1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348 |
- package cmd
- import (
- "bufio"
- "context"
- "crypto/ed25519"
- "crypto/rand"
- "encoding/json"
- "encoding/pem"
- "errors"
- "fmt"
- "io"
- "log"
- "math"
- "net"
- "net/http"
- "os"
- "os/signal"
- "path/filepath"
- "runtime"
- "strconv"
- "strings"
- "sync/atomic"
- "syscall"
- "time"
- "github.com/containerd/console"
- "github.com/mattn/go-runewidth"
- "github.com/olekukonko/tablewriter"
- "github.com/spf13/cobra"
- "golang.org/x/crypto/ssh"
- "golang.org/x/term"
- "github.com/ollama/ollama/api"
- "github.com/ollama/ollama/envconfig"
- "github.com/ollama/ollama/format"
- "github.com/ollama/ollama/llama"
- "github.com/ollama/ollama/parser"
- "github.com/ollama/ollama/progress"
- "github.com/ollama/ollama/runner"
- "github.com/ollama/ollama/server"
- "github.com/ollama/ollama/types/model"
- "github.com/ollama/ollama/version"
- )
- var errModelfileNotFound = errors.New("specified Modelfile wasn't found")
- func getModelfileName(cmd *cobra.Command) (string, error) {
- filename, _ := cmd.Flags().GetString("file")
- if filename == "" {
- filename = "Modelfile"
- }
- absName, err := filepath.Abs(filename)
- if err != nil {
- return "", err
- }
- _, err = os.Stat(absName)
- if err != nil {
- return "", err
- }
- return absName, nil
- }
- func CreateHandler(cmd *cobra.Command, args []string) error {
- p := progress.NewProgress(os.Stderr)
- defer p.Stop()
- var reader io.Reader
- filename, err := getModelfileName(cmd)
- if os.IsNotExist(err) {
- if filename == "" {
- reader = strings.NewReader("FROM .\n")
- } else {
- return errModelfileNotFound
- }
- } else if err != nil {
- return err
- } else {
- f, err := os.Open(filename)
- if err != nil {
- return err
- }
- reader = f
- defer f.Close()
- }
- modelfile, err := parser.ParseFile(reader)
- if err != nil {
- return err
- }
- status := "gathering model components"
- spinner := progress.NewSpinner(status)
- p.Add(status, spinner)
- req, err := modelfile.CreateRequest(filepath.Dir(filename))
- if err != nil {
- return err
- }
- spinner.Stop()
- req.Name = args[0]
- quantize, _ := cmd.Flags().GetString("quantize")
- if quantize != "" {
- req.Quantize = quantize
- }
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- if len(req.Files) > 0 {
- fileMap := map[string]string{}
- for f, digest := range req.Files {
- if _, err := createBlob(cmd, client, f, digest, p); err != nil {
- return err
- }
- fileMap[filepath.Base(f)] = digest
- }
- req.Files = fileMap
- }
- if len(req.Adapters) > 0 {
- fileMap := map[string]string{}
- for f, digest := range req.Adapters {
- if _, err := createBlob(cmd, client, f, digest, p); err != nil {
- return err
- }
- fileMap[filepath.Base(f)] = digest
- }
- req.Adapters = fileMap
- }
- bars := make(map[string]*progress.Bar)
- fn := func(resp api.ProgressResponse) error {
- if resp.Digest != "" {
- bar, ok := bars[resp.Digest]
- if !ok {
- bar = progress.NewBar(fmt.Sprintf("pulling %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
- 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
- }
- if err := client.Create(cmd.Context(), req, fn); err != nil {
- if strings.Contains(err.Error(), "path or Modelfile are required") {
- return fmt.Errorf("the ollama server must be updated to use `ollama create` with this client")
- }
- return err
- }
- return nil
- }
- func createBlob(cmd *cobra.Command, client *api.Client, path string, digest string, p *progress.Progress) (string, error) {
- realPath, err := filepath.EvalSymlinks(path)
- if err != nil {
- return "", err
- }
- bin, err := os.Open(realPath)
- 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()
- var pw progressWriter
- status := fmt.Sprintf("copying file %s 0%%", digest)
- spinner := progress.NewSpinner(status)
- p.Add(status, spinner)
- defer spinner.Stop()
- 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("copying file %s %d%%", digest, int(100*pw.n.Load()/fileSize)))
- case <-done:
- spinner.SetMessage(fmt.Sprintf("copying file %s 100%%", digest))
- return
- }
- }
- }()
- 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 loadOrUnloadModel(cmd *cobra.Command, opts *runOptions) error {
- p := progress.NewProgress(os.Stderr)
- defer p.StopAndClear()
- spinner := progress.NewSpinner("")
- p.Add("", spinner)
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- req := &api.GenerateRequest{
- Model: opts.Model,
- KeepAlive: opts.KeepAlive,
- }
- return client.Generate(cmd.Context(), req, func(api.GenerateResponse) error { return nil })
- }
- func StopHandler(cmd *cobra.Command, args []string) error {
- opts := &runOptions{
- Model: args[0],
- KeepAlive: &api.Duration{Duration: 0},
- }
- if err := loadOrUnloadModel(cmd, opts); err != nil {
- if strings.Contains(err.Error(), "not found") {
- return fmt.Errorf("couldn't find model \"%s\" to stop", args[0])
- }
- }
- return nil
- }
- func RunHandler(cmd *cobra.Command, args []string) error {
- interactive := true
- opts := runOptions{
- 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
- }
- // Be quiet if we're redirecting to a pipe or file
- if !term.IsTerminal(int(os.Stdout.Fd())) {
- interactive = false
- }
- nowrap, err := cmd.Flags().GetBool("nowordwrap")
- if err != nil {
- return err
- }
- opts.WordWrap = !nowrap
- // 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
- }
- // TODO(jessegross): We should either find another way to know if this is
- // a vision model or remove the logic. Also consider that other modalities will
- // need different behavior anyways.
- opts.MultiModal = len(info.ProjectorInfo) != 0 || envconfig.NewEngine()
- opts.ParentModel = info.Details.ParentModel
- if interactive {
- if err := loadOrUnloadModel(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()
- }
- }
- return generateInteractive(cmd, opts)
- }
- return generate(cmd, opts)
- }
- func PushHandler(cmd *cobra.Command, args []string) error {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- insecure, err := cmd.Flags().GetBool("insecure")
- if err != nil {
- return err
- }
- p := progress.NewProgress(os.Stderr)
- defer p.Stop()
- bars := make(map[string]*progress.Bar)
- var status string
- var spinner *progress.Spinner
- fn := func(resp api.ProgressResponse) error {
- if resp.Digest != "" {
- if spinner != nil {
- spinner.Stop()
- }
- bar, ok := bars[resp.Digest]
- if !ok {
- bar = progress.NewBar(fmt.Sprintf("pushing %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
- bars[resp.Digest] = bar
- p.Add(resp.Digest, bar)
- }
- bar.Set(resp.Completed)
- } else if status != resp.Status {
- if spinner != nil {
- spinner.Stop()
- }
- status = resp.Status
- spinner = progress.NewSpinner(status)
- p.Add(status, spinner)
- }
- return nil
- }
- request := api.PushRequest{Name: args[0], Insecure: insecure}
- n := model.ParseName(args[0])
- 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")
- }
- return err
- }
- p.Stop()
- spinner.Stop()
- destination := n.String()
- if strings.HasSuffix(n.Host, ".ollama.ai") || strings.HasSuffix(n.Host, ".ollama.com") {
- destination = "https://ollama.com/" + strings.TrimSuffix(n.DisplayShortest(), ":latest")
- }
- fmt.Printf("\nYou can find your model at:\n\n")
- fmt.Printf("\t%s\n", destination)
- return nil
- }
- func ListHandler(cmd *cobra.Command, args []string) error {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- models, err := client.List(cmd.Context())
- if err != nil {
- return err
- }
- var data [][]string
- for _, m := range models.Models {
- if len(args) == 0 || strings.HasPrefix(strings.ToLower(m.Name), strings.ToLower(args[0])) {
- data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), format.HumanTime(m.ModifiedAt, "Never")})
- }
- }
- table := tablewriter.NewWriter(os.Stdout)
- table.SetHeader([]string{"NAME", "ID", "SIZE", "MODIFIED"})
- 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
- }
- 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))
- }
- var until string
- delta := time.Since(m.ExpiresAt)
- if delta > 0 {
- until = "Stopping..."
- } else {
- until = format.HumanTime(m.ExpiresAt, "Never")
- }
- data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), procStr, until})
- }
- }
- 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
- }
- func DeleteHandler(cmd *cobra.Command, args []string) error {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- // Unload the model if it's running before deletion
- opts := &runOptions{
- Model: args[0],
- KeepAlive: &api.Duration{Duration: 0},
- }
- if err := loadOrUnloadModel(cmd, opts); err != nil {
- if !strings.Contains(err.Error(), "not found") {
- return fmt.Errorf("unable to stop existing running model \"%s\": %s", args[0], err)
- }
- }
- for _, name := range args {
- req := api.DeleteRequest{Name: name}
- if err := client.Delete(cmd.Context(), &req); err != nil {
- return err
- }
- fmt.Printf("deleted '%s'\n", name)
- }
- return nil
- }
- func ShowHandler(cmd *cobra.Command, args []string) error {
- client, err := api.ClientFromEnvironment()
- 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.Print(resp.System)
- case "template":
- fmt.Print(resp.Template)
- }
- return nil
- }
- 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)
- }
- 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
- }
- func CopyHandler(cmd *cobra.Command, args []string) error {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- req := api.CopyRequest{Source: args[0], Destination: args[1]}
- if err := client.Copy(cmd.Context(), &req); err != nil {
- return err
- }
- fmt.Printf("copied '%s' to '%s'\n", args[0], args[1])
- return nil
- }
- func PullHandler(cmd *cobra.Command, args []string) error {
- insecure, err := cmd.Flags().GetBool("insecure")
- if err != nil {
- return err
- }
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- p := progress.NewProgress(os.Stderr)
- defer p.Stop()
- bars := make(map[string]*progress.Bar)
- var status string
- var spinner *progress.Spinner
- fn := func(resp api.ProgressResponse) error {
- if resp.Digest != "" {
- if spinner != nil {
- spinner.Stop()
- }
- bar, ok := bars[resp.Digest]
- if !ok {
- bar = progress.NewBar(fmt.Sprintf("pulling %s...", resp.Digest[7:19]), resp.Total, resp.Completed)
- bars[resp.Digest] = bar
- p.Add(resp.Digest, bar)
- }
- bar.Set(resp.Completed)
- } else if status != resp.Status {
- if spinner != nil {
- spinner.Stop()
- }
- status = resp.Status
- spinner = progress.NewSpinner(status)
- p.Add(status, spinner)
- }
- return nil
- }
- request := api.PullRequest{Name: args[0], Insecure: insecure}
- if err := client.Pull(cmd.Context(), &request, fn); err != nil {
- return err
- }
- return nil
- }
- type generateContextKey string
- type runOptions struct {
- 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 {
- 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 {
- 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
- }
- 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
- }
- if opts.Format == "json" {
- opts.Format = `"` + opts.Format + `"`
- }
- req := &api.ChatRequest{
- Model: opts.Model,
- Messages: opts.Messages,
- Format: json.RawMessage(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 {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- p := progress.NewProgress(os.Stderr)
- defer p.StopAndClear()
- spinner := progress.NewSpinner("")
- p.Add("", spinner)
- var latest api.GenerateResponse
- generateContext, ok := cmd.Context().Value(generateContextKey("context")).([]int)
- if !ok {
- generateContext = []int{}
- }
- 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 {
- p.StopAndClear()
- latest = response
- content := response.Response
- displayResponse(content, opts.WordWrap, state)
- return nil
- }
- if opts.MultiModal {
- opts.Prompt, opts.Images, err = extractFileData(opts.Prompt)
- if err != nil {
- return err
- }
- }
- if opts.Format == "json" {
- opts.Format = `"` + opts.Format + `"`
- }
- request := api.GenerateRequest{
- Model: opts.Model,
- Prompt: opts.Prompt,
- Context: generateContext,
- Images: opts.Images,
- Format: json.RawMessage(opts.Format),
- System: opts.System,
- Options: opts.Options,
- KeepAlive: opts.KeepAlive,
- }
- if err := client.Generate(ctx, &request, fn); err != nil {
- if errors.Is(err, context.Canceled) {
- return nil
- }
- return err
- }
- if opts.Prompt != "" {
- fmt.Println()
- fmt.Println()
- }
- if !latest.Done {
- return nil
- }
- verbose, err := cmd.Flags().GetBool("verbose")
- if err != nil {
- return err
- }
- if verbose {
- latest.Summary()
- }
- ctx = context.WithValue(cmd.Context(), generateContextKey("context"), latest.Context)
- cmd.SetContext(ctx)
- return nil
- }
- func RunServer(_ *cobra.Command, _ []string) error {
- if err := initializeKeypair(); err != nil {
- return err
- }
- ln, err := net.Listen("tcp", envconfig.Host().Host)
- if err != nil {
- return err
- }
- err = server.Serve(ln)
- if errors.Is(err, http.ErrServerClosed) {
- return nil
- }
- return err
- }
- 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
- }
- func checkServerHeartbeat(cmd *cobra.Command, _ []string) error {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return err
- }
- if err := client.Heartbeat(cmd.Context()); err != nil {
- if !strings.Contains(err.Error(), " refused") {
- return err
- }
- if err := startApp(cmd.Context(), client); err != nil {
- return errors.New("could not connect to ollama app, is it running?")
- }
- }
- return nil
- }
- func versionHandler(cmd *cobra.Command, _ []string) {
- client, err := api.ClientFromEnvironment()
- if err != nil {
- return
- }
- serverVersion, err := client.Version(cmd.Context())
- if err != nil {
- fmt.Println("Warning: could not connect to a running Ollama instance")
- }
- if serverVersion != "" {
- fmt.Printf("ollama version is %s\n", serverVersion)
- }
- if serverVersion != version.Version {
- fmt.Printf("Warning: client version is %s\n", version.Version)
- }
- }
- 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)
- cobra.EnableCommandSorting = false
- if runtime.GOOS == "windows" && term.IsTerminal(int(os.Stdout.Fd())) {
- console.ConsoleFromFile(os.Stdin) //nolint:errcheck
- }
- rootCmd := &cobra.Command{
- Use: "ollama",
- Short: "Large language model runner",
- SilenceUsage: true,
- SilenceErrors: true,
- CompletionOptions: cobra.CompletionOptions{
- DisableDefaultCmd: true,
- },
- Run: func(cmd *cobra.Command, args []string) {
- if version, _ := cmd.Flags().GetBool("version"); version {
- versionHandler(cmd, args)
- return
- }
- cmd.Print(cmd.UsageString())
- },
- }
- rootCmd.Flags().BoolP("version", "v", false, "Show version information")
- createCmd := &cobra.Command{
- Use: "create MODEL",
- Short: "Create a model from a Modelfile",
- Args: cobra.ExactArgs(1),
- PreRunE: checkServerHeartbeat,
- RunE: CreateHandler,
- }
- createCmd.Flags().StringP("file", "f", "", "Name of the Modelfile (default \"Modelfile\"")
- createCmd.Flags().StringP("quantize", "q", "", "Quantize model to this level (e.g. q4_0)")
- showCmd := &cobra.Command{
- Use: "show MODEL",
- Short: "Show information for a model",
- Args: cobra.ExactArgs(1),
- 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")
- 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)")
- runCmd.Flags().Bool("verbose", false, "Show timings for response")
- runCmd.Flags().Bool("insecure", false, "Use an insecure registry")
- runCmd.Flags().Bool("nowordwrap", false, "Don't wrap words to the next line automatically")
- runCmd.Flags().String("format", "", "Response format (e.g. json)")
- stopCmd := &cobra.Command{
- Use: "stop MODEL",
- Short: "Stop a running model",
- Args: cobra.ExactArgs(1),
- PreRunE: checkServerHeartbeat,
- RunE: StopHandler,
- }
- serveCmd := &cobra.Command{
- Use: "serve",
- Aliases: []string{"start"},
- Short: "Start ollama",
- Args: cobra.ExactArgs(0),
- RunE: RunServer,
- }
- pullCmd := &cobra.Command{
- Use: "pull MODEL",
- Short: "Pull a model from a registry",
- 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",
- Args: cobra.ExactArgs(1),
- PreRunE: checkServerHeartbeat,
- RunE: PushHandler,
- }
- pushCmd.Flags().Bool("insecure", false, "Use an insecure registry")
- listCmd := &cobra.Command{
- Use: "list",
- Aliases: []string{"ls"},
- Short: "List models",
- PreRunE: checkServerHeartbeat,
- RunE: ListHandler,
- }
- psCmd := &cobra.Command{
- Use: "ps",
- Short: "List running models",
- PreRunE: checkServerHeartbeat,
- RunE: ListRunningHandler,
- }
- copyCmd := &cobra.Command{
- Use: "cp SOURCE DESTINATION",
- Short: "Copy a model",
- Args: cobra.ExactArgs(2),
- PreRunE: checkServerHeartbeat,
- RunE: CopyHandler,
- }
- deleteCmd := &cobra.Command{
- Use: "rm MODEL [MODEL...]",
- Short: "Remove a model",
- Args: cobra.MinimumNArgs(1),
- PreRunE: checkServerHeartbeat,
- RunE: DeleteHandler,
- }
- runnerCmd := &cobra.Command{
- Use: "runner",
- Short: llama.PrintSystemInfo(),
- Hidden: true,
- RunE: func(cmd *cobra.Command, args []string) error {
- return runner.Execute(os.Args[1:])
- },
- FParseErrWhitelist: cobra.FParseErrWhitelist{UnknownFlags: true},
- }
- runnerCmd.SetHelpFunc(func(cmd *cobra.Command, args []string) {
- _ = runner.Execute(args[1:])
- })
- envVars := envconfig.AsMap()
- envs := []envconfig.EnvVar{envVars["OLLAMA_HOST"]}
- for _, cmd := range []*cobra.Command{
- createCmd,
- showCmd,
- runCmd,
- stopCmd,
- 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"],
- envVars["OLLAMA_SCHED_SPREAD"],
- envVars["OLLAMA_TMPDIR"],
- envVars["OLLAMA_FLASH_ATTENTION"],
- envVars["OLLAMA_KV_CACHE_TYPE"],
- envVars["OLLAMA_LLM_LIBRARY"],
- envVars["OLLAMA_GPU_OVERHEAD"],
- envVars["OLLAMA_LOAD_TIMEOUT"],
- })
- default:
- appendEnvDocs(cmd, envs)
- }
- }
- rootCmd.AddCommand(
- serveCmd,
- createCmd,
- showCmd,
- runCmd,
- stopCmd,
- pullCmd,
- pushCmd,
- listCmd,
- psCmd,
- copyCmd,
- deleteCmd,
- runnerCmd,
- )
- return rootCmd
- }
|