content := response.Response
2023-07-13 01:18:06 +00:00
displayResponse(content, opts.WordWrap, state)
2023-09-18 19:26:56 +00:00
return nil
}
2023-07-07 21:04:43 +00:00
if opts.MultiModal {
opts.Prompt, opts.Images, err = extractFileData(opts.Prompt)
if err != nil {
return err
}
}
2023-10-12 22:56:40 +00:00
request := api.GenerateRequest{
Model: opts.Model,
Prompt: opts.Prompt,
Context: generateContext,
Images: opts.Images,
Format: opts.Format,
System: opts.System,
Options: opts.Options,
KeepAlive: opts.KeepAlive,
2023-10-12 22:56:40 +00:00
}
if err := client.Generate(ctx, &request, fn); err != nil {
if errors.Is(err, context.Canceled) {
return nil
2023-07-07 21:04:43 +00:00
}
return err
2023-09-18 19:26:56 +00:00
}
if opts.Prompt != "" {
2023-07-07 17:12:58 +00:00
fmt.Println()
fmt.Println()
2023-09-18 19:26:56 +00:00
}
2023-07-13 01:18:06 +00:00
if !latest.Done {
return nil
}
2023-09-18 19:26:56 +00:00
verbose, err := cmd.Flags().GetBool("verbose")
if err != nil {
return err
2023-09-18 19:26:56 +00:00
}
2023-07-18 18:59:42 +00:00
2023-09-18 19:26:56 +00:00
if verbose {
latest.Summary()
2023-07-07 17:12:58 +00:00
}
2023-07-06 20:49:31 +00:00
ctx = context.WithValue(cmd.Context(), generateContextKey("context"), latest.Context)
cmd.SetContext(ctx)
return nil
2023-07-06 20:49:31 +00:00
}
func RunServer(_ *cobra.Command, _ []string) error {
2023-09-21 00:49:48 +00:00
if err := initializeKeypair(); err != nil {
return err
}
2024-07-03 23:44:57 +00:00
ln, err := net.Listen("tcp", envconfig.Host().Host)
if err != nil {
return err
}
2023-07-04 04:47:00 +00:00
err = server.Serve(ln)
if errors.Is(err, http.ErrServerClosed) {
return nil
}
return err
2023-07-04 04:47:00 +00:00
}
func initializeKeypair() error {
home, err := os.UserHomeDir()
if err != nil {
return err
}
privKeyPath := filepath.Join(home, ".ollama", "id_ed25519")
pubKeyPath := filepath.Join(home, ".ollama", "id_ed25519.pub")
_, err = os.Stat(privKeyPath)
if os.IsNotExist(err) {
fmt.Printf("Couldn't find '%s'. Generating new private key.\n", privKeyPath)
cryptoPublicKey, cryptoPrivateKey, err := ed25519.GenerateKey(rand.Reader)
if err != nil {
return err
}
privateKeyBytes, err := ssh.MarshalPrivateKey(cryptoPrivateKey, "")
if err != nil {
return err
}
if err := os.MkdirAll(filepath.Dir(privKeyPath), 0o755); err != nil {
return fmt.Errorf("could not create directory %w", err)
}
if err := os.WriteFile(privKeyPath, pem.EncodeToMemory(privateKeyBytes), 0o600); err != nil {
return err
}
sshPublicKey, err := ssh.NewPublicKey(cryptoPublicKey)
if err != nil {
return err
}
publicKeyBytes := ssh.MarshalAuthorizedKey(sshPublicKey)
if err := os.WriteFile(pubKeyPath, publicKeyBytes, 0o644); err != nil {
return err
}
fmt.Printf("Your new public key is: \n\n%s\n", publicKeyBytes)
}
return nil
}
2023-10-12 22:56:40 +00:00
func checkServerHeartbeat(cmd *cobra.Command, _ []string) error {
2023-10-09 19:18:26 +00:00
client, err := api.ClientFromEnvironment()
if err != nil {
return err
}
2023-10-12 22:56:40 +00:00
if err := client.Heartbeat(cmd.Context()); err != nil {
if !strings.Contains(err.Error(), " refused") {
2023-07-31 21:38:10 +00:00
return err
}
if err := startApp(cmd.Context(), client); err != nil {
2024-08-01 21:52:15 +00:00
return errors.New("could not connect to ollama app, is it running?")
}
}
return nil
}
2023-11-22 17:41:02 +00:00
func versionHandler(cmd *cobra.Command, _ []string) {
client, err := api.ClientFromEnvironment()
if err != nil {
return
}
serverVersion, err := client.Version(cmd.Context())
if err != nil {
2023-12-01 20:10:27 +00:00
fmt.Println("Warning: could not connect to a running Ollama instance")
}
if serverVersion != "" {
fmt.Printf("ollama version is %s\n", serverVersion)
2023-11-22 17:41:02 +00:00
}
if serverVersion != version.Version {
2023-12-01 20:10:27 +00:00
fmt.Printf("Warning: client version is %s\n", version.Version)
}
2023-11-22 17:41:02 +00:00
}
func appendEnvDocs(cmd *cobra.Command, envs []envconfig.EnvVar) {
if len(envs) == 0 {
return
}
envUsage := `
Environment Variables:
`
for _, e := range envs {
envUsage += fmt.Sprintf(" %-24s %s\n", e.Name, e.Description)
}
cmd.SetUsageTemplate(cmd.UsageTemplate() + envUsage)
}
func NewCLI() *cobra.Command {
log.SetFlags(log.LstdFlags | log.Lshortfile)
2023-11-22 17:41:02 +00:00
cobra.EnableCommandSorting = false
if runtime.GOOS == "windows" {
console.ConsoleFromFile(os.Stdin) //nolint:errcheck
}
rootCmd := &cobra.Command{
2023-08-14 18:15:53 +00:00
Use: "ollama",
Short: "Large language model runner",
SilenceUsage: true,
SilenceErrors: true,
CompletionOptions: cobra.CompletionOptions{
DisableDefaultCmd: true,
},
2023-11-22 17:41:02 +00:00
Run: func(cmd *cobra.Command, args []string) {
if version, _ := cmd.Flags().GetBool("version"); version {
versionHandler(cmd, args)
return
}
cmd.Print(cmd.UsageString())
},
}
2023-11-22 17:41:02 +00:00
rootCmd.Flags().BoolP("version", "v", false, "Show version information")
createCmd := &cobra.Command{
Use: "create MODEL",
Short: "Create a model from a Modelfile",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(1),
PreRunE: checkServerHeartbeat,
RunE: CreateHandler,
}
createCmd.Flags().StringP("file", "f", "Modelfile", "Name of the Modelfile")
createCmd.Flags().StringP("quantize", "q", "", "Quantize model to this level (e.g. q4_0)")
2023-09-06 18:04:17 +00:00
showCmd := &cobra.Command{
Use: "show MODEL",
Short: "Show information for a model",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(1),
2023-09-06 18:04:17 +00:00
PreRunE: checkServerHeartbeat,
RunE: ShowHandler,
}
showCmd.Flags().Bool("license", false, "Show license of a model")
showCmd.Flags().Bool("modelfile", false, "Show Modelfile of a model")
showCmd.Flags().Bool("parameters", false, "Show parameters of a model")
showCmd.Flags().Bool("template", false, "Show template of a model")
showCmd.Flags().Bool("system", false, "Show system message of a model")
2023-09-06 18:04:17 +00:00
runCmd := &cobra.Command{
Use: "run MODEL [PROMPT]",
Short: "Run a model",
Args: cobra.MinimumNArgs(1),
PreRunE: checkServerHeartbeat,
RunE: RunHandler,
}
runCmd.Flags().String("keepalive", "", "Duration to keep a model loaded (e.g. 5m)")
2023-07-13 01:18:06 +00:00
runCmd.Flags().Bool("verbose", false, "Show timings for response")
2023-08-22 04:56:56 +00:00
runCmd.Flags().Bool("insecure", false, "Use an insecure registry")
runCmd.Flags().Bool("nowordwrap", false, "Don't wrap words to the next line automatically")
2023-11-14 02:54:02 +00:00
runCmd.Flags().String("format", "", "Response format (e.g. json)")
serveCmd := &cobra.Command{
Use: "serve",
Aliases: []string{"start"},
Short: "Start ollama",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(0),
2023-07-06 20:49:31 +00:00
RunE: RunServer,
}
pullCmd := &cobra.Command{
Use: "pull MODEL",
Short: "Pull a model from a registry",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(1),
PreRunE: checkServerHeartbeat,
RunE: PullHandler,
}
pullCmd.Flags().Bool("insecure", false, "Use an insecure registry")
pushCmd := &cobra.Command{
Use: "push MODEL",
Short: "Push a model to a registry",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(1),
PreRunE: checkServerHeartbeat,
RunE: PushHandler,
}
pushCmd.Flags().Bool("insecure", false, "Use an insecure registry")
2023-07-18 16:09:45 +00:00
listCmd := &cobra.Command{
Use: "list",
2023-07-20 22:28:27 +00:00
Aliases: []string{"ls"},
Short: "List models",
PreRunE: checkServerHeartbeat,
RunE: ListHandler,
2023-07-20 23:09:23 +00:00
}
psCmd := &cobra.Command{
Use: "ps",
Short: "List running models",
PreRunE: checkServerHeartbeat,
RunE: ListRunningHandler,
}
2023-07-24 15:27:28 +00:00
copyCmd := &cobra.Command{
2024-05-01 19:39:05 +00:00
Use: "cp SOURCE DESTINATION",
Short: "Copy a model",
2023-10-18 18:57:22 +00:00
Args: cobra.ExactArgs(2),
PreRunE: checkServerHeartbeat,
RunE: CopyHandler,
2023-07-24 15:27:28 +00:00
}
2023-07-20 23:09:23 +00:00
deleteCmd := &cobra.Command{
2023-10-18 18:57:22 +00:00
Use: "rm MODEL [MODEL...]",
Short: "Remove a model",
Args: cobra.MinimumNArgs(1),
PreRunE: checkServerHeartbeat,
RunE: DeleteHandler,
2023-07-18 16:09:45 +00:00
}
envVars := envconfig.AsMap()
envs := []envconfig.EnvVar{envVars["OLLAMA_HOST"]}
for _, cmd := range []*cobra.Command{
createCmd,
showCmd,
runCmd,
pullCmd,
pushCmd,
listCmd,
psCmd,
copyCmd,
deleteCmd,
serveCmd,
} {
switch cmd {
case runCmd:
appendEnvDocs(cmd, []envconfig.EnvVar{envVars["OLLAMA_HOST"], envVars["OLLAMA_NOHISTORY"]})
case serveCmd:
appendEnvDocs(cmd, []envconfig.EnvVar{
envVars["OLLAMA_DEBUG"],
envVars["OLLAMA_HOST"],
envVars["OLLAMA_KEEP_ALIVE"],
envVars["OLLAMA_MAX_LOADED_MODELS"],
envVars["OLLAMA_MAX_QUEUE"],
envVars["OLLAMA_MODELS"],
envVars["OLLAMA_NUM_PARALLEL"],
envVars["OLLAMA_NOPRUNE"],
envVars["OLLAMA_ORIGINS"],
2024-07-23 22:14:28 +00:00
envVars["OLLAMA_SCHED_SPREAD"],
envVars["OLLAMA_TMPDIR"],
envVars["OLLAMA_FLASH_ATTENTION"],
envVars["OLLAMA_LLM_LIBRARY"],
envVars["OLLAMA_GPU_OVERHEAD"],
envVars["OLLAMA_LOAD_TIMEOUT"],
})
default:
appendEnvDocs(cmd, envs)
}
}
rootCmd.AddCommand(
serveCmd,
createCmd,
2023-09-06 18:04:17 +00:00
showCmd,
runCmd,
pullCmd,
pushCmd,
2023-07-18 16:09:45 +00:00
listCmd,
psCmd,
2023-07-24 15:27:28 +00:00
copyCmd,
2023-07-20 23:09:23 +00:00
deleteCmd,
)
return rootCmd
}