api embed docs (#5282)

auth/auth.go

@@ -3,68 +3,49 @@ package auth
 import (
 	"bytes"
 	"context"
-	"crypto/ed25519"
 	"crypto/rand"
 	"encoding/base64"
-	"encoding/pem"
-	"errors"
 	"fmt"
 	"io"
 	"log/slog"
 	"os"
 	"path/filepath"
+	"strings"
 
 	"golang.org/x/crypto/ssh"
 )
 
 const defaultPrivateKey = "id_ed25519"
 
-func privateKey() (ssh.Signer, error) {
+func keyPath() (string, error) {
 	home, err := os.UserHomeDir()
 	if err != nil {
-		return nil, err
+		return "", err
 	}
 
-	keyPath := filepath.Join(home, ".ollama", defaultPrivateKey)
-	privateKeyFile, err := os.ReadFile(keyPath)
-	if errors.Is(err, os.ErrNotExist) {
-		err := initializeKeypair()
-		if err != nil {
-			return nil, err
-		}
-
-		return privateKey()
-	} else if err != nil {
-		slog.Info(fmt.Sprintf("Failed to load private key: %v", err))
-		return nil, err
-	}
-
-	return ssh.ParsePrivateKey(privateKeyFile)
+	return filepath.Join(home, ".ollama", defaultPrivateKey), nil
 }
 
-func GetPublicKey() (ssh.PublicKey, error) {
-	// try to read pubkey first
-	home, err := os.UserHomeDir()
+func GetPublicKey() (string, error) {
+	keyPath, err := keyPath()
 	if err != nil {
-		return nil, err
+		return "", err
 	}
 
-	pubkeyPath := filepath.Join(home, ".ollama", defaultPrivateKey+".pub")
-	pubKeyFile, err := os.ReadFile(pubkeyPath)
-	if errors.Is(err, os.ErrNotExist) {
-		// try from privateKey
-		privateKey, err := privateKey()
-		if err != nil {
-			return nil, fmt.Errorf("failed to read public key: %w", err)
-		}
-
-		return privateKey.PublicKey(), nil
-	} else if err != nil {
-		return nil, fmt.Errorf("failed to read public key: %w", err)
+	privateKeyFile, err := os.ReadFile(keyPath)
+	if err != nil {
+		slog.Info(fmt.Sprintf("Failed to load private key: %v", err))
+		return "", err
 	}
 
-	pubKey, _, _, _, err := ssh.ParseAuthorizedKey(pubKeyFile)
-	return pubKey, err
+	privateKey, err := ssh.ParsePrivateKey(privateKeyFile)
+	if err != nil {
+		return "", err
+	}
+
+	publicKey := ssh.MarshalAuthorizedKey(privateKey.PublicKey())
+
+	return strings.TrimSpace(string(publicKey)), nil
 }
 
 func NewNonce(r io.Reader, length int) (string, error) {
@@ -77,20 +58,25 @@ func NewNonce(r io.Reader, length int) (string, error) {
 }
 
 func Sign(ctx context.Context, bts []byte) (string, error) {
-	privateKey, err := privateKey()
+	keyPath, err := keyPath()
+	if err != nil {
+		return "", err
+	}
+
+	privateKeyFile, err := os.ReadFile(keyPath)
+	if err != nil {
+		slog.Info(fmt.Sprintf("Failed to load private key: %v", err))
+		return "", err
+	}
+
+	privateKey, err := ssh.ParsePrivateKey(privateKeyFile)
 	if err != nil {
 		return "", err
 	}
 
 	// get the pubkey, but remove the type
-	publicKey, err := GetPublicKey()
-	if err != nil {
-		return "", err
-	}
-
-	publicKeyBytes := ssh.MarshalAuthorizedKey(publicKey)
-
-	parts := bytes.Split(publicKeyBytes, []byte(" "))
+	publicKey := ssh.MarshalAuthorizedKey(privateKey.PublicKey())
+	parts := bytes.Split(publicKey, []byte(" "))
 	if len(parts) < 2 {
 		return "", fmt.Errorf("malformed public key")
 	}
@@ -103,49 +89,3 @@ func Sign(ctx context.Context, bts []byte) (string, error) {
 	// signature is <pubkey>:<signature>
 	return fmt.Sprintf("%s:%s", bytes.TrimSpace(parts[1]), base64.StdEncoding.EncodeToString(signedData.Blob)), nil
 }
-
-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 errors.Is(err, os.ErrNotExist) {
-		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
-}

cmd/cmd.go

@@ -4,7 +4,10 @@ import (
 	"archive/zip"
 	"bytes"
 	"context"
+	"crypto/ed25519"
+	"crypto/rand"
 	"crypto/sha256"
+	"encoding/pem"
 	"errors"
 	"fmt"
 	"io"
@@ -376,12 +379,11 @@ func errFromUnknownKey(unknownKeyErr error) error {
 	if len(matches) > 0 {
 		serverPubKey := matches[0]
 
-		publicKey, err := auth.GetPublicKey()
+		localPubKey, err := auth.GetPublicKey()
 		if err != nil {
 			return unknownKeyErr
 		}
 
-		localPubKey := strings.TrimSpace(string(ssh.MarshalAuthorizedKey(publicKey)))
 		if runtime.GOOS == "linux" && serverPubKey != localPubKey {
 			// try the ollama service public key
 			svcPubKey, err := os.ReadFile("/usr/share/ollama/.ollama/id_ed25519.pub")
@@ -1070,7 +1072,7 @@ func generate(cmd *cobra.Command, opts runOptions) error {
 }
 
 func RunServer(cmd *cobra.Command, _ []string) error {
-	if _, err := auth.GetPublicKey(); err != nil {
+	if err := initializeKeypair(); err != nil {
 		return err
 	}
 
@@ -1087,6 +1089,52 @@ func RunServer(cmd *cobra.Command, _ []string) error {
 	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 {

convert/mistral.go

@@ -71,6 +71,11 @@ func (m *MistralModel) WriteGGUF(ws io.WriteSeeker) error {
 		"tokenizer.ggml.unknown_token_id": uint32(0),
 	}
 
+	if m.Params.HeadDimension > 0 {
+		kv["llama.attention.key_length"] = uint32(m.Params.HeadDimension)
+		kv["llama.attention.value_length"] = uint32(m.Params.HeadDimension)
+	}
+
 	return llm.NewGGUFV3(m.Params.ByteOrder).Encode(ws, kv, m.Tensors)
 }
 

docs/api.md

@@ -1026,7 +1026,7 @@ If `stream` is set to `false`, then the response is a single JSON object:
 ## Generate Embeddings
 
 ```shell
-POST /api/embeddings
+POST /api/embed
 ```
 
 Generate embeddings from a model
@@ -1034,10 +1034,11 @@ Generate embeddings from a model
 ### Parameters
 
 - `model`: name of model to generate embeddings from
-- `prompt`: text to generate embeddings for
+- `input`: text or list of text to generate embeddings for
 
 Advanced parameters:
 
+- `truncate`: truncates the end of each input to fit within context length. Returns error if `false` and context length is exceeded. Defaults to `true`
 - `options`: additional model parameters listed in the documentation for the [Modelfile](./modelfile.md#valid-parameters-and-values) such as `temperature`
 - `keep_alive`: controls how long the model will stay loaded into memory following the request (default: `5m`)
 
@@ -1046,9 +1047,9 @@ Advanced parameters:
 #### Request
 
 ```shell
-curl http://localhost:11434/api/embeddings -d '{
+curl http://localhost:11434/api/embed -d '{
   "model": "all-minilm",
-  "prompt": "Here is an article about llamas..."
+  "input": "Why is the sky blue?"
 }'
 ```
 
@@ -1056,10 +1057,35 @@ curl http://localhost:11434/api/embeddings -d '{
 
 ```json
 {
-  "embedding": [
-    0.5670403838157654, 0.009260174818336964, 0.23178744316101074, -0.2916173040866852, -0.8924556970596313,
-    0.8785552978515625, -0.34576427936553955, 0.5742510557174683, -0.04222835972905159, -0.137906014919281
-  ]
+  "model": "all-minilm",
+  "embeddings": [[
+    0.010071029, -0.0017594862, 0.05007221, 0.04692972, 0.054916814,
+    0.008599704, 0.105441414, -0.025878139, 0.12958129, 0.031952348
+  ]]
+}
+```
+
+#### Request (Multiple input)
+
+```shell
+curl http://localhost:11434/api/embed -d '{
+  "model": "all-minilm",
+  "input": ["Why is the sky blue?", "Why is the grass green?"]
+}'
+```
+
+#### Response
+
+```json
+{
+  "model": "all-minilm",
+  "embeddings": [[
+    0.010071029, -0.0017594862, 0.05007221, 0.04692972, 0.054916814,
+    0.008599704, 0.105441414, -0.025878139, 0.12958129, 0.031952348
+  ],[
+    -0.0098027075, 0.06042469, 0.025257962, -0.006364387, 0.07272725,
+    0.017194884, 0.09032035, -0.051705178, 0.09951512, 0.09072481
+  ]]
 }
 ```
 
@@ -1106,3 +1132,45 @@ A single JSON object will be returned.
   ]
 }
 ```
+
+## Generate Embedding
+
+> Note: this endpoint has been superseded by `/api/embed`
+
+```shell
+POST /api/embeddings
+```
+
+Generate embeddings from a model
+
+### Parameters
+
+- `model`: name of model to generate embeddings from
+- `prompt`: text to generate embeddings for
+
+Advanced parameters:
+
+- `options`: additional model parameters listed in the documentation for the [Modelfile](./modelfile.md#valid-parameters-and-values) such as `temperature`
+- `keep_alive`: controls how long the model will stay loaded into memory following the request (default: `5m`)
+
+### Examples
+
+#### Request
+
+```shell
+curl http://localhost:11434/api/embeddings -d '{
+  "model": "all-minilm",
+  "prompt": "Here is an article about llamas..."
+}'
+```
+
+#### Response
+
+```json
+{
+  "embedding": [
+    0.5670403838157654, 0.009260174818336964, 0.23178744316101074, -0.2916173040866852, -0.8924556970596313,
+    0.8785552978515625, -0.34576427936553955, 0.5742510557174683, -0.04222835972905159, -0.137906014919281
+  ]
+}
+```

server/images.go

@@ -32,7 +32,6 @@ import (
 	"github.com/ollama/ollama/types/errtypes"
 	"github.com/ollama/ollama/types/model"
 	"github.com/ollama/ollama/version"
-	"golang.org/x/crypto/ssh"
 )
 
 var (
@@ -1089,12 +1088,11 @@ func makeRequestWithRetry(ctx context.Context, method string, requestURL *url.UR
 	if anonymous {
 		// no user is associated with the public key, and the request requires non-anonymous access
 		pubKey, nestedErr := auth.GetPublicKey()
-		localPubKey := strings.TrimSpace(string(ssh.MarshalAuthorizedKey(pubKey)))
 		if nestedErr != nil {
 			slog.Error(fmt.Sprintf("couldn't get public key: %v", nestedErr))
 			return nil, errUnauthorized
 		}
-		return nil, &errtypes.UnknownOllamaKey{Key: localPubKey}
+		return nil, &errtypes.UnknownOllamaKey{Key: pubKey}
 	}
 	// user is associated with the public key, but is not authorized to make the request
 	return nil, errUnauthorized