embeddings: added embedding command for cl (#12795)

Co-authored-by: A-Akhil <akhilrahul70@gmail.com>

This PR introduces a new ollama embed command that allows users to generate embeddings directly from the command line.

Added ollama embed MODEL [TEXT...] command for generating text embeddings
Supports both direct text arguments and stdin piping for scripted workflows

Outputs embeddings as JSON arrays (one per line)

cmd/cmd.go

@@ -322,6 +322,44 @@ func StopHandler(cmd *cobra.Command, args []string) error {
 	return nil
 }
 
+func generateEmbedding(cmd *cobra.Command, modelName, input string, keepAlive *api.Duration, truncate *bool, dimensions int) error {
+	client, err := api.ClientFromEnvironment()
+	if err != nil {
+		return err
+	}
+
+	req := &api.EmbedRequest{
+		Model: modelName,
+		Input: input,
+	}
+	if keepAlive != nil {
+		req.KeepAlive = keepAlive
+	}
+	if truncate != nil {
+		req.Truncate = truncate
+	}
+	if dimensions > 0 {
+		req.Dimensions = dimensions
+	}
+
+	resp, err := client.Embed(cmd.Context(), req)
+	if err != nil {
+		return err
+	}
+
+	if len(resp.Embeddings) == 0 {
+		return errors.New("no embeddings returned")
+	}
+
+	output, err := json.Marshal(resp.Embeddings[0])
+	if err != nil {
+		return err
+	}
+	fmt.Println(string(output))
+
+	return nil
+}
+
 func RunHandler(cmd *cobra.Command, args []string) error {
 	interactive := true
 
@@ -386,7 +424,11 @@ func RunHandler(cmd *cobra.Command, args []string) error {
 			return err
 		}
 
-		prompts = append([]string{string(in)}, prompts...)
+		// Only prepend stdin content if it's not empty
+		stdinContent := string(in)
+		if len(stdinContent) > 0 {
+			prompts = append([]string{stdinContent}, prompts...)
+		}
 		opts.ShowConnect = false
 		opts.WordWrap = false
 		interactive = false
@@ -452,6 +494,29 @@ func RunHandler(cmd *cobra.Command, args []string) error {
 
 	opts.ParentModel = info.Details.ParentModel
 
+	// Check if this is an embedding model
+	isEmbeddingModel := slices.Contains(info.Capabilities, model.CapabilityEmbedding)
+
+	// If it's an embedding model, handle embedding generation
+	if isEmbeddingModel {
+		if opts.Prompt == "" {
+			return errors.New("embedding models require input text. Usage: ollama run " + name + " \"your text here\"")
+		}
+
+		// Get embedding-specific flags
+		var truncate *bool
+		if truncateFlag, err := cmd.Flags().GetBool("truncate"); err == nil && cmd.Flags().Changed("truncate") {
+			truncate = &truncateFlag
+		}
+
+		dimensions, err := cmd.Flags().GetInt("dimensions")
+		if err != nil {
+			return err
+		}
+
+		return generateEmbedding(cmd, name, opts.Prompt, opts.KeepAlive, truncate, dimensions)
+	}
+
 	if interactive {
 		if err := loadOrUnloadModel(cmd, &opts); err != nil {
 			var sErr api.AuthorizationError
@@ -1684,6 +1749,8 @@ func NewCLI() *cobra.Command {
 	runCmd.Flags().String("think", "", "Enable thinking mode: true/false or high/medium/low for supported models")
 	runCmd.Flags().Lookup("think").NoOptDefVal = "true"
 	runCmd.Flags().Bool("hidethinking", false, "Hide thinking output (if provided)")
+	runCmd.Flags().Bool("truncate", false, "For embedding models: truncate inputs exceeding context length (default: true). Set --truncate=false to error instead")
+	runCmd.Flags().Int("dimensions", 0, "Truncate output embeddings to specified dimension (embedding models only)")
 
 	stopCmd := &cobra.Command{
 		Use:     "stop MODEL",