fix: Initialize sudo reference before its first actual use (#7360)

Double sudo invocation fix in the install script

.air.toml

@@ -0,0 +1,8 @@
+# .air.toml
+[build]
+cmd = "make build"
+bin = "./local-ai"
+args_bin = [ "--debug" ]
+include_ext = ["go", "html", "yaml", "toml", "json", "txt", "md"]
+exclude_dir = ["pkg/grpc/proto"]
+delay = 1000

.github/gallery-agent/agent.go

@@ -0,0 +1,288 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"slices"
+	"strings"
+
+	hfapi "github.com/mudler/LocalAI/pkg/huggingface-api"
+	cogito "github.com/mudler/cogito"
+
+	"github.com/mudler/cogito/structures"
+	"github.com/sashabaranov/go-openai/jsonschema"
+)
+
+var (
+	openAIModel      = os.Getenv("OPENAI_MODEL")
+	openAIKey        = os.Getenv("OPENAI_KEY")
+	openAIBaseURL    = os.Getenv("OPENAI_BASE_URL")
+	galleryIndexPath = os.Getenv("GALLERY_INDEX_PATH")
+	//defaultclient
+	llm = cogito.NewOpenAILLM(openAIModel, openAIKey, openAIBaseURL)
+)
+
+// cleanTextContent removes trailing spaces, tabs, and normalizes line endings
+// to prevent YAML linting issues like trailing spaces and multiple empty lines
+func cleanTextContent(text string) string {
+	lines := strings.Split(text, "\n")
+	var cleanedLines []string
+	var prevEmpty bool
+	for _, line := range lines {
+		// Remove all trailing whitespace (spaces, tabs, etc.)
+		trimmed := strings.TrimRight(line, " \t\r")
+		// Avoid multiple consecutive empty lines
+		if trimmed == "" {
+			if !prevEmpty {
+				cleanedLines = append(cleanedLines, "")
+			}
+			prevEmpty = true
+		} else {
+			cleanedLines = append(cleanedLines, trimmed)
+			prevEmpty = false
+		}
+	}
+	// Remove trailing empty lines from the result
+	result := strings.Join(cleanedLines, "\n")
+	return strings.TrimRight(result, "\n")
+}
+
+// isModelExisting checks if a specific model ID exists in the gallery using text search
+func isModelExisting(modelID string) (bool, error) {
+	indexPath := getGalleryIndexPath()
+	content, err := os.ReadFile(indexPath)
+	if err != nil {
+		return false, fmt.Errorf("failed to read %s: %w", indexPath, err)
+	}
+
+	contentStr := string(content)
+	// Simple text search - if the model ID appears anywhere in the file, it exists
+	return strings.Contains(contentStr, modelID), nil
+}
+
+// filterExistingModels removes models that already exist in the gallery
+func filterExistingModels(models []ProcessedModel) ([]ProcessedModel, error) {
+	var filteredModels []ProcessedModel
+	for _, model := range models {
+		exists, err := isModelExisting(model.ModelID)
+		if err != nil {
+			fmt.Printf("Error checking if model %s exists: %v, skipping\n", model.ModelID, err)
+			continue
+		}
+
+		if !exists {
+			filteredModels = append(filteredModels, model)
+		} else {
+			fmt.Printf("Skipping existing model: %s\n", model.ModelID)
+		}
+	}
+
+	fmt.Printf("Filtered out %d existing models, %d new models remaining\n",
+		len(models)-len(filteredModels), len(filteredModels))
+
+	return filteredModels, nil
+}
+
+// getGalleryIndexPath returns the gallery index file path, with a default fallback
+func getGalleryIndexPath() string {
+	if galleryIndexPath != "" {
+		return galleryIndexPath
+	}
+	return "gallery/index.yaml"
+}
+
+func getRealReadme(ctx context.Context, repository string) (string, error) {
+	// Create a conversation fragment
+	fragment := cogito.NewEmptyFragment().
+		AddMessage("user",
+			`Your task is to get a clear description of a large language model from huggingface by using the provided tool. I will share with you a repository that might be quantized, and as such probably not by the original model author. We need to get the real  description of the model, and not the one that might be quantized. You will have to call the tool to get the readme more than once by figuring out from the quantized readme which is the base model readme. This is the repository: `+repository)
+
+	// Execute with tools
+	result, err := cogito.ExecuteTools(llm, fragment,
+		cogito.WithIterations(3),
+		cogito.WithMaxAttempts(3),
+		cogito.WithTools(&HFReadmeTool{client: hfapi.NewClient()}))
+	if err != nil {
+		return "", err
+	}
+
+	result = result.AddMessage("user", "Describe the model in a clear and concise way that can be shared in a model gallery.")
+
+	// Get a response
+	newFragment, err := llm.Ask(ctx, result)
+	if err != nil {
+		return "", err
+	}
+
+	content := newFragment.LastMessage().Content
+	return cleanTextContent(content), nil
+}
+
+func selectMostInterestingModels(ctx context.Context, searchResult *SearchResult) ([]ProcessedModel, error) {
+	// Create a conversation fragment
+	fragment := cogito.NewEmptyFragment().
+		AddMessage("user",
+			`Your task is to analyze a list of AI models and select the most interesting ones for a model gallery. You will be given detailed information about multiple models including their metadata, file information, and README content.
+
+Consider the following criteria when selecting models:
+1. Model popularity (download count)
+2. Model recency (last modified date)
+3. Model completeness (has preferred model file, README, etc.)
+4. Model uniqueness (not duplicates or very similar models)
+5. Model quality (based on README content and description)
+6. Model utility (practical applications)
+
+You should select models that would be most valuable for users browsing a model gallery. Prioritize models that are:
+- Well-documented with clear READMEs
+- Recently updated
+- Popular (high download count)
+- Have the preferred quantization format available
+- Offer unique capabilities or are from reputable authors
+
+Return your analysis and selection reasoning.`)
+
+	// Add the search results as context
+	modelsInfo := fmt.Sprintf("Found %d models matching '%s' with quantization preference '%s':\n\n",
+		searchResult.TotalModelsFound, searchResult.SearchTerm, searchResult.Quantization)
+
+	for i, model := range searchResult.Models {
+		modelsInfo += fmt.Sprintf("Model %d:\n", i+1)
+		modelsInfo += fmt.Sprintf("  ID: %s\n", model.ModelID)
+		modelsInfo += fmt.Sprintf("  Author: %s\n", model.Author)
+		modelsInfo += fmt.Sprintf("  Downloads: %d\n", model.Downloads)
+		modelsInfo += fmt.Sprintf("  Last Modified: %s\n", model.LastModified)
+		modelsInfo += fmt.Sprintf("  Files: %d files\n", len(model.Files))
+
+		if model.PreferredModelFile != nil {
+			modelsInfo += fmt.Sprintf("  Preferred Model File: %s (%d bytes)\n",
+				model.PreferredModelFile.Path, model.PreferredModelFile.Size)
+		} else {
+			modelsInfo += "  No preferred model file found\n"
+		}
+
+		if model.ReadmeContent != "" {
+			modelsInfo += fmt.Sprintf("  README: %s\n", model.ReadmeContent)
+		}
+
+		if model.ProcessingError != "" {
+			modelsInfo += fmt.Sprintf("  Processing Error: %s\n", model.ProcessingError)
+		}
+
+		modelsInfo += "\n"
+	}
+
+	fragment = fragment.AddMessage("user", modelsInfo)
+
+	fragment = fragment.AddMessage("user", "Based on your analysis, select the top 5 most interesting models and provide a brief explanation for each selection. Also, create a filtered SearchResult with only the selected models. Return just a list of repositories IDs, you will later be asked to output it as a JSON array with the json tool.")
+
+	// Get a response
+	newFragment, err := llm.Ask(ctx, fragment)
+	if err != nil {
+		return nil, err
+	}
+
+	fmt.Println(newFragment.LastMessage().Content)
+	repositories := struct {
+		Repositories []string `json:"repositories"`
+	}{}
+
+	s := structures.Structure{
+		Schema: jsonschema.Definition{
+			Type:                 jsonschema.Object,
+			AdditionalProperties: false,
+			Properties: map[string]jsonschema.Definition{
+				"repositories": {
+					Type:        jsonschema.Array,
+					Items:       &jsonschema.Definition{Type: jsonschema.String},
+					Description: "The trending repositories IDs",
+				},
+			},
+			Required: []string{"repositories"},
+		},
+		Object: &repositories,
+	}
+
+	err = newFragment.ExtractStructure(ctx, llm, s)
+	if err != nil {
+		return nil, err
+	}
+
+	filteredModels := []ProcessedModel{}
+	for _, m := range searchResult.Models {
+		if slices.Contains(repositories.Repositories, m.ModelID) {
+			filteredModels = append(filteredModels, m)
+		}
+	}
+
+	return filteredModels, nil
+}
+
+// ModelFamily represents a YAML anchor/family
+type ModelFamily struct {
+	Anchor string `json:"anchor"`
+	Name   string `json:"name"`
+}
+
+// selectModelFamily selects the appropriate model family/anchor for a given model
+func selectModelFamily(ctx context.Context, model ProcessedModel, availableFamilies []ModelFamily) (string, error) {
+	// Create a conversation fragment
+	fragment := cogito.NewEmptyFragment().
+		AddMessage("user",
+			`Your task is to select the most appropriate model family/anchor for a given AI model. You will be provided with:
+1. Information about the model (name, description, etc.)
+2. A list of available model families/anchors
+
+You need to select the family that best matches the model's architecture, capabilities, or characteristics. Consider:
+- Model architecture (e.g., Llama, Qwen, Mistral, etc.)
+- Model capabilities (e.g., vision, coding, chat, etc.)
+- Model size/type (e.g., small, medium, large)
+- Model purpose (e.g., general purpose, specialized, etc.)
+
+Return the anchor name that best fits the model.`)
+
+	// Add model information
+	modelInfo := "Model Information:\n"
+	modelInfo += fmt.Sprintf("  ID: %s\n", model.ModelID)
+	modelInfo += fmt.Sprintf("  Author: %s\n", model.Author)
+	modelInfo += fmt.Sprintf("  Downloads: %d\n", model.Downloads)
+	modelInfo += fmt.Sprintf("  Description: %s\n", model.ReadmeContentPreview)
+
+	fragment = fragment.AddMessage("user", modelInfo)
+
+	// Add available families
+	familiesInfo := "Available Model Families:\n"
+	for _, family := range availableFamilies {
+		familiesInfo += fmt.Sprintf("  - %s (%s)\n", family.Anchor, family.Name)
+	}
+
+	fragment = fragment.AddMessage("user", familiesInfo)
+	fragment = fragment.AddMessage("user", "Select the most appropriate family anchor for this model. Return just the anchor name.")
+
+	// Get a response
+	newFragment, err := llm.Ask(ctx, fragment)
+	if err != nil {
+		return "", err
+	}
+
+	// Extract the selected family
+	selectedFamily := strings.TrimSpace(newFragment.LastMessage().Content)
+
+	// Validate that the selected family exists in our list
+	for _, family := range availableFamilies {
+		if family.Anchor == selectedFamily {
+			return selectedFamily, nil
+		}
+	}
+
+	// If no exact match, try to find a close match
+	for _, family := range availableFamilies {
+		if strings.Contains(strings.ToLower(family.Anchor), strings.ToLower(selectedFamily)) ||
+			strings.Contains(strings.ToLower(selectedFamily), strings.ToLower(family.Anchor)) {
+			return family.Anchor, nil
+		}
+	}
+
+	// Default fallback
+	return "llama3", nil
+}

.github/gallery-agent/gallery.go

@@ -0,0 +1,203 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"strings"
+)
+
+// generateYAMLEntry generates a YAML entry for a model using the specified anchor
+func generateYAMLEntry(model ProcessedModel, familyAnchor string) string {
+	// Extract model name from ModelID
+	parts := strings.Split(model.ModelID, "/")
+	modelName := model.ModelID
+	if len(parts) > 0 {
+		modelName = strings.ToLower(parts[len(parts)-1])
+	}
+	// Remove common suffixes
+	modelName = strings.ReplaceAll(modelName, "-gguf", "")
+	modelName = strings.ReplaceAll(modelName, "-q4_k_m", "")
+	modelName = strings.ReplaceAll(modelName, "-q4_k_s", "")
+	modelName = strings.ReplaceAll(modelName, "-q3_k_m", "")
+	modelName = strings.ReplaceAll(modelName, "-q2_k", "")
+
+	fileName := ""
+	checksum := ""
+	if model.PreferredModelFile != nil {
+		fileParts := strings.Split(model.PreferredModelFile.Path, "/")
+		if len(fileParts) > 0 {
+			fileName = fileParts[len(fileParts)-1]
+		}
+		checksum = model.PreferredModelFile.SHA256
+	} else {
+		fileName = model.ModelID
+	}
+
+	description := model.ReadmeContent
+	if description == "" {
+		description = fmt.Sprintf("AI model: %s", modelName)
+	}
+
+	// Clean up description to prevent YAML linting issues
+	description = cleanTextContent(description)
+
+	// Format description for YAML (indent each line and ensure no trailing spaces)
+	lines := strings.Split(description, "\n")
+	var formattedLines []string
+	for _, line := range lines {
+		if strings.TrimSpace(line) == "" {
+			// Keep empty lines as empty (no indentation)
+			formattedLines = append(formattedLines, "")
+		} else {
+			// Add indentation to non-empty lines
+			formattedLines = append(formattedLines, "    "+line)
+		}
+	}
+	formattedDescription := strings.Join(formattedLines, "\n")
+	// Remove any trailing spaces from the formatted description
+	formattedDescription = strings.TrimRight(formattedDescription, " \t")
+	yamlTemplate := ""
+	if checksum != "" {
+		yamlTemplate = `- !!merge <<: *%s
+  name: "%s"
+  urls:
+    - https://huggingface.co/%s
+  description: |
+%s
+  overrides:
+    parameters:
+      model: %s
+  files:
+    - filename: %s
+      sha256: %s
+      uri: huggingface://%s/%s`
+		return fmt.Sprintf(yamlTemplate,
+			familyAnchor,
+			modelName,
+			model.ModelID,
+			formattedDescription,
+			fileName,
+			fileName,
+			checksum,
+			model.ModelID,
+			fileName,
+		)
+	} else {
+		yamlTemplate = `- !!merge <<: *%s
+  name: "%s"
+  urls:
+    - https://huggingface.co/%s
+  description: |
+%s
+  overrides:
+    parameters:
+      model: %s`
+		return fmt.Sprintf(yamlTemplate,
+			familyAnchor,
+			modelName,
+			model.ModelID,
+			formattedDescription,
+			fileName,
+		)
+	}
+}
+
+// extractModelFamilies extracts all YAML anchors from the gallery index.yaml file
+func extractModelFamilies() ([]ModelFamily, error) {
+	// Read the index.yaml file
+	indexPath := getGalleryIndexPath()
+	content, err := os.ReadFile(indexPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read %s: %w", indexPath, err)
+	}
+
+	lines := strings.Split(string(content), "\n")
+	var families []ModelFamily
+
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+		// Look for YAML anchors (lines starting with "- &")
+		if strings.HasPrefix(line, "- &") {
+			// Extract the anchor name (everything after "- &")
+			anchor := strings.TrimPrefix(line, "- &")
+			// Remove any trailing colon or other characters
+			anchor = strings.Split(anchor, ":")[0]
+			anchor = strings.Split(anchor, " ")[0]
+
+			if anchor != "" {
+				families = append(families, ModelFamily{
+					Anchor: anchor,
+					Name:   anchor, // Use anchor as name for now
+				})
+			}
+		}
+	}
+
+	return families, nil
+}
+
+// generateYAMLForModels generates YAML entries for selected models and appends to index.yaml
+func generateYAMLForModels(ctx context.Context, models []ProcessedModel) error {
+	// Extract available model families
+	families, err := extractModelFamilies()
+	if err != nil {
+		return fmt.Errorf("failed to extract model families: %w", err)
+	}
+
+	fmt.Printf("Found %d model families: %v\n", len(families),
+		func() []string {
+			var names []string
+			for _, f := range families {
+				names = append(names, f.Anchor)
+			}
+			return names
+		}())
+
+	// Generate YAML entries for each model
+	var yamlEntries []string
+	for _, model := range models {
+		fmt.Printf("Selecting family for model: %s\n", model.ModelID)
+
+		// Select appropriate family for this model
+		familyAnchor, err := selectModelFamily(ctx, model, families)
+		if err != nil {
+			fmt.Printf("Error selecting family for %s: %v, using default\n", model.ModelID, err)
+			familyAnchor = "llama3" // Default fallback
+		}
+
+		fmt.Printf("Selected family '%s' for model %s\n", familyAnchor, model.ModelID)
+
+		// Generate YAML entry
+		yamlEntry := generateYAMLEntry(model, familyAnchor)
+		yamlEntries = append(yamlEntries, yamlEntry)
+	}
+
+	// Append to index.yaml
+	if len(yamlEntries) > 0 {
+		indexPath := getGalleryIndexPath()
+		fmt.Printf("Appending YAML entries to %s...\n", indexPath)
+
+		// Read current content
+		content, err := os.ReadFile(indexPath)
+		if err != nil {
+			return fmt.Errorf("failed to read %s: %w", indexPath, err)
+		}
+
+		// Append new entries
+		// Remove trailing whitespace from existing content and join entries without extra newlines
+		existingContent := strings.TrimRight(string(content), " \t\n\r")
+		yamlBlock := strings.Join(yamlEntries, "\n")
+		newContent := existingContent + "\n" + yamlBlock + "\n"
+
+		// Write back to file
+		err = os.WriteFile(indexPath, []byte(newContent), 0644)
+		if err != nil {
+			return fmt.Errorf("failed to write %s: %w", indexPath, err)
+		}
+
+		fmt.Printf("Successfully added %d models to %s\n", len(yamlEntries), indexPath)
+	}
+
+	return nil
+}

.github/gallery-agent/main.go

@@ -0,0 +1,351 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+	"time"
+
+	hfapi "github.com/mudler/LocalAI/pkg/huggingface-api"
+)
+
+// ProcessedModelFile represents a processed model file with additional metadata
+type ProcessedModelFile struct {
+	Path     string `json:"path"`
+	Size     int64  `json:"size"`
+	SHA256   string `json:"sha256"`
+	IsReadme bool   `json:"is_readme"`
+	FileType string `json:"file_type"` // "model", "readme", "other"
+}
+
+// ProcessedModel represents a processed model with all gathered metadata
+type ProcessedModel struct {
+	ModelID                 string               `json:"model_id"`
+	Author                  string               `json:"author"`
+	Downloads               int                  `json:"downloads"`
+	LastModified            string               `json:"last_modified"`
+	Files                   []ProcessedModelFile `json:"files"`
+	PreferredModelFile      *ProcessedModelFile  `json:"preferred_model_file,omitempty"`
+	ReadmeFile              *ProcessedModelFile  `json:"readme_file,omitempty"`
+	ReadmeContent           string               `json:"readme_content,omitempty"`
+	ReadmeContentPreview    string               `json:"readme_content_preview,omitempty"`
+	QuantizationPreferences []string             `json:"quantization_preferences"`
+	ProcessingError         string               `json:"processing_error,omitempty"`
+}
+
+// SearchResult represents the complete result of searching and processing models
+type SearchResult struct {
+	SearchTerm       string           `json:"search_term"`
+	Limit            int              `json:"limit"`
+	Quantization     string           `json:"quantization"`
+	TotalModelsFound int              `json:"total_models_found"`
+	Models           []ProcessedModel `json:"models"`
+	FormattedOutput  string           `json:"formatted_output"`
+}
+
+// AddedModelSummary represents a summary of models added to the gallery
+type AddedModelSummary struct {
+	SearchTerm     string   `json:"search_term"`
+	TotalFound     int      `json:"total_found"`
+	ModelsAdded    int      `json:"models_added"`
+	AddedModelIDs  []string `json:"added_model_ids"`
+	AddedModelURLs []string `json:"added_model_urls"`
+	Quantization   string   `json:"quantization"`
+	ProcessingTime string   `json:"processing_time"`
+}
+
+func main() {
+	startTime := time.Now()
+
+	// Check for synthetic mode
+	syntheticMode := os.Getenv("SYNTHETIC_MODE")
+	if syntheticMode == "true" || syntheticMode == "1" {
+		fmt.Println("Running in SYNTHETIC MODE - generating random test data")
+		err := runSyntheticMode()
+		if err != nil {
+			fmt.Fprintf(os.Stderr, "Error in synthetic mode: %v\n", err)
+			os.Exit(1)
+		}
+		return
+	}
+
+	// Get configuration from environment variables
+	searchTerm := os.Getenv("SEARCH_TERM")
+	if searchTerm == "" {
+		searchTerm = "GGUF"
+	}
+
+	limitStr := os.Getenv("LIMIT")
+	if limitStr == "" {
+		limitStr = "5"
+	}
+	limit, err := strconv.Atoi(limitStr)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error parsing LIMIT: %v\n", err)
+		os.Exit(1)
+	}
+
+	quantization := os.Getenv("QUANTIZATION")
+
+	maxModels := os.Getenv("MAX_MODELS")
+	if maxModels == "" {
+		maxModels = "1"
+	}
+	maxModelsInt, err := strconv.Atoi(maxModels)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error parsing MAX_MODELS: %v\n", err)
+		os.Exit(1)
+	}
+
+	// Print configuration
+	fmt.Printf("Gallery Agent Configuration:\n")
+	fmt.Printf("  Search Term: %s\n", searchTerm)
+	fmt.Printf("  Limit: %d\n", limit)
+	fmt.Printf("  Quantization: %s\n", quantization)
+	fmt.Printf("  Max Models to Add: %d\n", maxModelsInt)
+	fmt.Printf("  Gallery Index Path: %s\n", os.Getenv("GALLERY_INDEX_PATH"))
+	fmt.Println()
+
+	result, err := searchAndProcessModels(searchTerm, limit, quantization)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error: %v\n", err)
+		os.Exit(1)
+	}
+
+	fmt.Println(result.FormattedOutput)
+
+	// Use AI agent to select the most interesting models
+	fmt.Println("Using AI agent to select the most interesting models...")
+	models, err := selectMostInterestingModels(context.Background(), result)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error in model selection: %v\n", err)
+		// Continue with original result if selection fails
+		models = result.Models
+	}
+
+	fmt.Print(models)
+
+	// Filter out models that already exist in the gallery
+	fmt.Println("Filtering out existing models...")
+	models, err = filterExistingModels(models)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error filtering existing models: %v\n", err)
+		os.Exit(1)
+	}
+
+	// Limit to maxModelsInt after filtering
+	if len(models) > maxModelsInt {
+		models = models[:maxModelsInt]
+	}
+
+	// Track added models for summary
+	var addedModelIDs []string
+	var addedModelURLs []string
+
+	// Generate YAML entries and append to gallery/index.yaml
+	if len(models) > 0 {
+		for _, model := range models {
+			addedModelIDs = append(addedModelIDs, model.ModelID)
+			// Generate Hugging Face URL for the model
+			modelURL := fmt.Sprintf("https://huggingface.co/%s", model.ModelID)
+			addedModelURLs = append(addedModelURLs, modelURL)
+		}
+		fmt.Println("Generating YAML entries for selected models...")
+		err = generateYAMLForModels(context.Background(), models)
+		if err != nil {
+			fmt.Fprintf(os.Stderr, "Error generating YAML entries: %v\n", err)
+			os.Exit(1)
+		}
+	} else {
+		fmt.Println("No new models to add to the gallery.")
+	}
+
+	// Create and write summary
+	processingTime := time.Since(startTime).String()
+	summary := AddedModelSummary{
+		SearchTerm:     searchTerm,
+		TotalFound:     result.TotalModelsFound,
+		ModelsAdded:    len(addedModelIDs),
+		AddedModelIDs:  addedModelIDs,
+		AddedModelURLs: addedModelURLs,
+		Quantization:   quantization,
+		ProcessingTime: processingTime,
+	}
+
+	// Write summary to file
+	summaryData, err := json.MarshalIndent(summary, "", "  ")
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error marshaling summary: %v\n", err)
+	} else {
+		err = os.WriteFile("gallery-agent-summary.json", summaryData, 0644)
+		if err != nil {
+			fmt.Fprintf(os.Stderr, "Error writing summary file: %v\n", err)
+		} else {
+			fmt.Printf("Summary written to gallery-agent-summary.json\n")
+		}
+	}
+}
+
+func searchAndProcessModels(searchTerm string, limit int, quantization string) (*SearchResult, error) {
+	client := hfapi.NewClient()
+	var outputBuilder strings.Builder
+
+	fmt.Println("Searching for models...")
+	// Initialize the result struct
+	result := &SearchResult{
+		SearchTerm:   searchTerm,
+		Limit:        limit,
+		Quantization: quantization,
+		Models:       []ProcessedModel{},
+	}
+
+	models, err := client.GetLatest(searchTerm, limit)
+	if err != nil {
+		return nil, fmt.Errorf("failed to fetch models: %w", err)
+	}
+
+	fmt.Println("Models found:", len(models))
+	result.TotalModelsFound = len(models)
+
+	if len(models) == 0 {
+		outputBuilder.WriteString("No models found.\n")
+		result.FormattedOutput = outputBuilder.String()
+		return result, nil
+	}
+
+	outputBuilder.WriteString(fmt.Sprintf("Found %d models matching '%s':\n\n", len(models), searchTerm))
+
+	// Process each model
+	for i, model := range models {
+		outputBuilder.WriteString(fmt.Sprintf("%d. Processing Model: %s\n", i+1, model.ModelID))
+		outputBuilder.WriteString(fmt.Sprintf("   Author: %s\n", model.Author))
+		outputBuilder.WriteString(fmt.Sprintf("   Downloads: %d\n", model.Downloads))
+		outputBuilder.WriteString(fmt.Sprintf("   Last Modified: %s\n", model.LastModified))
+
+		// Initialize processed model struct
+		processedModel := ProcessedModel{
+			ModelID:                 model.ModelID,
+			Author:                  model.Author,
+			Downloads:               model.Downloads,
+			LastModified:            model.LastModified,
+			QuantizationPreferences: []string{quantization, "Q4_K_M", "Q4_K_S", "Q3_K_M", "Q2_K"},
+		}
+
+		// Get detailed model information
+		details, err := client.GetModelDetails(model.ModelID)
+		if err != nil {
+			errorMsg := fmt.Sprintf("   Error getting model details: %v\n", err)
+			outputBuilder.WriteString(errorMsg)
+			processedModel.ProcessingError = err.Error()
+			result.Models = append(result.Models, processedModel)
+			continue
+		}
+
+		// Define quantization preferences (in order of preference)
+		quantizationPreferences := []string{quantization, "Q4_K_M", "Q4_K_S", "Q3_K_M", "Q2_K"}
+
+		// Find preferred model file
+		preferredModelFile := hfapi.FindPreferredModelFile(details.Files, quantizationPreferences)
+
+		// Process files
+		processedFiles := make([]ProcessedModelFile, len(details.Files))
+		for j, file := range details.Files {
+			fileType := "other"
+			if file.IsReadme {
+				fileType = "readme"
+			} else if preferredModelFile != nil && file.Path == preferredModelFile.Path {
+				fileType = "model"
+			}
+
+			processedFiles[j] = ProcessedModelFile{
+				Path:     file.Path,
+				Size:     file.Size,
+				SHA256:   file.SHA256,
+				IsReadme: file.IsReadme,
+				FileType: fileType,
+			}
+		}
+
+		processedModel.Files = processedFiles
+
+		// Set preferred model file
+		if preferredModelFile != nil {
+			for _, file := range processedFiles {
+				if file.Path == preferredModelFile.Path {
+					processedModel.PreferredModelFile = &file
+					break
+				}
+			}
+		}
+
+		// Print file information
+		outputBuilder.WriteString(fmt.Sprintf("   Files found: %d\n", len(details.Files)))
+
+		if preferredModelFile != nil {
+			outputBuilder.WriteString(fmt.Sprintf("   Preferred Model File: %s (SHA256: %s)\n",
+				preferredModelFile.Path,
+				preferredModelFile.SHA256))
+		} else {
+			outputBuilder.WriteString(fmt.Sprintf("   No model file found with quantization preferences: %v\n", quantizationPreferences))
+		}
+
+		if details.ReadmeFile != nil {
+			outputBuilder.WriteString(fmt.Sprintf("   README File: %s\n", details.ReadmeFile.Path))
+
+			// Find and set readme file
+			for _, file := range processedFiles {
+				if file.IsReadme {
+					processedModel.ReadmeFile = &file
+					break
+				}
+			}
+
+			fmt.Println("Getting real readme for", model.ModelID, "waiting...")
+			// Use agent to get the real readme and prepare the model description
+			readmeContent, err := getRealReadme(context.Background(), model.ModelID)
+			if err == nil {
+				processedModel.ReadmeContent = readmeContent
+				processedModel.ReadmeContentPreview = truncateString(readmeContent, 200)
+				outputBuilder.WriteString(fmt.Sprintf("   README Content Preview: %s\n",
+					processedModel.ReadmeContentPreview))
+			} else {
+				continue
+			}
+			fmt.Println("Real readme got", readmeContent)
+			// Get README content
+			// readmeContent, err := client.GetReadmeContent(model.ModelID, details.ReadmeFile.Path)
+			// if err == nil {
+			// 	processedModel.ReadmeContent = readmeContent
+			// 	processedModel.ReadmeContentPreview = truncateString(readmeContent, 200)
+			// 	outputBuilder.WriteString(fmt.Sprintf("   README Content Preview: %s\n",
+			// 		processedModel.ReadmeContentPreview))
+			// }
+		}
+
+		// Print all files with their checksums
+		outputBuilder.WriteString("   All Files:\n")
+		for _, file := range processedFiles {
+			outputBuilder.WriteString(fmt.Sprintf("     - %s (%s, %d bytes", file.Path, file.FileType, file.Size))
+			if file.SHA256 != "" {
+				outputBuilder.WriteString(fmt.Sprintf(", SHA256: %s", file.SHA256))
+			}
+			outputBuilder.WriteString(")\n")
+		}
+
+		outputBuilder.WriteString("\n")
+		result.Models = append(result.Models, processedModel)
+	}
+
+	result.FormattedOutput = outputBuilder.String()
+	return result, nil
+}
+
+func truncateString(s string, maxLen int) string {
+	if len(s) <= maxLen {
+		return s
+	}
+	return s[:maxLen] + "..."
+}

.github/gallery-agent/testing.go

@@ -0,0 +1,190 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"math/rand"
+	"strings"
+	"time"
+)
+
+// runSyntheticMode generates synthetic test data and appends it to the gallery
+func runSyntheticMode() error {
+	generator := NewSyntheticDataGenerator()
+
+	// Generate a random number of synthetic models (1-3)
+	numModels := generator.rand.Intn(3) + 1
+	fmt.Printf("Generating %d synthetic models for testing...\n", numModels)
+
+	var models []ProcessedModel
+	for i := 0; i < numModels; i++ {
+		model := generator.GenerateProcessedModel()
+		models = append(models, model)
+		fmt.Printf("Generated synthetic model: %s\n", model.ModelID)
+	}
+
+	// Generate YAML entries and append to gallery/index.yaml
+	fmt.Println("Generating YAML entries for synthetic models...")
+	err := generateYAMLForModels(context.Background(), models)
+	if err != nil {
+		return fmt.Errorf("error generating YAML entries: %w", err)
+	}
+
+	fmt.Printf("Successfully added %d synthetic models to the gallery for testing!\n", len(models))
+	return nil
+}
+
+// SyntheticDataGenerator provides methods to generate synthetic test data
+type SyntheticDataGenerator struct {
+	rand *rand.Rand
+}
+
+// NewSyntheticDataGenerator creates a new synthetic data generator
+func NewSyntheticDataGenerator() *SyntheticDataGenerator {
+	return &SyntheticDataGenerator{
+		rand: rand.New(rand.NewSource(time.Now().UnixNano())),
+	}
+}
+
+// GenerateProcessedModelFile creates a synthetic ProcessedModelFile
+func (g *SyntheticDataGenerator) GenerateProcessedModelFile() ProcessedModelFile {
+	fileTypes := []string{"model", "readme", "other"}
+	fileType := fileTypes[g.rand.Intn(len(fileTypes))]
+
+	var path string
+	var isReadme bool
+
+	switch fileType {
+	case "model":
+		path = fmt.Sprintf("model-%s.gguf", g.randomString(8))
+		isReadme = false
+	case "readme":
+		path = "README.md"
+		isReadme = true
+	default:
+		path = fmt.Sprintf("file-%s.txt", g.randomString(6))
+		isReadme = false
+	}
+
+	return ProcessedModelFile{
+		Path:     path,
+		Size:     int64(g.rand.Intn(1000000000) + 1000000), // 1MB to 1GB
+		SHA256:   g.randomSHA256(),
+		IsReadme: isReadme,
+		FileType: fileType,
+	}
+}
+
+// GenerateProcessedModel creates a synthetic ProcessedModel
+func (g *SyntheticDataGenerator) GenerateProcessedModel() ProcessedModel {
+	authors := []string{"microsoft", "meta", "google", "openai", "anthropic", "mistralai", "huggingface"}
+	modelNames := []string{"llama", "gpt", "claude", "mistral", "gemma", "phi", "qwen", "codellama"}
+
+	author := authors[g.rand.Intn(len(authors))]
+	modelName := modelNames[g.rand.Intn(len(modelNames))]
+	modelID := fmt.Sprintf("%s/%s-%s", author, modelName, g.randomString(6))
+
+	// Generate files
+	numFiles := g.rand.Intn(5) + 2 // 2-6 files
+	files := make([]ProcessedModelFile, numFiles)
+
+	// Ensure at least one model file and one readme
+	hasModelFile := false
+	hasReadme := false
+
+	for i := 0; i < numFiles; i++ {
+		files[i] = g.GenerateProcessedModelFile()
+		if files[i].FileType == "model" {
+			hasModelFile = true
+		}
+		if files[i].FileType == "readme" {
+			hasReadme = true
+		}
+	}
+
+	// Add required files if missing
+	if !hasModelFile {
+		modelFile := g.GenerateProcessedModelFile()
+		modelFile.FileType = "model"
+		modelFile.Path = fmt.Sprintf("%s-Q4_K_M.gguf", modelName)
+		files = append(files, modelFile)
+	}
+
+	if !hasReadme {
+		readmeFile := g.GenerateProcessedModelFile()
+		readmeFile.FileType = "readme"
+		readmeFile.Path = "README.md"
+		readmeFile.IsReadme = true
+		files = append(files, readmeFile)
+	}
+
+	// Find preferred model file
+	var preferredModelFile *ProcessedModelFile
+	for i := range files {
+		if files[i].FileType == "model" {
+			preferredModelFile = &files[i]
+			break
+		}
+	}
+
+	// Find readme file
+	var readmeFile *ProcessedModelFile
+	for i := range files {
+		if files[i].FileType == "readme" {
+			readmeFile = &files[i]
+			break
+		}
+	}
+
+	readmeContent := g.generateReadmeContent(modelName, author)
+
+	return ProcessedModel{
+		ModelID:                 modelID,
+		Author:                  author,
+		Downloads:               g.rand.Intn(1000000) + 1000,
+		LastModified:            g.randomDate(),
+		Files:                   files,
+		PreferredModelFile:      preferredModelFile,
+		ReadmeFile:              readmeFile,
+		ReadmeContent:           readmeContent,
+		ReadmeContentPreview:    truncateString(readmeContent, 200),
+		QuantizationPreferences: []string{"Q4_K_M", "Q4_K_S", "Q3_K_M", "Q2_K"},
+		ProcessingError:         "",
+	}
+}
+
+// Helper methods for synthetic data generation
+func (g *SyntheticDataGenerator) randomString(length int) string {
+	const charset = "abcdefghijklmnopqrstuvwxyz0123456789"
+	b := make([]byte, length)
+	for i := range b {
+		b[i] = charset[g.rand.Intn(len(charset))]
+	}
+	return string(b)
+}
+
+func (g *SyntheticDataGenerator) randomSHA256() string {
+	const charset = "0123456789abcdef"
+	b := make([]byte, 64)
+	for i := range b {
+		b[i] = charset[g.rand.Intn(len(charset))]
+	}
+	return string(b)
+}
+
+func (g *SyntheticDataGenerator) randomDate() string {
+	now := time.Now()
+	daysAgo := g.rand.Intn(365) // Random date within last year
+	pastDate := now.AddDate(0, 0, -daysAgo)
+	return pastDate.Format("2006-01-02T15:04:05.000Z")
+}
+
+func (g *SyntheticDataGenerator) generateReadmeContent(modelName, author string) string {
+	templates := []string{
+		fmt.Sprintf("# %s Model\n\nThis is a %s model developed by %s. It's designed for various natural language processing tasks including text generation, question answering, and conversation.\n\n## Features\n\n- High-quality text generation\n- Efficient inference\n- Multiple quantization options\n- Easy to use with LocalAI\n\n## Usage\n\nUse this model with LocalAI for various AI tasks.", strings.Title(modelName), modelName, author),
+		fmt.Sprintf("# %s\n\nA powerful language model from %s. This model excels at understanding and generating human-like text across multiple domains.\n\n## Capabilities\n\n- Text completion\n- Code generation\n- Creative writing\n- Technical documentation\n\n## Model Details\n\n- Architecture: Transformer-based\n- Training: Large-scale supervised learning\n- Quantization: Available in multiple formats", strings.Title(modelName), author),
+		fmt.Sprintf("# %s Language Model\n\nDeveloped by %s, this model represents state-of-the-art performance in natural language understanding and generation.\n\n## Key Features\n\n- Multilingual support\n- Context-aware responses\n- Efficient memory usage\n- Fast inference speed\n\n## Applications\n\n- Chatbots and virtual assistants\n- Content generation\n- Code completion\n- Educational tools", strings.Title(modelName), author),
+	}
+
+	return templates[g.rand.Intn(len(templates))]
+}

.github/gallery-agent/tools.go

@@ -0,0 +1,46 @@
+package main
+
+import (
+	"fmt"
+
+	hfapi "github.com/mudler/LocalAI/pkg/huggingface-api"
+	openai "github.com/sashabaranov/go-openai"
+	jsonschema "github.com/sashabaranov/go-openai/jsonschema"
+)
+
+// Get repository README from HF
+type HFReadmeTool struct {
+	client *hfapi.Client
+}
+
+func (s *HFReadmeTool) Execute(args map[string]any) (string, error) {
+	q, ok := args["repository"].(string)
+	if !ok {
+		return "", fmt.Errorf("no query")
+	}
+	readme, err := s.client.GetReadmeContent(q, "README.md")
+	if err != nil {
+		return "", err
+	}
+	return readme, nil
+}
+
+func (s *HFReadmeTool) Tool() openai.Tool {
+	return openai.Tool{
+		Type: openai.ToolTypeFunction,
+		Function: &openai.FunctionDefinition{
+			Name:        "hf_readme",
+			Description: "A tool to get the README content of a huggingface repository",
+			Parameters: jsonschema.Definition{
+				Type: jsonschema.Object,
+				Properties: map[string]jsonschema.Definition{
+					"repository": {
+						Type:        jsonschema.String,
+						Description: "The huggingface repository to get the README content of",
+					},
+				},
+				Required: []string{"repository"},
+			},
+		},
+	}
+}

.github/workflows/backend.yml

@@ -111,6 +111,18 @@ jobs:
             backend: "diffusers"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
+          - build-type: ''
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-cpu-chatterbox'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"
+            skip-drivers: 'true'
+            backend: "chatterbox"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
           # CUDA 11 additional backends
           - build-type: 'cublas'
             cuda-major-version: "11"
@@ -477,6 +489,18 @@ jobs:
             backend: "diffusers"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
+          - build-type: 'l4t'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-nvidia-l4t-kokoro'
+            runs-on: 'ubuntu-24.04-arm'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            skip-drivers: 'true'
+            backend: "kokoro"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
           # SYCL additional backends
           - build-type: 'intel'
             cuda-major-version: ""
@@ -763,7 +787,7 @@ jobs:
             cuda-minor-version: ""
             platforms: 'linux/amd64'
             tag-latest: 'auto'
-            tag-suffix: '-gpu-hipblas-whisper'
+            tag-suffix: '-gpu-rocm-hipblas-whisper'
             base-image: "rocm/dev-ubuntu-22.04:6.4.3"
             runs-on: 'ubuntu-latest'
             skip-drivers: 'false'
@@ -858,7 +882,7 @@ jobs:
             backend: "rfdetr"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
-          - build-type: 'cublas'
+          - build-type: 'l4t'
             cuda-major-version: "12"
             cuda-minor-version: "0"
             platforms: 'linux/arm64'
@@ -931,6 +955,18 @@ jobs:
             backend: "exllama2"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
+          - build-type: 'l4t'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            skip-drivers: 'true'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-nvidia-l4t-arm64-chatterbox'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            runs-on: 'ubuntu-24.04-arm'
+            backend: "chatterbox"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
           # runs out of space on the runner
           # - build-type: 'hipblas'
           #   cuda-major-version: ""
@@ -957,6 +993,55 @@ jobs:
             backend: "kitten-tts"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
+          # neutts
+          - build-type: ''
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64,linux/arm64'
+            tag-latest: 'auto'
+            tag-suffix: '-cpu-neutts'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"
+            skip-drivers: 'false'
+            backend: "neutts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
+          - build-type: 'cublas'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-nvidia-cuda-12-neutts'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"
+            skip-drivers: 'false'
+            backend: "neutts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
+          - build-type: 'hipblas'
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-rocm-hipblas-neutts'
+            runs-on: 'arc-runner-set'
+            base-image: "rocm/dev-ubuntu-22.04:6.4.3"
+            skip-drivers: 'false'
+            backend: "neutts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
+          - build-type: 'l4t'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            skip-drivers: 'true'
+            tag-latest: 'auto'
+            tag-suffix: '-nvidia-l4t-arm64-neutts'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            runs-on: 'ubuntu-24.04-arm'
+            backend: "neutts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
   backend-jobs-darwin:
     uses: ./.github/workflows/backend_build_darwin.yml
     strategy:
@@ -968,6 +1053,9 @@ jobs:
           - backend: "mlx"
             tag-suffix: "-metal-darwin-arm64-mlx"
             build-type: "mps"
+          - backend: "chatterbox"
+            tag-suffix: "-metal-darwin-arm64-chatterbox"
+            build-type: "mps"
           - backend: "mlx-vlm"
             tag-suffix: "-metal-darwin-arm64-mlx-vlm"
             build-type: "mps"
@@ -1002,7 +1090,7 @@ jobs:
         go-version: ['1.21.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
@@ -1021,7 +1109,7 @@ jobs:
           make protogen-go
           make backends/llama-cpp-darwin
       - name: Upload llama-cpp.tar
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v5
         with:
           name: llama-cpp-tar
           path: backend-images/llama-cpp.tar
@@ -1031,7 +1119,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Download llama-cpp.tar
-        uses: actions/download-artifact@v5
+        uses: actions/download-artifact@v6
         with:
           name: llama-cpp-tar
           path: .
@@ -1088,7 +1176,7 @@ jobs:
         go-version: ['1.21.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
@@ -1109,7 +1197,7 @@ jobs:
           export PLATFORMARCH=darwin/amd64
           make backends/llama-cpp-darwin
       - name: Upload llama-cpp.tar
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v5
         with:
           name: llama-cpp-tar-x86
           path: backend-images/llama-cpp.tar
@@ -1119,7 +1207,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Download llama-cpp.tar
-        uses: actions/download-artifact@v5
+        uses: actions/download-artifact@v6
         with:
           name: llama-cpp-tar-x86
           path: .

.github/workflows/backend_build.yml

@@ -97,7 +97,7 @@ jobs:
           && sudo apt-get install -y git
 
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Release space from worker
         if: inputs.runs-on == 'ubuntu-latest'

.github/workflows/backend_build_darwin.yml

@@ -50,7 +50,7 @@ jobs:
         go-version: ['${{ inputs.go-version }}']
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
 
@@ -74,7 +74,7 @@ jobs:
           BACKEND=${{ inputs.backend }} BUILD_TYPE=${{ inputs.build-type }} USE_PIP=${{ inputs.use-pip }} make build-darwin-${{ inputs.lang }}-backend
 
       - name: Upload ${{ inputs.backend }}.tar
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v5
         with:
           name: ${{ inputs.backend }}-tar
           path: backend-images/${{ inputs.backend }}.tar
@@ -85,7 +85,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Download ${{ inputs.backend }}.tar
-        uses: actions/download-artifact@v5
+        uses: actions/download-artifact@v6
         with:
           name: ${{ inputs.backend }}-tar
           path: .

.github/workflows/backend_pr.yml

@@ -17,7 +17,7 @@ jobs:
       has-backends-darwin: ${{ steps.set-matrix.outputs.has-backends-darwin }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Setup Bun
         uses: oven-sh/setup-bun@v2

.github/workflows/build-test.yaml

@@ -11,13 +11,13 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
-          go-version: 1.23
+          go-version: 1.25
       - name: Run GoReleaser
         run: |
           make dev-dist
@@ -25,19 +25,19 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
-          go-version: 1.23
+          go-version: 1.25
       - name: Build launcher for macOS ARM64
         run: |
           make build-launcher-darwin
           ls -liah dist
       - name: Upload macOS launcher artifacts
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v5
         with:
           name: launcher-macos
           path: dist/
@@ -47,20 +47,20 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
-          go-version: 1.23
+          go-version: 1.25
       - name: Build launcher for Linux
         run: |
           sudo apt-get update
           sudo apt-get install golang gcc libgl1-mesa-dev xorg-dev libxkbcommon-dev
           make build-launcher-linux
       - name: Upload Linux launcher artifacts
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v5
         with:
           name: launcher-linux
           path: local-ai-launcher-linux.tar.xz

.github/workflows/bump_deps.yaml

@@ -1,10 +1,10 @@
-name: Bump dependencies
+name: Bump Backend dependencies
 on:
   schedule:
     - cron: 0 20 * * *
   workflow_dispatch:
 jobs:
-  bump:
+  bump-backends:
     strategy:
       fail-fast: false
       matrix:
@@ -31,7 +31,7 @@ jobs:
             file: "backend/go/piper/Makefile"
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - name: Bump dependencies 🔧
         id: bump
         run: |

.github/workflows/bump_docs.yaml

@@ -1,10 +1,10 @@
-name: Bump dependencies
+name: Bump Documentation
 on:
   schedule:
     - cron: 0 20 * * *
   workflow_dispatch:
 jobs:
-  bump:
+  bump-docs:
     strategy:
       fail-fast: false
       matrix:
@@ -12,7 +12,7 @@ jobs:
           - repository: "mudler/LocalAI"
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - name: Bump dependencies 🔧
         run: |
           bash .github/bump_docs.sh ${{ matrix.repository }}

.github/workflows/checksum_checker.yaml

@@ -15,7 +15,7 @@ jobs:
           && sudo add-apt-repository -y ppa:git-core/ppa \
           && sudo apt-get update \
           && sudo apt-get install -y git
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - name: Install dependencies
         run: |
           sudo apt-get update

.github/workflows/dependabot_auto.yml

@@ -20,7 +20,7 @@ jobs:
           skip-commit-verification: true
 
       - name: Checkout repository
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Approve a PR if not already approved
         run: |

.github/workflows/deploy-explorer.yaml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - uses: actions/setup-go@v5
@@ -33,7 +33,7 @@ jobs:
         run: |
           CGO_ENABLED=0 make build
       - name: rm
-        uses: appleboy/ssh-action@v1.2.2
+        uses: appleboy/ssh-action@v1.2.3
         with:
             host: ${{ secrets.EXPLORER_SSH_HOST }}
             username: ${{ secrets.EXPLORER_SSH_USERNAME }}
@@ -53,7 +53,7 @@ jobs:
             rm: true
             target: ./local-ai
       - name: restarting
-        uses: appleboy/ssh-action@v1.2.2
+        uses: appleboy/ssh-action@v1.2.3
         with:
             host: ${{ secrets.EXPLORER_SSH_HOST }}
             username: ${{ secrets.EXPLORER_SSH_USERNAME }}

.github/workflows/gallery-agent.yaml

@@ -0,0 +1,119 @@
+name: Gallery Agent
+on:
+
+  schedule:
+    - cron: '0 */3 * * *'  # Run every 4 hours
+  workflow_dispatch:
+    inputs:
+      search_term:
+        description: 'Search term for models'
+        required: false
+        default: 'GGUF'
+        type: string
+      limit:
+        description: 'Maximum number of models to process'
+        required: false
+        default: '15'
+        type: string
+      quantization:
+        description: 'Preferred quantization format'
+        required: false
+        default: 'Q4_K_M'
+        type: string
+      max_models:
+        description: 'Maximum number of models to add to the gallery'
+        required: false
+        default: '1'
+        type: string
+jobs:
+  gallery-agent:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Set up Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: '1.21'
+
+
+      - name: Run gallery agent
+        env:
+          OPENAI_MODEL: ${{ secrets.OPENAI_MODEL }}
+          OPENAI_KEY: ${{ secrets.OPENAI_KEY }}
+          OPENAI_BASE_URL: ${{ secrets.OPENAI_BASE_URL }}
+          SEARCH_TERM: ${{ github.event.inputs.search_term || 'GGUF' }}
+          LIMIT: ${{ github.event.inputs.limit || '15' }}
+          QUANTIZATION: ${{ github.event.inputs.quantization || 'Q4_K_M' }}
+          MAX_MODELS: ${{ github.event.inputs.max_models || '1' }}
+        run: |
+          export GALLERY_INDEX_PATH=$PWD/gallery/index.yaml
+          go run .github/gallery-agent
+
+      - name: Check for changes
+        id: check_changes
+        run: |
+          if git diff --quiet gallery/index.yaml; then
+            echo "changes=false" >> $GITHUB_OUTPUT
+            echo "No changes detected in gallery/index.yaml"
+          else
+            echo "changes=true" >> $GITHUB_OUTPUT
+            echo "Changes detected in gallery/index.yaml"
+            git diff gallery/index.yaml
+          fi
+
+      - name: Read gallery agent summary
+        id: read_summary
+        if: steps.check_changes.outputs.changes == 'true'
+        run: |
+          if [ -f ".github/gallery-agent/gallery-agent-summary.json" ]; then
+            echo "summary_exists=true" >> $GITHUB_OUTPUT
+            # Extract summary data using jq
+            echo "search_term=$(jq -r '.search_term' .github/gallery-agent/gallery-agent-summary.json)" >> $GITHUB_OUTPUT
+            echo "total_found=$(jq -r '.total_found' .github/gallery-agent/gallery-agent-summary.json)" >> $GITHUB_OUTPUT
+            echo "models_added=$(jq -r '.models_added' .github/gallery-agent/gallery-agent-summary.json)" >> $GITHUB_OUTPUT
+            echo "quantization=$(jq -r '.quantization' .github/gallery-agent/gallery-agent-summary.json)" >> $GITHUB_OUTPUT
+            echo "processing_time=$(jq -r '.processing_time' .github/gallery-agent/gallery-agent-summary.json)" >> $GITHUB_OUTPUT
+            
+            # Create a formatted list of added models with URLs
+            added_models=$(jq -r 'range(0; .added_model_ids | length) as $i | "- [\(.added_model_ids[$i])](\(.added_model_urls[$i]))"' .github/gallery-agent/gallery-agent-summary.json | tr '\n' '\n')
+            echo "added_models<<EOF" >> $GITHUB_OUTPUT
+            echo "$added_models" >> $GITHUB_OUTPUT
+            echo "EOF" >> $GITHUB_OUTPUT
+            rm -f .github/gallery-agent/gallery-agent-summary.json
+          else
+            echo "summary_exists=false" >> $GITHUB_OUTPUT
+          fi
+
+      - name: Create Pull Request
+        if: steps.check_changes.outputs.changes == 'true'
+        uses: peter-evans/create-pull-request@v7
+        with:
+          token: ${{ secrets.UPDATE_BOT_TOKEN }}
+          push-to-fork: ci-forks/LocalAI
+          commit-message: 'chore(model gallery): :robot: add new models via gallery agent'
+          title: 'chore(model gallery): :robot: add ${{ steps.read_summary.outputs.models_added || 0 }} new models via gallery agent'
+          # Branch has to be unique so PRs are not overriding each other
+          branch-suffix: timestamp
+          body: |
+            This PR was automatically created by the gallery agent workflow.
+            
+            **Summary:**
+            - **Search Term:** ${{ steps.read_summary.outputs.search_term || github.event.inputs.search_term || 'GGUF' }}
+            - **Models Found:** ${{ steps.read_summary.outputs.total_found || 'N/A' }}
+            - **Models Added:** ${{ steps.read_summary.outputs.models_added || '0' }}
+            - **Quantization:** ${{ steps.read_summary.outputs.quantization || github.event.inputs.quantization || 'Q4_K_M' }}
+            - **Processing Time:** ${{ steps.read_summary.outputs.processing_time || 'N/A' }}
+            
+            **Added Models:**
+            ${{ steps.read_summary.outputs.added_models || '- No models added' }}
+            
+            **Workflow Details:**
+            - Triggered by: `${{ github.event_name }}`
+            - Run ID: `${{ github.run_id }}`
+            - Commit: `${{ github.sha }}`
+          signoff: true
+          delete-branch: true

.github/workflows/generate_grpc_cache.yaml

@@ -73,7 +73,7 @@ jobs:
         uses: docker/setup-buildx-action@master
 
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Cache GRPC
         uses: docker/build-push-action@v6

.github/workflows/generate_intel_image.yaml

@@ -16,7 +16,7 @@ jobs:
       matrix:
         include:
           - base-image: intel/oneapi-basekit:2025.2.0-0-devel-ubuntu22.04
-            runs-on: 'ubuntu-latest'
+            runs-on: 'arc-runner-set'
             platforms: 'linux/amd64'
     runs-on: ${{matrix.runs-on}}
     steps:
@@ -43,7 +43,7 @@ jobs:
         uses: docker/setup-buildx-action@master
 
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Cache Intel images
         uses: docker/build-push-action@v6

.github/workflows/image_build.yml

@@ -94,7 +94,7 @@ jobs:
           && sudo apt-get update \
           && sudo apt-get install -y git
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Release space from worker
         if: inputs.runs-on == 'ubuntu-latest'

.github/workflows/labeler.yml

@@ -9,4 +9,4 @@ jobs:
       pull-requests: write
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/labeler@v5
+    - uses: actions/labeler@v6

.github/workflows/localaibot_automerge.yml

@@ -6,14 +6,15 @@ permissions:
   contents: write
   pull-requests: write
   packages: read
-
+  issues: write # for Homebrew/actions/post-comment
+  actions: write # to dispatch publish workflow
 jobs:
   dependabot:
     runs-on: ubuntu-latest
-    if: ${{ github.actor == 'localai-bot' }}
+    if: ${{ github.actor == 'localai-bot' && !contains(github.event.pull_request.title, 'chore(model gallery):') }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
 
       - name: Approve a PR if not already approved
         run: |

.github/workflows/notify-models.yaml

@@ -1,22 +1,27 @@
 name: Notifications for new models
 on:
-  pull_request:
+  pull_request_target:
      types:
        - closed
 
+permissions:
+  contents: read
+  pull-requests: read
+
 jobs:
   notify-discord:
     if: ${{ (github.event.pull_request.merged == true) && (contains(github.event.pull_request.labels.*.name, 'area/ai-model')) }}
     env:
-        MODEL_NAME: gemma-3-12b-it
+        MODEL_NAME: gemma-3-12b-it-qat
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v5
+    - uses: actions/checkout@v6
       with:
         fetch-depth: 0 # needed to checkout all branches for this Action to work
+        ref: ${{ github.event.pull_request.head.sha }} # Checkout the PR head to get the actual changes
     - uses: mudler/localai-github-action@v1
       with:
-        model: 'gemma-3-12b-it' # Any from models.localai.io, or from huggingface.com with: "huggingface://<repository>/file"
+        model: 'gemma-3-12b-it-qat' # Any from models.localai.io, or from huggingface.com with: "huggingface://<repository>/file"
         # Check the PR diff using the current branch and the base branch of the PR
     - uses: GrantBirki/git-diff-action@v2.8.1
       id: git-diff-action
@@ -79,7 +84,7 @@ jobs:
         args: ${{ steps.summarize.outputs.message }}
     - name: Setup tmate session if fails
       if: ${{ failure() }}
-      uses: mxschmitt/action-tmate@v3.22
+      uses: mxschmitt/action-tmate@v3.23
       with:
         detached: true
         connect-timeout-seconds: 180
@@ -87,12 +92,13 @@ jobs:
   notify-twitter:
     if: ${{ (github.event.pull_request.merged == true) && (contains(github.event.pull_request.labels.*.name, 'area/ai-model')) }}
     env:
-        MODEL_NAME: gemma-3-12b-it
+        MODEL_NAME: gemma-3-12b-it-qat
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v5
+    - uses: actions/checkout@v6
       with:
         fetch-depth: 0 # needed to checkout all branches for this Action to work
+        ref: ${{ github.event.pull_request.head.sha }} # Checkout the PR head to get the actual changes
     - name: Start LocalAI
       run: |
         echo "Starting LocalAI..."
@@ -161,7 +167,7 @@ jobs:
         TWITTER_ACCESS_TOKEN_SECRET: ${{ secrets.TWITTER_ACCESS_TOKEN_SECRET }}
     - name: Setup tmate session if fails
       if: ${{ failure() }}
-      uses: mxschmitt/action-tmate@v3.22
+      uses: mxschmitt/action-tmate@v3.23
       with:
         detached: true
         connect-timeout-seconds: 180

.github/workflows/notify-releases.yaml

@@ -11,10 +11,11 @@ jobs:
         RELEASE_BODY: ${{ github.event.release.body }}
         RELEASE_TITLE: ${{ github.event.release.name }}
         RELEASE_TAG_NAME: ${{ github.event.release.tag_name }}
+        MODEL_NAME: gemma-3-12b-it-qat
     steps:
     - uses: mudler/localai-github-action@v1
       with:
-        model: 'gemma-3-12b-it' # Any from models.localai.io, or from huggingface.com with: "huggingface://<repository>/file"
+        model: 'gemma-3-12b-it-qat' # Any from models.localai.io, or from huggingface.com with: "huggingface://<repository>/file"
     - name: Summarize
       id: summarize
       run: |

.github/workflows/release.yaml

@@ -10,7 +10,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -28,7 +28,7 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -41,12 +41,12 @@ jobs:
       - name: Upload DMG to Release
         uses: softprops/action-gh-release@v2
         with:
-          files: ./dist/LocalAI-Launcher.dmg
+          files: ./dist/LocalAI.dmg
   launcher-build-linux:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
       - name: Set up Go
@@ -61,4 +61,4 @@ jobs:
       - name: Upload Linux launcher artifacts
         uses: softprops/action-gh-release@v2
         with:
-          files: ./local-ai-launcher-linux.tar.xz
+          files: ./local-ai-launcher-linux.tar.xz

.github/workflows/secscan.yaml

@@ -14,17 +14,17 @@ jobs:
       GO111MODULE: on
     steps:
       - name: Checkout Source
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         if: ${{ github.actor != 'dependabot[bot]' }}
       - name: Run Gosec Security Scanner
         if: ${{ github.actor != 'dependabot[bot]' }}
-        uses: securego/gosec@v2.22.8
+        uses: securego/gosec@v2.22.9
         with:
           # we let the report trigger content trigger a failure using the GitHub Security features.
           args: '-no-fail -fmt sarif -out results.sarif ./...'
       - name: Upload SARIF file
         if: ${{ github.actor != 'dependabot[bot]' }}
-        uses: github/codeql-action/upload-sarif@v3
+        uses: github/codeql-action/upload-sarif@v4
         with:
           # Path to SARIF file relative to the root of the repository
           sarif_file: results.sarif

.github/workflows/stalebot.yml

@@ -10,7 +10,7 @@ jobs:
   stale:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/stale@5bef64f19d7facfb25b37b414482c7164d639639 # v9
+      - uses: actions/stale@5f858e3efba33a5ca4407a664cc011ad407f2008 # v9
         with:
           stale-issue-message: 'This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 5 days.'
           stale-pr-message: 'This PR is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 10 days.'

.github/workflows/test-extra.yml

@@ -19,7 +19,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v5
+  #       uses: actions/checkout@v6
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -40,7 +40,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Dependencies
@@ -61,7 +61,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Dependencies
@@ -83,7 +83,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Dependencies
@@ -104,7 +104,7 @@ jobs:
   #  runs-on: ubuntu-latest
   #  steps:
   #    - name: Clone
-  #      uses: actions/checkout@v5
+  #      uses: actions/checkout@v6
   #      with:
   #        submodules: true
   #    - name: Dependencies
@@ -124,7 +124,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v5
+  #       uses: actions/checkout@v6
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -186,7 +186,7 @@ jobs:
   #           sudo rm -rf "$AGENT_TOOLSDIRECTORY" || true
   #           df -h
   #     - name: Clone
-  #       uses: actions/checkout@v5
+  #       uses: actions/checkout@v6
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -211,7 +211,7 @@ jobs:
   #   runs-on: ubuntu-latest
   #   steps:
   #     - name: Clone
-  #       uses: actions/checkout@v5
+  #       uses: actions/checkout@v6
   #       with:
   #         submodules: true
   #     - name: Dependencies
@@ -232,7 +232,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Dependencies

.github/workflows/test.yml

@@ -21,7 +21,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        go-version: ['1.21.x']
+        go-version: ['1.25.x']
     steps:
       - name: Free Disk Space (Ubuntu)
         uses: jlumbroso/free-disk-space@main
@@ -70,7 +70,7 @@ jobs:
           sudo rm -rfv build || true
           df -h
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
@@ -124,7 +124,7 @@ jobs:
           PATH="$PATH:/root/go/bin" GO_TAGS="tts" make --jobs 5 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3.22
+        uses: mxschmitt/action-tmate@v3.23
         with:
           detached: true
           connect-timeout-seconds: 180
@@ -166,7 +166,7 @@ jobs:
           sudo rm -rfv build || true
           df -h
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Dependencies
@@ -183,7 +183,7 @@ jobs:
             PATH="$PATH:$HOME/go/bin" make backends/local-store backends/silero-vad backends/llama-cpp backends/whisper backends/piper backends/stablediffusion-ggml docker-build-aio e2e-aio
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3.22
+        uses: mxschmitt/action-tmate@v3.23
         with:
           detached: true
           connect-timeout-seconds: 180
@@ -193,10 +193,10 @@ jobs:
     runs-on: macOS-14
     strategy:
       matrix:
-        go-version: ['1.21.x']
+        go-version: ['1.25.x']
     steps:
       - name: Clone
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
         with:
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
@@ -226,7 +226,7 @@ jobs:
           PATH="$PATH:$HOME/go/bin" BUILD_TYPE="GITHUB_CI_HAS_BROKEN_METAL" CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF" make --jobs 4 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3.22
+        uses: mxschmitt/action-tmate@v3.23
         with:
           detached: true
           connect-timeout-seconds: 180

.github/workflows/update_swagger.yaml

@@ -9,7 +9,7 @@ jobs:
       fail-fast: false
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - uses: actions/setup-go@v5
         with:
           go-version: 'stable'

.gitmodules

@@ -1,6 +1,3 @@
 [submodule "docs/themes/hugo-theme-relearn"]
 	path = docs/themes/hugo-theme-relearn
 	url = https://github.com/McShelby/hugo-theme-relearn.git
-[submodule "docs/themes/lotusdocs"]
-	path = docs/themes/lotusdocs
-	url = https://github.com/colinwilson/lotusdocs

CONTRIBUTING.md

@@ -30,6 +30,7 @@ Thank you for your interest in contributing to LocalAI! We appreciate your time
 3. Install the required dependencies ( see https://localai.io/basics/build/#build-localai-locally )
 4. Build LocalAI: `make build`
 5. Run LocalAI: `./local-ai`
+6. To Build and live reload: `make build-dev`
 
 ## Contributing
 
@@ -76,7 +77,7 @@ LOCALAI_IMAGE_TAG=test LOCALAI_IMAGE=local-ai-aio make run-e2e-aio
 ## Documentation
 
 We are welcome the contribution of the documents, please open new PR or create a new issue. The documentation is available under `docs/` https://github.com/mudler/LocalAI/tree/master/docs
- 
+
 ## Community and Communication
 
 - You can reach out via the Github issue tracker.

Dockerfile

@@ -78,6 +78,16 @@ RUN <<EOT bash
     fi
 EOT
 
+# https://github.com/NVIDIA/Isaac-GR00T/issues/343
+RUN <<EOT bash
+    if [ "${BUILD_TYPE}" = "cublas" ] && [ "${TARGETARCH}" = "arm64" ]; then
+        wget https://developer.download.nvidia.com/compute/cudss/0.6.0/local_installers/cudss-local-tegra-repo-ubuntu2204-0.6.0_0.6.0-1_arm64.deb && \
+        dpkg -i cudss-local-tegra-repo-ubuntu2204-0.6.0_0.6.0-1_arm64.deb && \
+        cp /var/cudss-local-tegra-repo-ubuntu2204-0.6.0/cudss-*-keyring.gpg /usr/share/keyrings/ && \
+        apt-get update && apt-get -y install cudss
+    fi
+EOT
+
 # If we are building with clblas support, we need the libraries for the builds
 RUN if [ "${BUILD_TYPE}" = "clblas" ] && [ "${SKIP_DRIVERS}" = "false" ]; then \
         apt-get update && \
@@ -322,6 +332,6 @@ RUN mkdir -p /models /backends
 HEALTHCHECK --interval=1m --timeout=10m --retries=10 \
   CMD curl -f ${HEALTHCHECK_ENDPOINT} || exit 1
 
-VOLUME /models /backends
+VOLUME /models /backends /configuration
 EXPOSE 8080
 ENTRYPOINT [ "/entrypoint.sh" ]

Makefile

@@ -103,6 +103,10 @@ build-launcher: ## Build the launcher application
 
 build-all: build build-launcher ## Build both server and launcher
 
+build-dev: ## Run LocalAI in dev mode with live reload
+	@command -v air >/dev/null 2>&1 || go install github.com/air-verse/air@latest
+	air -c .air.toml
+
 dev-dist:
 	$(GORELEASER) build --snapshot --clean
 
@@ -117,8 +121,8 @@ run: ## run local-ai
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" $(GOCMD) run ./
 
 test-models/testmodel.ggml:
-	mkdir test-models
-	mkdir test-dir
+	mkdir -p test-models
+	mkdir -p test-dir
 	wget -q https://huggingface.co/mradermacher/gpt2-alpaca-gpt4-GGUF/resolve/main/gpt2-alpaca-gpt4.Q4_K_M.gguf -O test-models/testmodel.ggml
 	wget -q https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.en.bin -O test-models/whisper-en
 	wget -q https://huggingface.co/mudler/all-MiniLM-L6-v2/resolve/main/ggml-model-q4_0.bin -O test-models/bert
@@ -369,10 +373,16 @@ backends/kitten-tts: docker-build-kitten-tts docker-save-kitten-tts build
 backends/kokoro: docker-build-kokoro docker-save-kokoro build
 	./local-ai backends install "ocifile://$(abspath ./backend-images/kokoro.tar)"
 
+backends/chatterbox: docker-build-chatterbox docker-save-chatterbox build
+	./local-ai backends install "ocifile://$(abspath ./backend-images/chatterbox.tar)"
+
 backends/llama-cpp-darwin: build
 	bash ./scripts/build/llama-cpp-darwin.sh
 	./local-ai backends install "ocifile://$(abspath ./backend-images/llama-cpp.tar)"
 
+backends/neutts: docker-build-neutts docker-save-neutts build
+	./local-ai backends install "ocifile://$(abspath ./backend-images/neutts.tar)"
+
 build-darwin-python-backend: build
 	bash ./scripts/build/python-darwin.sh
 
@@ -426,6 +436,15 @@ docker-build-kitten-tts:
 docker-save-kitten-tts: backend-images
 	docker save local-ai-backend:kitten-tts -o backend-images/kitten-tts.tar
 
+docker-save-chatterbox: backend-images
+	docker save local-ai-backend:chatterbox -o backend-images/chatterbox.tar
+
+docker-build-neutts:
+	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:neutts -f backend/Dockerfile.python --build-arg BACKEND=neutts ./backend
+
+docker-save-neutts: backend-images
+	docker save local-ai-backend:neutts -o backend-images/neutts.tar
+
 docker-build-kokoro:
 	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:kokoro -f backend/Dockerfile.python --build-arg BACKEND=kokoro ./backend
 
@@ -493,7 +512,7 @@ docker-build-bark:
 	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:bark -f backend/Dockerfile.python --build-arg BACKEND=bark .
 
 docker-build-chatterbox:
-	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:chatterbox -f backend/Dockerfile.python --build-arg BACKEND=chatterbox .
+	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:chatterbox -f backend/Dockerfile.python --build-arg BACKEND=chatterbox ./backend
 
 docker-build-exllama2:
 	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:exllama2 -f backend/Dockerfile.python --build-arg BACKEND=exllama2 .

README.md

@@ -43,7 +43,7 @@
 
 > :bulb: Get help - [❓FAQ](https://localai.io/faq/) [💭Discussions](https://github.com/go-skynet/LocalAI/discussions) [:speech_balloon: Discord](https://discord.gg/uJAeKSAGDy) [:book: Documentation website](https://localai.io/)
 >
-> [💻 Quickstart](https://localai.io/basics/getting_started/) [🖼️ Models](https://models.localai.io/) [🚀 Roadmap](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap) [🥽 Demo](https://demo.localai.io) [🌍 Explorer](https://explorer.localai.io) [🛫 Examples](https://github.com/mudler/LocalAI-examples) Try on 
+> [💻 Quickstart](https://localai.io/basics/getting_started/) [🖼️ Models](https://models.localai.io/) [🚀 Roadmap](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap) [🛫 Examples](https://github.com/mudler/LocalAI-examples) Try on 
 [![Telegram](https://img.shields.io/badge/Telegram-2CA5E0?style=for-the-badge&logo=telegram&logoColor=white)](https://t.me/localaiofficial_bot)
 
 [![tests](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml)[![Build and Release](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml)[![build container images](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml)[![Bump dependencies](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml)[![Artifact Hub](https://img.shields.io/endpoint?url=https://artifacthub.io/badge/repository/localai)](https://artifacthub.io/packages/search?repo=localai)
@@ -108,10 +108,25 @@ Run the installer script:
 curl https://localai.io/install.sh | sh
 ```
 
-For more installation options, see [Installer Options](https://localai.io/docs/advanced/installer/).
+For more installation options, see [Installer Options](https://localai.io/installation/).
+
+### macOS Download:
+
+<a href="https://github.com/mudler/LocalAI/releases/latest/download/LocalAI.dmg">
+  <img src="https://img.shields.io/badge/Download-macOS-blue?style=for-the-badge&logo=apple&logoColor=white" alt="Download LocalAI for macOS"/>
+</a>
+
+> Note: the DMGs are not signed by Apple as quarantined. See https://github.com/mudler/LocalAI/issues/6268 for a workaround, fix is tracked here: https://github.com/mudler/LocalAI/issues/6244
 
 Or run with docker:
 
+> **💡 Docker Run vs Docker Start**
+> 
+> - `docker run` creates and starts a new container. If a container with the same name already exists, this command will fail.
+> - `docker start` starts an existing container that was previously created with `docker run`.
+> 
+> If you've already run LocalAI before and want to start it again, use: `docker start -i local-ai`
+
 ### CPU only image:
 
 ```bash
@@ -187,10 +202,13 @@ local-ai run oci://localai/phi-2:latest
 
 > ⚡ **Automatic Backend Detection**: When you install models from the gallery or YAML files, LocalAI automatically detects your system's GPU capabilities (NVIDIA, AMD, Intel) and downloads the appropriate backend. For advanced configuration options, see [GPU Acceleration](https://localai.io/features/gpu-acceleration/#automatic-backend-detection).
 
-For more information, see [💻 Getting started](https://localai.io/basics/getting_started/index.html)
+For more information, see [💻 Getting started](https://localai.io/basics/getting_started/index.html), if you are interested in our roadmap items and future enhancements, you can see the [Issues labeled as Roadmap here](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap)
 
 ## 📰 Latest project news
 
+- November 2025: Major improvements to the UX. Among these: [Import models via URL](https://github.com/mudler/LocalAI/pull/7245) and [Multiple chats and history](https://github.com/mudler/LocalAI/pull/7325)
+- October 2025: 🔌 [Model Context Protocol (MCP)](https://localai.io/docs/features/mcp/) support added for agentic capabilities with external tools
+- September 2025: New Launcher application for MacOS and Linux, extended support to many backends for Mac and Nvidia L4T devices. Models: Added MLX-Audio, WAN 2.2. WebUI improvements and Python-based backends now ships portable python environments.
 - August 2025: MLX, MLX-VLM, Diffusers and llama.cpp are now supported on Mac M1/M2/M3+ chips ( with `development` suffix in the gallery ): https://github.com/mudler/LocalAI/pull/6049 https://github.com/mudler/LocalAI/pull/6119 https://github.com/mudler/LocalAI/pull/6121 https://github.com/mudler/LocalAI/pull/6060
 - July/August 2025: 🔍 [Object Detection](https://localai.io/features/object-detection/) added to the API featuring [rf-detr](https://github.com/roboflow/rf-detr)
 - July 2025: All backends migrated outside of the main binary. LocalAI is now more lightweight, small, and automatically downloads the required backend to run the model. [Read the release notes](https://github.com/mudler/LocalAI/releases/tag/v3.2.0)
@@ -229,7 +247,7 @@ Roadmap items: [List of issues](https://github.com/mudler/LocalAI/issues?q=is%3A
 - 🔍 [Object Detection](https://localai.io/features/object-detection/)
 - 📈 [Reranker API](https://localai.io/features/reranker/)
 - 🆕🖧 [P2P Inferencing](https://localai.io/features/distribute/)
-- [Agentic capabilities](https://github.com/mudler/LocalAGI)
+- 🆕🔌 [Model Context Protocol (MCP)](https://localai.io/docs/features/mcp/) - Agentic capabilities with external tools and [LocalAGI's Agentic capabilities](https://github.com/mudler/LocalAGI)
 - 🔊 Voice activity detection (Silero-VAD support)
 - 🌍 Integrated WebUI!
 
@@ -260,6 +278,7 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 | **piper** | Fast neural TTS system | CPU |
 | **kitten-tts** | Kitten TTS models | CPU |
 | **silero-vad** | Voice Activity Detection | CPU |
+| **neutts** | Text-to-speech with voice cloning | CUDA 12, ROCm, CPU |
 
 ### Image & Video Generation
 | Backend | Description | Acceleration Support |
@@ -281,7 +300,7 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 |-------------------|-------------------|------------------|
 | **NVIDIA CUDA 11** | llama.cpp, whisper, stablediffusion, diffusers, rerankers, bark, chatterbox | Nvidia hardware |
 | **NVIDIA CUDA 12** | All CUDA-compatible backends | Nvidia hardware |
-| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark | AMD Graphics |
+| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark, neutts | AMD Graphics |
 | **Intel oneAPI** | llama.cpp, whisper, stablediffusion, vllm, transformers, diffusers, rfdetr, rerankers, exllama2, coqui, kokoro, bark | Intel Arc, Intel iGPUs |
 | **Apple Metal** | llama.cpp, whisper, diffusers, MLX, MLX-VLM, bark-cpp | Apple M1/M2/M3+ |
 | **Vulkan** | llama.cpp, whisper, stablediffusion | Cross-platform GPUs |
@@ -298,6 +317,12 @@ WebUIs:
 - https://github.com/go-skynet/LocalAI-frontend
 - QA-Pilot(An interactive chat project that leverages LocalAI LLMs for rapid understanding and navigation of GitHub code repository) https://github.com/reid41/QA-Pilot
 
+Agentic Libraries:
+- https://github.com/mudler/cogito
+
+MCPs:
+- https://github.com/mudler/MCPs
+
 Model galleries
 - https://github.com/go-skynet/model-gallery
 

aio/cpu/vision.yaml

@@ -2,10 +2,10 @@ context_size: 4096
 f16: true
 backend: llama-cpp
 mmap: true
-mmproj: minicpm-v-2_6-mmproj-f16.gguf
+mmproj: minicpm-v-4_5-mmproj-f16.gguf
 name: gpt-4o
 parameters:
-  model: minicpm-v-2_6-Q4_K_M.gguf
+  model: minicpm-v-4_5-Q4_K_M.gguf
 stopwords:
 - <|im_end|>
 - <dummy32000>
@@ -42,9 +42,9 @@ template:
     <|im_start|>assistant
 
 download_files:
-- filename: minicpm-v-2_6-Q4_K_M.gguf
-  sha256: 3a4078d53b46f22989adbf998ce5a3fd090b6541f112d7e936eb4204a04100b1
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/ggml-model-Q4_K_M.gguf
-- filename: minicpm-v-2_6-mmproj-f16.gguf
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/mmproj-model-f16.gguf
-  sha256: 4485f68a0f1aa404c391e788ea88ea653c100d8e98fe572698f701e5809711fd
+- filename: minicpm-v-4_5-Q4_K_M.gguf
+  sha256: c1c3c33100b15b4caf7319acce4e23c0eb0ce1cbd12f70e8d24f05aa67b7512f
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/ggml-model-Q4_K_M.gguf
+- filename: minicpm-v-4_5-mmproj-f16.gguf
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/mmproj-model-f16.gguf
+  sha256: 7a7225a32e8d453aaa3d22d8c579b5bf833c253f784cdb05c99c9a76fd616df8

aio/gpu-8g/vision.yaml

@@ -2,10 +2,10 @@ context_size: 4096
 backend: llama-cpp
 f16: true
 mmap: true
-mmproj: minicpm-v-2_6-mmproj-f16.gguf
+mmproj: minicpm-v-4_5-mmproj-f16.gguf
 name: gpt-4o
 parameters:
-  model: minicpm-v-2_6-Q4_K_M.gguf
+  model: minicpm-v-4_5-Q4_K_M.gguf
 stopwords:
 - <|im_end|>
 - <dummy32000>
@@ -42,9 +42,9 @@ template:
     <|im_start|>assistant
 
 download_files:
-- filename: minicpm-v-2_6-Q4_K_M.gguf
-  sha256: 3a4078d53b46f22989adbf998ce5a3fd090b6541f112d7e936eb4204a04100b1
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/ggml-model-Q4_K_M.gguf
-- filename: minicpm-v-2_6-mmproj-f16.gguf
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/mmproj-model-f16.gguf
-  sha256: 4485f68a0f1aa404c391e788ea88ea653c100d8e98fe572698f701e5809711fd
+- filename: minicpm-v-4_5-Q4_K_M.gguf
+  sha256: c1c3c33100b15b4caf7319acce4e23c0eb0ce1cbd12f70e8d24f05aa67b7512f
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/ggml-model-Q4_K_M.gguf
+- filename: minicpm-v-4_5-mmproj-f16.gguf
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/mmproj-model-f16.gguf
+  sha256: 7a7225a32e8d453aaa3d22d8c579b5bf833c253f784cdb05c99c9a76fd616df8

aio/intel/vision.yaml

@@ -2,10 +2,10 @@ context_size: 4096
 backend: llama-cpp
 f16: true
 mmap: true
-mmproj: minicpm-v-2_6-mmproj-f16.gguf
+mmproj: minicpm-v-4_5-mmproj-f16.gguf
 name: gpt-4o
 parameters:
-  model: minicpm-v-2_6-Q4_K_M.gguf
+  model: minicpm-v-4_5-Q4_K_M.gguf
 stopwords:
 - <|im_end|>
 - <dummy32000>
@@ -43,9 +43,9 @@ template:
 
 
 download_files:
-- filename: minicpm-v-2_6-Q4_K_M.gguf
-  sha256: 3a4078d53b46f22989adbf998ce5a3fd090b6541f112d7e936eb4204a04100b1
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/ggml-model-Q4_K_M.gguf
-- filename: minicpm-v-2_6-mmproj-f16.gguf
-  uri: huggingface://openbmb/MiniCPM-V-2_6-gguf/mmproj-model-f16.gguf
-  sha256: 4485f68a0f1aa404c391e788ea88ea653c100d8e98fe572698f701e5809711fd
+- filename: minicpm-v-4_5-Q4_K_M.gguf
+  sha256: c1c3c33100b15b4caf7319acce4e23c0eb0ce1cbd12f70e8d24f05aa67b7512f
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/ggml-model-Q4_K_M.gguf
+- filename: minicpm-v-4_5-mmproj-f16.gguf
+  uri: huggingface://openbmb/MiniCPM-V-4_5-gguf/mmproj-model-f16.gguf
+  sha256: 7a7225a32e8d453aaa3d22d8c579b5bf833c253f784cdb05c99c9a76fd616df8

backend/Dockerfile.llama-cpp

@@ -197,7 +197,7 @@ EOT
 
 
 # Copy libraries using a script to handle architecture differences
-RUN make -C /LocalAI/backend/cpp/llama-cpp package
+RUN make -BC /LocalAI/backend/cpp/llama-cpp package
 
 
 FROM scratch

backend/Dockerfile.python

@@ -28,7 +28,7 @@ RUN apt-get update && \
         curl python3-pip \
         python-is-python3 \
         python3-dev llvm \
-        python3-venv make && \
+        python3-venv make cmake && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/* && \
     pip install --upgrade pip

backend/backend.proto

@@ -154,6 +154,10 @@ message PredictOptions {
   repeated string Videos = 45;
   repeated string Audios = 46;
   string CorrelationId = 47;
+  string Tools = 48;  // JSON array of available tools/functions for tool calling
+  string ToolChoice = 49;  // JSON string or object specifying tool choice behavior
+  int32 Logprobs = 50;  // Number of top logprobs to return (maps to OpenAI logprobs parameter)
+  int32 TopLogprobs = 51;  // Number of top logprobs to return per token (maps to OpenAI top_logprobs parameter)
 }
 
 // The response message containing the result
@@ -164,6 +168,7 @@ message Reply {
   double timing_prompt_processing = 4;
   double timing_token_generation = 5;
   bytes audio = 6;
+  bytes logprobs = 7;  // JSON-encoded logprobs data matching OpenAI format
 }
 
 message GrammarTrigger {
@@ -276,6 +281,7 @@ message TranscriptRequest {
   string language = 3;
   uint32 threads = 4;
   bool translate = 5;
+  bool diarize = 6;
 }
 
 message TranscriptResult {
@@ -305,7 +311,7 @@ message GenerateImageRequest {
   // Diffusers
   string EnableParameters = 10;
   int32 CLIPSkip = 11;
-  
+
   // Reference images for models that support them (e.g., Flux Kontext)
   repeated string ref_images = 12;
 }
@@ -381,6 +387,11 @@ message StatusResponse {
 message Message {
   string role = 1;
   string content = 2;
+  // Optional fields for OpenAI-compatible message format
+  string name = 3;                    // Tool name (for tool messages)
+  string tool_call_id = 4;            // Tool call ID (for tool messages)
+  string reasoning_content = 5;       // Reasoning content (for thinking models)
+  string tool_calls = 6;              // Tool calls as JSON string (for assistant messages with tool calls)
 }
 
 message DetectOptions {

backend/cpp/llama-cpp/CMakeLists.txt

@@ -57,7 +57,7 @@ add_library(hw_grpc_proto
   ${hw_proto_srcs}
   ${hw_proto_hdrs} )
 
-add_executable(${TARGET} grpc-server.cpp utils.hpp json.hpp httplib.h)
+add_executable(${TARGET} grpc-server.cpp json.hpp httplib.h)
 
 target_include_directories(${TARGET} PRIVATE ../llava)
 target_include_directories(${TARGET} PRIVATE ${CMAKE_SOURCE_DIR})

backend/cpp/llama-cpp/Makefile

@@ -1,5 +1,5 @@
 
-LLAMA_VERSION?=3de008208b9b8a33f49f979097a99b4d59e6e521
+LLAMA_VERSION?=583cb83416467e8abf9b37349dcf1f6a0083745a
 LLAMA_REPO?=https://github.com/ggerganov/llama.cpp
 
 CMAKE_ARGS?=
@@ -14,7 +14,7 @@ CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF -DLLAMA_CURL=OFF
 
 CURRENT_MAKEFILE_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))
 ifeq ($(NATIVE),false)
-	CMAKE_ARGS+=-DGGML_NATIVE=OFF
+	CMAKE_ARGS+=-DGGML_NATIVE=OFF -DLLAMA_OPENSSL=OFF
 endif
 # If build type is cublas, then we set -DGGML_CUDA=ON to CMAKE_ARGS automatically
 ifeq ($(BUILD_TYPE),cublas)

backend/cpp/llama-cpp/prepare.sh

@@ -9,10 +9,13 @@ done
 
 set -e
 
+for file in $(ls llama.cpp/tools/server/); do
+    cp -rfv llama.cpp/tools/server/$file llama.cpp/tools/grpc-server/
+done
+
 cp -r CMakeLists.txt llama.cpp/tools/grpc-server/
 cp -r grpc-server.cpp llama.cpp/tools/grpc-server/
 cp -rfv llama.cpp/vendor/nlohmann/json.hpp llama.cpp/tools/grpc-server/
-cp -rfv llama.cpp/tools/server/utils.hpp llama.cpp/tools/grpc-server/
 cp -rfv llama.cpp/vendor/cpp-httplib/httplib.h llama.cpp/tools/grpc-server/
 
 set +e

backend/go/stablediffusion-ggml/.gitignore

@@ -1,4 +1,6 @@
 package/
 sources/
+.cache/
+build/
 libgosd.so
 stablediffusion-ggml

backend/go/stablediffusion-ggml/Makefile

@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)
 
 # stablediffusion.cpp (ggml)
 STABLEDIFFUSION_GGML_REPO?=https://github.com/leejet/stable-diffusion.cpp
-STABLEDIFFUSION_GGML_VERSION?=4c6475f9176bf99271ccf5a2817b30a490b83db0
+STABLEDIFFUSION_GGML_VERSION?=0ebe6fe118f125665939b27c89f34ed38716bff8
 
 CMAKE_ARGS+=-DGGML_MAX_NAME=128
 

backend/go/stablediffusion-ggml/gosd.cpp

@@ -4,17 +4,11 @@
 #include <stdio.h>
 #include <string.h>
 #include <time.h>
-#include <iostream>
-#include <random>
 #include <string>
 #include <vector>
 #include <filesystem>
 #include "gosd.h"
 
-// #include "preprocessing.hpp"
-#include "flux.hpp"
-#include "stable-diffusion.h"
-
 #define STB_IMAGE_IMPLEMENTATION
 #define STB_IMAGE_STATIC
 #include "stb_image.h"
@@ -29,7 +23,7 @@
 
 // Names of the sampler method, same order as enum sample_method in stable-diffusion.h
 const char* sample_method_str[] = {
-    "euler_a",
+    "default",
     "euler",
     "heun",
     "dpm2",
@@ -41,19 +35,27 @@ const char* sample_method_str[] = {
     "lcm",
     "ddim_trailing",
     "tcd",
+    "euler_a",
 };
 
+static_assert(std::size(sample_method_str) == SAMPLE_METHOD_COUNT, "sample method mismatch");
+
 // Names of the sigma schedule overrides, same order as sample_schedule in stable-diffusion.h
-const char* schedule_str[] = {
+const char* schedulers[] = {
     "default",
     "discrete",
     "karras",
     "exponential",
     "ays",
     "gits",
+    "smoothstep",
 };
 
+static_assert(std::size(schedulers) == SCHEDULE_COUNT, "schedulers mismatch");
+
 sd_ctx_t* sd_c;
+// Moved from the context (load time) to generation time params
+scheduler_t scheduler = scheduler_t::DEFAULT;
 
 sample_method_t sample_method;
 
@@ -105,7 +107,7 @@ int load_model(const char *model, char *model_path, char* options[], int threads
     const char *clip_g_path  = "";
     const char *t5xxl_path  = "";
     const char *vae_path  = "";
-    const char *scheduler = "";
+    const char *scheduler_str = "";
     const char *sampler = "";
     char *lora_dir = model_path;
     bool lora_dir_allocated = false;
@@ -133,7 +135,7 @@ int load_model(const char *model, char *model_path, char* options[], int threads
             vae_path = optval;
         }
         if (!strcmp(optname, "scheduler")) {
-            scheduler = optval;
+            scheduler_str = optval;
         }
         if (!strcmp(optname, "sampler")) {
             sampler = optval;
@@ -166,26 +168,17 @@ int load_model(const char *model, char *model_path, char* options[], int threads
     }
     if (sample_method_found == -1) {
         fprintf(stderr, "Invalid sample method, default to EULER_A!\n");
-        sample_method_found = EULER_A;
+        sample_method_found = sample_method_t::SAMPLE_METHOD_DEFAULT;
     }
     sample_method = (sample_method_t)sample_method_found;
 
-    int schedule_found            = -1;
     for (int d = 0; d < SCHEDULE_COUNT; d++) {
-        if (!strcmp(scheduler, schedule_str[d])) {
-            schedule_found = d;
-                fprintf (stderr, "Found scheduler: %s\n", scheduler);
-
+        if (!strcmp(scheduler_str, schedulers[d])) {
+            scheduler = (scheduler_t)d;
+            fprintf (stderr, "Found scheduler: %s\n", scheduler_str);
         }
     }
 
-    if (schedule_found == -1) {
-        fprintf (stderr, "Invalid scheduler! using DEFAULT\n");
-        schedule_found = DEFAULT;
-    }
-
-    schedule_t schedule = (schedule_t)schedule_found;
-
     fprintf (stderr, "Creating context\n");
     sd_ctx_params_t ctx_params;
     sd_ctx_params_init(&ctx_params);
@@ -199,13 +192,10 @@ int load_model(const char *model, char *model_path, char* options[], int threads
     ctx_params.control_net_path = "";
     ctx_params.lora_model_dir = lora_dir;
     ctx_params.embedding_dir = "";
-    ctx_params.stacked_id_embed_dir = "";
     ctx_params.vae_decode_only = false;
-    ctx_params.vae_tiling = false;
     ctx_params.free_params_immediately = false;
     ctx_params.n_threads = threads;
     ctx_params.rng_type = STD_DEFAULT_RNG;
-    ctx_params.schedule = schedule;
     sd_ctx_t* sd_ctx = new_sd_ctx(&ctx_params);
 
     if (sd_ctx == NULL) {
@@ -228,7 +218,49 @@ int load_model(const char *model, char *model_path, char* options[], int threads
     return 0;
 }
 
-int gen_image(char *text, char *negativeText, int width, int height, int steps, int64_t seed, char *dst, float cfg_scale, char *src_image, float strength, char *mask_image, char **ref_images, int ref_images_count) {
+void sd_tiling_params_set_enabled(sd_tiling_params_t *params, bool enabled) {
+    params->enabled = enabled;
+}
+
+void sd_tiling_params_set_tile_sizes(sd_tiling_params_t *params, int tile_size_x, int tile_size_y) {
+    params->tile_size_x = tile_size_x;
+    params->tile_size_y = tile_size_y;
+}
+
+void sd_tiling_params_set_rel_sizes(sd_tiling_params_t *params, float rel_size_x, float rel_size_y) {
+    params->rel_size_x = rel_size_x;
+    params->rel_size_y = rel_size_y;
+}
+
+void sd_tiling_params_set_target_overlap(sd_tiling_params_t *params, float target_overlap) {
+    params->target_overlap = target_overlap;
+}
+
+sd_tiling_params_t* sd_img_gen_params_get_vae_tiling_params(sd_img_gen_params_t *params) {
+    return &params->vae_tiling_params;
+}
+
+sd_img_gen_params_t* sd_img_gen_params_new(void) {
+    sd_img_gen_params_t *params = (sd_img_gen_params_t *)std::malloc(sizeof(sd_img_gen_params_t));
+    sd_img_gen_params_init(params);
+    return params;
+}
+
+void sd_img_gen_params_set_prompts(sd_img_gen_params_t *params, const char *prompt, const char *negative_prompt) {
+    params->prompt = prompt;
+    params->negative_prompt = negative_prompt;
+}
+
+void sd_img_gen_params_set_dimensions(sd_img_gen_params_t *params, int width, int height) {
+    params->width = width;
+    params->height = height;
+}
+
+void sd_img_gen_params_set_seed(sd_img_gen_params_t *params, int64_t seed) {
+    params->seed = seed;
+}
+
+int gen_image(sd_img_gen_params_t *p, int steps, char *dst, float cfg_scale, char *src_image, float strength, char *mask_image, char **ref_images, int ref_images_count) {
 
     sd_image_t* results;
 
@@ -236,20 +268,15 @@ int gen_image(char *text, char *negativeText, int width, int height, int steps,
 
     fprintf (stderr, "Generating image\n");
 
-    sd_img_gen_params_t p;
-    sd_img_gen_params_init(&p);
+    p->sample_params.guidance.txt_cfg = cfg_scale;
+    p->sample_params.guidance.slg.layers = skip_layers.data();
+    p->sample_params.guidance.slg.layer_count = skip_layers.size();
+    p->sample_params.sample_method = sample_method;
+    p->sample_params.sample_steps = steps;
+    p->sample_params.scheduler = scheduler;
 
-    p.prompt = text;
-    p.negative_prompt = negativeText;
-    p.guidance.txt_cfg = cfg_scale;
-    p.guidance.slg.layers = skip_layers.data();
-    p.guidance.slg.layer_count = skip_layers.size();
-    p.width = width;
-    p.height = height;
-    p.sample_method = sample_method;
-    p.sample_steps = steps;
-    p.seed = seed;
-    p.input_id_images_path = "";
+    int width = p->width;
+    int height = p->height;
 
     // Handle input image for img2img
     bool has_input_image = (src_image != NULL && strlen(src_image) > 0);
@@ -298,13 +325,13 @@ int gen_image(char *text, char *negativeText, int width, int height, int steps,
             input_image_buffer = resized_image_buffer;
         }
 
-        p.init_image = {(uint32_t)width, (uint32_t)height, 3, input_image_buffer};
-        p.strength = strength;
+        p->init_image = {(uint32_t)width, (uint32_t)height, 3, input_image_buffer};
+        p->strength = strength;
         fprintf(stderr, "Using img2img with strength: %.2f\n", strength);
     } else {
         // No input image, use empty image for text-to-image
-        p.init_image = {(uint32_t)width, (uint32_t)height, 3, NULL};
-        p.strength = 0.0f;
+        p->init_image = {(uint32_t)width, (uint32_t)height, 3, NULL};
+        p->strength = 0.0f;
     }
 
     // Handle mask image for inpainting
@@ -344,12 +371,12 @@ int gen_image(char *text, char *negativeText, int width, int height, int steps,
             mask_image_buffer = resized_mask_buffer;
         }
 
-        p.mask_image = {(uint32_t)width, (uint32_t)height, 1, mask_image_buffer};
+        p->mask_image = {(uint32_t)width, (uint32_t)height, 1, mask_image_buffer};
         fprintf(stderr, "Using inpainting with mask\n");
     } else {
         // No mask image, create default full mask
         default_mask_image_vec.resize(width * height, 255);
-        p.mask_image = {(uint32_t)width, (uint32_t)height, 1, default_mask_image_vec.data()};
+        p->mask_image = {(uint32_t)width, (uint32_t)height, 1, default_mask_image_vec.data()};
     }
 
     // Handle reference images
@@ -407,13 +434,15 @@ int gen_image(char *text, char *negativeText, int width, int height, int steps,
         }
 
         if (!ref_images_vec.empty()) {
-            p.ref_images = ref_images_vec.data();
-            p.ref_images_count = ref_images_vec.size();
+            p->ref_images = ref_images_vec.data();
+            p->ref_images_count = ref_images_vec.size();
             fprintf(stderr, "Using %zu reference images\n", ref_images_vec.size());
         }
     }
 
-    results = generate_image(sd_c, &p);
+    results = generate_image(sd_c, p);
+
+    std::free(p);
 
     if (results == NULL) {
         fprintf (stderr, "NO results\n");

backend/go/stablediffusion-ggml/gosd.go

@@ -22,7 +22,18 @@ type SDGGML struct {
 
 var (
 	LoadModel func(model, model_apth string, options []uintptr, threads int32, diff int) int
-	GenImage  func(text, negativeText string, width, height, steps int, seed int64, dst string, cfgScale float32, srcImage string, strength float32, maskImage string, refImages []string, refImagesCount int) int
+	GenImage  func(params uintptr, steps int, dst string, cfgScale float32, srcImage string, strength float32, maskImage string, refImages []string, refImagesCount int) int
+
+	TilingParamsSetEnabled       func(params uintptr, enabled bool)
+	TilingParamsSetTileSizes     func(params uintptr, tileSizeX int, tileSizeY int)
+	TilingParamsSetRelSizes      func(params uintptr, relSizeX float32, relSizeY float32)
+	TilingParamsSetTargetOverlap func(params uintptr, targetOverlap float32)
+
+	ImgGenParamsNew                func() uintptr
+	ImgGenParamsSetPrompts         func(params uintptr, prompt string, negativePrompt string)
+	ImgGenParamsSetDimensions      func(params uintptr, width int, height int)
+	ImgGenParamsSetSeed            func(params uintptr, seed int64)
+	ImgGenParamsGetVaeTilingParams func(params uintptr) uintptr
 )
 
 // Copied from Purego internal/strings
@@ -120,7 +131,15 @@ func (sd *SDGGML) GenerateImage(opts *pb.GenerateImageRequest) error {
 	// Default strength for img2img (0.75 is a good default)
 	strength := float32(0.75)
 
-	ret := GenImage(t, negative, int(opts.Width), int(opts.Height), int(opts.Step), int64(opts.Seed), dst, sd.cfgScale, srcImage, strength, maskImage, refImages, refImagesCount)
+	// free'd by GenImage
+	p := ImgGenParamsNew()
+	ImgGenParamsSetPrompts(p, t, negative)
+	ImgGenParamsSetDimensions(p, int(opts.Width), int(opts.Height))
+	ImgGenParamsSetSeed(p, int64(opts.Seed))
+	vaep := ImgGenParamsGetVaeTilingParams(p)
+	TilingParamsSetEnabled(vaep, false)
+
+	ret := GenImage(p, int(opts.Step), dst, sd.cfgScale, srcImage, strength, maskImage, refImages, refImagesCount)
 	if ret != 0 {
 		return fmt.Errorf("inference failed")
 	}

backend/go/stablediffusion-ggml/gosd.h

@@ -1,8 +1,23 @@
+#include <cstdint>
+#include "stable-diffusion.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif
+
+void sd_tiling_params_set_enabled(sd_tiling_params_t *params, bool enabled);
+void sd_tiling_params_set_tile_sizes(sd_tiling_params_t *params, int tile_size_x, int tile_size_y);
+void sd_tiling_params_set_rel_sizes(sd_tiling_params_t *params, float rel_size_x, float rel_size_y);
+void sd_tiling_params_set_target_overlap(sd_tiling_params_t *params, float target_overlap);
+sd_tiling_params_t* sd_img_gen_params_get_vae_tiling_params(sd_img_gen_params_t *params);
+
+sd_img_gen_params_t* sd_img_gen_params_new(void);
+void sd_img_gen_params_set_prompts(sd_img_gen_params_t *params, const char *prompt, const char *negative_prompt);
+void sd_img_gen_params_set_dimensions(sd_img_gen_params_t *params, int width, int height);
+void sd_img_gen_params_set_seed(sd_img_gen_params_t *params, int64_t seed);
+
 int load_model(const char *model, char *model_path, char* options[], int threads, int diffusionModel);
-int gen_image(char *text, char *negativeText, int width, int height, int steps, int64_t seed, char *dst, float cfg_scale, char *src_image, float strength, char *mask_image, char **ref_images, int ref_images_count);
+int gen_image(sd_img_gen_params_t *p, int steps, char *dst, float cfg_scale, char *src_image, float strength, char *mask_image, char **ref_images, int ref_images_count);
 #ifdef __cplusplus
 }
 #endif

backend/go/stablediffusion-ggml/main.go

@@ -11,14 +11,35 @@ var (
 	addr = flag.String("addr", "localhost:50051", "the address to connect to")
 )
 
+type LibFuncs struct {
+	FuncPtr any
+	Name    string
+}
+
 func main() {
 	gosd, err := purego.Dlopen("./libgosd.so", purego.RTLD_NOW|purego.RTLD_GLOBAL)
 	if err != nil {
 		panic(err)
 	}
 
-	purego.RegisterLibFunc(&LoadModel, gosd, "load_model")
-	purego.RegisterLibFunc(&GenImage, gosd, "gen_image")
+	libFuncs := []LibFuncs{
+		{&LoadModel, "load_model"},
+		{&GenImage, "gen_image"},
+		{&TilingParamsSetEnabled, "sd_tiling_params_set_enabled"},
+		{&TilingParamsSetTileSizes, "sd_tiling_params_set_tile_sizes"},
+		{&TilingParamsSetRelSizes, "sd_tiling_params_set_rel_sizes"},
+		{&TilingParamsSetTargetOverlap, "sd_tiling_params_set_target_overlap"},
+
+		{&ImgGenParamsNew, "sd_img_gen_params_new"},
+		{&ImgGenParamsSetPrompts, "sd_img_gen_params_set_prompts"},
+		{&ImgGenParamsSetDimensions, "sd_img_gen_params_set_dimensions"},
+		{&ImgGenParamsSetSeed, "sd_img_gen_params_set_seed"},
+		{&ImgGenParamsGetVaeTilingParams, "sd_img_gen_params_get_vae_tiling_params"},
+	}
+
+	for _, lf := range libFuncs {
+		purego.RegisterLibFunc(lf.FuncPtr, gosd, lf.Name)
+	}
 
 	flag.Parse()
 

backend/go/whisper/Makefile

@@ -8,7 +8,8 @@ JOBS?=$(shell nproc --ignore=1)
 
 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggml-org/whisper.cpp
-WHISPER_CPP_VERSION?=7745fcf32846006128f16de429cfe1677c963b30
+WHISPER_CPP_VERSION?=19ceec8eac980403b714d603e5ca31653cd42a3f
+SO_TARGET?=libgowhisper.so
 
 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF
 
@@ -57,15 +58,18 @@ sources/whisper.cpp:
 	git checkout $(WHISPER_CPP_VERSION) && \
 	git submodule update --init --recursive --depth 1 --single-branch
 
-libgowhisper.so: sources/whisper.cpp CMakeLists.txt gowhisper.cpp gowhisper.h
-	mkdir -p build && \
-	cd build && \
-	cmake .. $(CMAKE_ARGS) && \
-	cmake --build . --config Release -j$(JOBS) && \
-	cd .. && \
-	mv build/libgowhisper.so ./
+# Detect OS
+UNAME_S := $(shell uname -s)
 
-whisper: main.go gowhisper.go libgowhisper.so
+# Only build CPU variants on Linux
+ifeq ($(UNAME_S),Linux)
+	VARIANT_TARGETS = libgowhisper-avx.so libgowhisper-avx2.so libgowhisper-avx512.so libgowhisper-fallback.so
+else
+	# On non-Linux (e.g., Darwin), build only fallback variant
+	VARIANT_TARGETS = libgowhisper-fallback.so
+endif
+
+whisper: main.go gowhisper.go $(VARIANT_TARGETS)
 	CGO_ENABLED=0 $(GOCMD) build -tags "$(GO_TAGS)" -o whisper ./
 
 package: whisper
@@ -73,5 +77,46 @@ package: whisper
 
 build: package
 
-clean:
-	rm -rf libgowhisper.o build whisper
+clean: purge
+	rm -rf libgowhisper*.so sources/whisper.cpp whisper
+
+purge:
+	rm -rf build*
+
+# Build all variants (Linux only)
+ifeq ($(UNAME_S),Linux)
+libgowhisper-avx.so: sources/whisper.cpp
+	$(MAKE) purge
+	$(info ${GREEN}I whisper build info:avx${RESET})
+	SO_TARGET=libgowhisper-avx.so CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off" $(MAKE) libgowhisper-custom
+	rm -rfv build*
+
+libgowhisper-avx2.so: sources/whisper.cpp
+	$(MAKE) purge
+	$(info ${GREEN}I whisper build info:avx2${RESET})
+	SO_TARGET=libgowhisper-avx2.so CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=on -DGGML_AVX512=off -DGGML_FMA=on -DGGML_F16C=on" $(MAKE) libgowhisper-custom
+	rm -rfv build*
+
+libgowhisper-avx512.so: sources/whisper.cpp
+	$(MAKE) purge
+	$(info ${GREEN}I whisper build info:avx512${RESET})
+	SO_TARGET=libgowhisper-avx512.so CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=off -DGGML_AVX512=on -DGGML_FMA=on -DGGML_F16C=on" $(MAKE) libgowhisper-custom
+	rm -rfv build*
+endif
+
+# Build fallback variant (all platforms)
+libgowhisper-fallback.so: sources/whisper.cpp
+	$(MAKE) purge
+	$(info ${GREEN}I whisper build info:fallback${RESET})
+	SO_TARGET=libgowhisper-fallback.so CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=off -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off" $(MAKE) libgowhisper-custom
+	rm -rfv build*
+
+libgowhisper-custom: CMakeLists.txt gowhisper.cpp gowhisper.h
+	mkdir -p build-$(SO_TARGET) && \
+	cd build-$(SO_TARGET) && \
+	cmake .. $(CMAKE_ARGS) && \
+	cmake --build . --config Release -j$(JOBS) && \
+	cd .. && \
+	mv build-$(SO_TARGET)/libgowhisper.so ./$(SO_TARGET)
+
+all: whisper package

backend/go/whisper/gowhisper.cpp

@@ -7,34 +7,35 @@ static struct whisper_vad_context *vctx;
 static struct whisper_context *ctx;
 static std::vector<float> flat_segs;
 
-static void ggml_log_cb(enum ggml_log_level level, const char* log, void* data) {
-    const char* level_str;
+static void ggml_log_cb(enum ggml_log_level level, const char *log,
+                        void *data) {
+  const char *level_str;
 
-    if (!log) {
-        return;
-    }
+  if (!log) {
+    return;
+  }
 
-    switch (level) {
-        case GGML_LOG_LEVEL_DEBUG:
-            level_str = "DEBUG";
-            break;
-        case GGML_LOG_LEVEL_INFO:
-            level_str = "INFO";
-            break;
-        case GGML_LOG_LEVEL_WARN:
-            level_str = "WARN";
-            break;
-        case GGML_LOG_LEVEL_ERROR:
-            level_str = "ERROR";
-            break;
-        default: /* Potential future-proofing */
-            level_str = "?????";
-            break;
-    }
+  switch (level) {
+  case GGML_LOG_LEVEL_DEBUG:
+    level_str = "DEBUG";
+    break;
+  case GGML_LOG_LEVEL_INFO:
+    level_str = "INFO";
+    break;
+  case GGML_LOG_LEVEL_WARN:
+    level_str = "WARN";
+    break;
+  case GGML_LOG_LEVEL_ERROR:
+    level_str = "ERROR";
+    break;
+  default: /* Potential future-proofing */
+    level_str = "?????";
+    break;
+  }
 
-    fprintf(stderr, "[%-5s] ", level_str);
-    fputs(log, stderr);
-    fflush(stderr);
+  fprintf(stderr, "[%-5s] ", level_str);
+  fputs(log, stderr);
+  fflush(stderr);
 }
 
 int load_model(const char *const model_path) {
@@ -105,8 +106,8 @@ int vad(float pcmf32[], size_t pcmf32_len, float **segs_out,
   return 0;
 }
 
-int transcribe(uint32_t threads, char *lang, bool translate, float pcmf32[],
-               size_t pcmf32_len, size_t *segs_out_len) {
+int transcribe(uint32_t threads, char *lang, bool translate, bool tdrz,
+               float pcmf32[], size_t pcmf32_len, size_t *segs_out_len) {
   whisper_full_params wparams =
       whisper_full_default_params(WHISPER_SAMPLING_GREEDY);
 
@@ -120,6 +121,9 @@ int transcribe(uint32_t threads, char *lang, bool translate, float pcmf32[],
   wparams.translate = translate;
   wparams.debug_mode = true;
   wparams.print_progress = true;
+  wparams.tdrz_enable = tdrz;
+
+  fprintf(stderr, "info: Enable tdrz: %d\n", tdrz);
 
   if (whisper_full(ctx, wparams, pcmf32, pcmf32_len)) {
     fprintf(stderr, "error: transcription failed\n");
@@ -144,3 +148,7 @@ int n_tokens(int i) { return whisper_full_n_tokens(ctx, i); }
 int32_t get_token_id(int i, int j) {
   return whisper_full_get_token_id(ctx, i, j);
 }
+
+bool get_segment_speaker_turn_next(int i) {
+  return whisper_full_get_segment_speaker_turn_next(ctx, i);
+}

backend/go/whisper/gowhisper.go

@@ -14,15 +14,16 @@ import (
 )
 
 var (
-	CppLoadModel       func(modelPath string) int
-	CppLoadModelVAD    func(modelPath string) int
-	CppVAD             func(pcmf32 []float32, pcmf32Size uintptr, segsOut unsafe.Pointer, segsOutLen unsafe.Pointer) int
-	CppTranscribe      func(threads uint32, lang string, translate bool, pcmf32 []float32, pcmf32Len uintptr, segsOutLen unsafe.Pointer) int
-	CppGetSegmentText  func(i int) string
-	CppGetSegmentStart func(i int) int64
-	CppGetSegmentEnd   func(i int) int64
-	CppNTokens         func(i int) int
-	CppGetTokenID      func(i int, j int) int
+	CppLoadModel                 func(modelPath string) int
+	CppLoadModelVAD              func(modelPath string) int
+	CppVAD                       func(pcmf32 []float32, pcmf32Size uintptr, segsOut unsafe.Pointer, segsOutLen unsafe.Pointer) int
+	CppTranscribe                func(threads uint32, lang string, translate bool, diarize bool, pcmf32 []float32, pcmf32Len uintptr, segsOutLen unsafe.Pointer) int
+	CppGetSegmentText            func(i int) string
+	CppGetSegmentStart           func(i int) int64
+	CppGetSegmentEnd             func(i int) int64
+	CppNTokens                   func(i int) int
+	CppGetTokenID                func(i int, j int) int
+	CppGetSegmentSpeakerTurnNext func(i int) bool
 )
 
 type Whisper struct {
@@ -122,7 +123,7 @@ func (w *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (pb.TranscriptR
 	segsLen := uintptr(0xdeadbeef)
 	segsLenPtr := unsafe.Pointer(&segsLen)
 
-	if ret := CppTranscribe(opts.Threads, opts.Language, opts.Translate, data, uintptr(len(data)), segsLenPtr); ret != 0 {
+	if ret := CppTranscribe(opts.Threads, opts.Language, opts.Translate, opts.Diarize, data, uintptr(len(data)), segsLenPtr); ret != 0 {
 		return pb.TranscriptResult{}, fmt.Errorf("Failed Transcribe")
 	}
 
@@ -134,6 +135,10 @@ func (w *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (pb.TranscriptR
 		txt := strings.Clone(CppGetSegmentText(i))
 		tokens := make([]int32, CppNTokens(i))
 
+		if opts.Diarize && CppGetSegmentSpeakerTurnNext(i) {
+			txt += " [SPEAKER_TURN]"
+		}
+
 		for j := range tokens {
 			tokens[j] = int32(CppGetTokenID(i, j))
 		}
@@ -151,6 +156,6 @@ func (w *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (pb.TranscriptR
 
 	return pb.TranscriptResult{
 		Segments: segments,
-		Text: strings.TrimSpace(text),
+		Text:     strings.TrimSpace(text),
 	}, nil
 }

backend/go/whisper/gowhisper.h

@@ -6,11 +6,12 @@ int load_model(const char *const model_path);
 int load_model_vad(const char *const model_path);
 int vad(float pcmf32[], size_t pcmf32_size, float **segs_out,
         size_t *segs_out_len);
-int transcribe(uint32_t threads, char *lang, bool translate, float pcmf32[],
-               size_t pcmf32_len, size_t *segs_out_len);
+int transcribe(uint32_t threads, char *lang, bool translate, bool tdrz,
+               float pcmf32[], size_t pcmf32_len, size_t *segs_out_len);
 const char *get_segment_text(int i);
 int64_t get_segment_t0(int i);
 int64_t get_segment_t1(int i);
 int n_tokens(int i);
 int32_t get_token_id(int i, int j);
+bool get_segment_speaker_turn_next(int i);
 }

backend/go/whisper/main.go

@@ -3,6 +3,7 @@ package main
 // Note: this is started internally by LocalAI and a server is allocated for each model
 import (
 	"flag"
+	"os"
 
 	"github.com/ebitengine/purego"
 	grpc "github.com/mudler/LocalAI/pkg/grpc"
@@ -18,7 +19,13 @@ type LibFuncs struct {
 }
 
 func main() {
-	gosd, err := purego.Dlopen("./libgowhisper.so", purego.RTLD_NOW|purego.RTLD_GLOBAL)
+	// Get library name from environment variable, default to fallback
+	libName := os.Getenv("WHISPER_LIBRARY")
+	if libName == "" {
+		libName = "./libgowhisper-fallback.so"
+	}
+
+	gosd, err := purego.Dlopen(libName, purego.RTLD_NOW|purego.RTLD_GLOBAL)
 	if err != nil {
 		panic(err)
 	}
@@ -33,6 +40,7 @@ func main() {
 		{&CppGetSegmentEnd, "get_segment_t1"},
 		{&CppNTokens, "n_tokens"},
 		{&CppGetTokenID, "get_token_id"},
+		{&CppGetSegmentSpeakerTurnNext, "get_segment_speaker_turn_next"},
 	}
 
 	for _, lf := range libFuncs {

backend/go/whisper/package.sh

@@ -10,7 +10,8 @@ CURDIR=$(dirname "$(realpath $0)")
 # Create lib directory
 mkdir -p $CURDIR/package/lib
 
-cp -avf $CURDIR/whisper $CURDIR/libgowhisper.so $CURDIR/package/
+cp -avf $CURDIR/whisper $CURDIR/package/
+cp -fv $CURDIR/libgowhisper-*.so $CURDIR/package/
 cp -fv $CURDIR/run.sh $CURDIR/package/
 
 # Detect architecture and copy appropriate libraries

backend/go/whisper/run.sh

@@ -1,14 +1,52 @@
 #!/bin/bash
 set -ex
 
+# Get the absolute current dir where the script is located
 CURDIR=$(dirname "$(realpath $0)")
 
+cd /
+
+echo "CPU info:"
+if [ "$(uname)" != "Darwin" ]; then
+	grep -e "model\sname" /proc/cpuinfo | head -1
+	grep -e "flags" /proc/cpuinfo | head -1
+fi
+
+LIBRARY="$CURDIR/libgowhisper-fallback.so"
+
+if [ "$(uname)" != "Darwin" ]; then
+	if grep -q -e "\savx\s" /proc/cpuinfo ; then
+		echo "CPU:    AVX    found OK"
+		if [ -e $CURDIR/libgowhisper-avx.so ]; then
+			LIBRARY="$CURDIR/libgowhisper-avx.so"
+		fi
+	fi
+
+	if grep -q -e "\savx2\s" /proc/cpuinfo ; then
+		echo "CPU:    AVX2   found OK"
+		if [ -e $CURDIR/libgowhisper-avx2.so ]; then
+			LIBRARY="$CURDIR/libgowhisper-avx2.so"
+		fi
+	fi
+
+	# Check avx 512
+	if grep -q -e "\savx512f\s" /proc/cpuinfo ; then
+		echo "CPU:    AVX512F found OK"
+		if [ -e $CURDIR/libgowhisper-avx512.so ]; then
+			LIBRARY="$CURDIR/libgowhisper-avx512.so"
+		fi
+	fi
+fi
+
 export LD_LIBRARY_PATH=$CURDIR/lib:$LD_LIBRARY_PATH
+export WHISPER_LIBRARY=$LIBRARY
 
 # If there is a lib/ld.so, use it
 if [ -f $CURDIR/lib/ld.so ]; then
 	echo "Using lib/ld.so"
+	echo "Using library: $LIBRARY"
 	exec $CURDIR/lib/ld.so $CURDIR/whisper "$@"
 fi
 
+echo "Using library: $LIBRARY"
 exec $CURDIR/whisper "$@"

backend/index.yaml

@@ -270,6 +270,7 @@
     nvidia: "cuda12-kokoro"
     intel: "intel-kokoro"
     amd: "rocm-kokoro"
+    nvidia-l4t: "nvidia-l4t-kokoro"
 - &coqui
   urls:
     - https://github.com/idiap/coqui-ai-TTS
@@ -350,6 +351,9 @@
   alias: "chatterbox"
   capabilities:
     nvidia: "cuda12-chatterbox"
+    metal: "metal-chatterbox"
+    default: "cpu-chatterbox"
+    nvidia-l4t: "nvidia-l4t-arm64-chatterbox"
 - &piper
   name: "piper"
   uri: "quay.io/go-skynet/local-ai-backends:latest-piper"
@@ -423,6 +427,68 @@
     - text-to-speech
     - TTS
   license: apache-2.0
+- &neutts
+  name: "neutts"
+  urls:
+    - https://github.com/neuphonic/neutts-air
+  description: |
+    NeuTTS Air is the world’s first super-realistic, on-device, TTS speech language model with instant voice cloning. Built off a 0.5B LLM backbone, NeuTTS Air brings natural-sounding speech, real-time performance, built-in security and speaker cloning to your local device - unlocking a new category of embedded voice agents, assistants, toys, and compliance-safe apps.
+  tags:
+    - text-to-speech
+    - TTS
+  license: apache-2.0
+  capabilities:
+    default: "cpu-neutts"
+    nvidia: "cuda12-neutts"
+    amd: "rocm-neutts"
+    nvidia-l4t: "nvidia-l4t-neutts"
+- !!merge <<: *neutts
+  name: "neutts-development"
+  capabilities:
+    default: "cpu-neutts-development"
+    nvidia: "cuda12-neutts-development"
+    amd: "rocm-neutts-development"
+    nvidia-l4t: "nvidia-l4t-neutts-development"
+- !!merge <<: *neutts
+  name: "cpu-neutts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-neutts"
+  mirrors:
+    - localai/localai-backends:latest-cpu-neutts
+- !!merge <<: *neutts
+  name: "cuda12-neutts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-12-neutts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-cuda-12-neutts
+- !!merge <<: *neutts
+  name: "rocm-neutts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-rocm-hipblas-neutts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-rocm-hipblas-neutts
+- !!merge <<: *neutts
+  name: "nvidia-l4t-neutts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-arm64-neutts"
+  mirrors:
+    - localai/localai-backends:latest-nvidia-l4t-arm64-neutts
+- !!merge <<: *neutts
+  name: "cpu-neutts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-neutts"
+  mirrors:
+    - localai/localai-backends:master-cpu-neutts
+- !!merge <<: *neutts
+  name: "cuda12-neutts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-12-neutts"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-cuda-12-neutts
+- !!merge <<: *neutts
+  name: "rocm-neutts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-rocm-hipblas-neutts"
+  mirrors:
+    - localai/localai-backends:master-gpu-rocm-hipblas-neutts
+- !!merge <<: *neutts
+  name: "nvidia-l4t-neutts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-arm64-neutts"
+  mirrors:
+    - localai/localai-backends:master-nvidia-l4t-arm64-neutts
 - !!merge <<: *mlx
   name: "mlx-development"
   uri: "quay.io/go-skynet/local-ai-backends:master-metal-darwin-arm64-mlx"
@@ -1047,6 +1113,7 @@
     nvidia: "cuda12-kokoro-development"
     intel: "intel-kokoro-development"
     amd: "rocm-kokoro-development"
+    nvidia-l4t: "nvidia-l4t-kokoro-development"
 - !!merge <<: *kokoro
   name: "cuda11-kokoro-development"
   uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-11-kokoro"
@@ -1072,6 +1139,16 @@
   uri: "quay.io/go-skynet/local-ai-backends:master-gpu-intel-kokoro"
   mirrors:
     - localai/localai-backends:master-gpu-intel-kokoro
+- !!merge <<: *kokoro
+  name: "nvidia-l4t-kokoro"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-l4t-kokoro"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-l4t-kokoro
+- !!merge <<: *kokoro
+  name: "nvidia-l4t-kokoro-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-l4t-kokoro"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-l4t-kokoro
 - !!merge <<: *kokoro
   name: "cuda11-kokoro"
   uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-11-kokoro"
@@ -1223,6 +1300,39 @@
   name: "chatterbox-development"
   capabilities:
     nvidia: "cuda12-chatterbox-development"
+    metal: "metal-chatterbox-development"
+    default: "cpu-chatterbox-development"
+    nvidia-l4t: "nvidia-l4t-arm64-chatterbox"
+- !!merge <<: *chatterbox
+  name: "cpu-chatterbox"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-chatterbox"
+  mirrors:
+    - localai/localai-backends:latest-cpu-chatterbox
+- !!merge <<: *chatterbox
+  name: "cpu-chatterbox-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-chatterbox"
+  mirrors:
+    - localai/localai-backends:master-cpu-chatterbox
+- !!merge <<: *chatterbox
+  name: "nvidia-l4t-arm64-chatterbox"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-l4t-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-l4t-arm64-chatterbox
+- !!merge <<: *chatterbox
+  name: "nvidia-l4t-arm64-chatterbox-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-l4t-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-l4t-arm64-chatterbox
+- !!merge <<: *chatterbox
+  name: "metal-chatterbox"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-metal-darwin-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:latest-metal-darwin-arm64-chatterbox
+- !!merge <<: *chatterbox
+  name: "metal-chatterbox-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-metal-darwin-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:master-metal-darwin-arm64-chatterbox
 - !!merge <<: *chatterbox
   name: "cuda12-chatterbox-development"
   uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-12-chatterbox"

backend/python/bark/requirements.txt

@@ -1,4 +1,4 @@
 bark==0.1.5
-grpcio==1.74.0
+grpcio==1.76.0
 protobuf
 certifi

backend/python/chatterbox/backend.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 """
-This is an extra gRPC server of LocalAI for Bark TTS
+This is an extra gRPC server of LocalAI for Chatterbox TTS
 """
 from concurrent import futures
 import time
@@ -14,15 +14,98 @@ import backend_pb2_grpc
 import torch
 import torchaudio as ta
 from chatterbox.tts import ChatterboxTTS
-
+from chatterbox.mtl_tts import ChatterboxMultilingualTTS
 import grpc
+import tempfile
 
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
+def split_text_at_word_boundary(text, max_length=250):
+    """
+    Split text at word boundaries without truncating words.
+    Returns a list of text chunks.
+    """
+    if not text or len(text) <= max_length:
+        return [text]
+    
+    chunks = []
+    words = text.split()
+    current_chunk = ""
+    
+    for word in words:
+        # Check if adding this word would exceed the limit
+        if len(current_chunk) + len(word) + 1 <= max_length:
+            if current_chunk:
+                current_chunk += " " + word
+            else:
+                current_chunk = word
+        else:
+            # If current chunk is not empty, add it to chunks
+            if current_chunk:
+                chunks.append(current_chunk)
+                current_chunk = word
+            else:
+                # If a single word is longer than max_length, we have to include it anyway
+                chunks.append(word)
+                current_chunk = ""
+    
+    # Add the last chunk if it's not empty
+    if current_chunk:
+        chunks.append(current_chunk)
+    
+    return chunks
+
+def merge_audio_files(audio_files, output_path, sample_rate):
+    """
+    Merge multiple audio files into a single audio file.
+    """
+    if not audio_files:
+        return
+    
+    if len(audio_files) == 1:
+        # If only one file, just copy it
+        import shutil
+        shutil.copy2(audio_files[0], output_path)
+        return
+    
+    # Load all audio files
+    waveforms = []
+    for audio_file in audio_files:
+        waveform, sr = ta.load(audio_file)
+        if sr != sample_rate:
+            # Resample if necessary
+            resampler = ta.transforms.Resample(sr, sample_rate)
+            waveform = resampler(waveform)
+        waveforms.append(waveform)
+    
+    # Concatenate all waveforms
+    merged_waveform = torch.cat(waveforms, dim=1)
+    
+    # Save the merged audio
+    ta.save(output_path, merged_waveform, sample_rate)
+    
+    # Clean up temporary files
+    for audio_file in audio_files:
+        if os.path.exists(audio_file):
+            os.remove(audio_file)
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
 MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
-COQUI_LANGUAGE = os.environ.get('COQUI_LANGUAGE', None)
 
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
@@ -47,6 +130,28 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if not torch.cuda.is_available() and request.CUDA:
             return backend_pb2.Result(success=False, message="CUDA is not available")
 
+
+        options = request.Options
+
+        # empty dict
+        self.options = {}
+
+        # The options are a list of strings in this form optname:optvalue
+        # We are storing all the options in a dict so we can use it later when
+        # generating the images
+        for opt in options:
+            if ":" not in opt:
+                continue
+            key, value = opt.split(":")
+            # if value is a number, convert it to the appropriate type
+            if is_float(value):
+                value = float(value)
+            elif is_int(value):
+                value = int(value)
+            elif value.lower() in ["true", "false"]:
+                value = value.lower() == "true"
+            self.options[key] = value
+
         self.AudioPath = None
 
         if os.path.isabs(request.AudioPath):
@@ -56,10 +161,14 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             modelFileBase = os.path.dirname(request.ModelFile)
             # modify LoraAdapter to be relative to modelFileBase
             self.AudioPath = os.path.join(modelFileBase, request.AudioPath)
-
         try:
             print("Preparing models, please wait", file=sys.stderr)
-            self.model = ChatterboxTTS.from_pretrained(device=device)
+            if "multilingual" in self.options:
+                # remove key from options
+                del self.options["multilingual"]
+                self.model = ChatterboxMultilingualTTS.from_pretrained(device=device)
+            else:
+                self.model = ChatterboxTTS.from_pretrained(device=device)
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         # Implement your logic here for the LoadModel service
@@ -68,14 +177,43 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
 
     def TTS(self, request, context):
         try:
-            # Generate audio using ChatterboxTTS
+            kwargs = {}
+
+            if "language" in self.options:
+                kwargs["language_id"] = self.options["language"]
             if self.AudioPath is not None:
-                wav = self.model.generate(request.text, audio_prompt_path=self.AudioPath)
+                kwargs["audio_prompt_path"] = self.AudioPath
+
+            # add options to kwargs
+            kwargs.update(self.options)
+
+            # Check if text exceeds 250 characters
+            # (chatterbox does not support long text)
+            # https://github.com/resemble-ai/chatterbox/issues/60
+            # https://github.com/resemble-ai/chatterbox/issues/110
+            if len(request.text) > 250:
+                # Split text at word boundaries
+                text_chunks = split_text_at_word_boundary(request.text, max_length=250)
+                print(f"Splitting text into chunks of 250 characters: {len(text_chunks)}", file=sys.stderr)
+                # Generate audio for each chunk
+                temp_audio_files = []
+                for i, chunk in enumerate(text_chunks):
+                    # Generate audio for this chunk
+                    wav = self.model.generate(chunk, **kwargs)
+                    
+                    # Create temporary file for this chunk
+                    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.wav')
+                    temp_file.close()
+                    ta.save(temp_file.name, wav, self.model.sr)
+                    temp_audio_files.append(temp_file.name)
+                
+                # Merge all audio files
+                merge_audio_files(temp_audio_files, request.dst, self.model.sr)
             else:
-                wav = self.model.generate(request.text)
-            
-            # Save the generated audio
-            ta.save(request.dst, wav, self.model.sr)
+                # Generate audio using ChatterboxTTS for short text
+                wav = self.model.generate(request.text, **kwargs)
+                # Save the generated audio
+                ta.save(request.dst, wav, self.model.sr)
             
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")

backend/python/chatterbox/install.sh

@@ -15,5 +15,6 @@ fi
 if [ "x${BUILD_PROFILE}" == "xintel" ]; then
     EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
 fi
+EXTRA_PIP_INSTALL_FLAGS+=" --no-build-isolation"
 
 installRequirements

backend/python/chatterbox/requirements-cpu.txt

@@ -1,5 +1,9 @@
+--extra-index-url https://download.pytorch.org/whl/cpu
 accelerate
-torch==2.6.0
-torchaudio==2.6.0
-transformers==4.46.3
-chatterbox-tts
+torch
+torchaudio
+numpy>=1.24.0,<1.26.0
+transformers
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
+#chatterbox-tts==0.1.4

backend/python/chatterbox/requirements-cublas11.txt

@@ -2,5 +2,7 @@
 torch==2.6.0+cu118
 torchaudio==2.6.0+cu118
 transformers==4.46.3
-chatterbox-tts
+numpy>=1.24.0,<1.26.0
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-cublas12.txt

@@ -1,5 +1,7 @@
-torch==2.6.0
-torchaudio==2.6.0
-transformers==4.46.3
-chatterbox-tts
+torch
+torchaudio
+transformers
+numpy>=1.24.0,<1.26.0
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-hipblas.txt

@@ -1,6 +1,8 @@
 --extra-index-url https://download.pytorch.org/whl/rocm6.0
 torch==2.6.0+rocm6.1
 torchaudio==2.6.0+rocm6.1
-transformers==4.46.3
-chatterbox-tts
+transformers
+numpy>=1.24.0,<1.26.0
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-intel.txt

@@ -2,10 +2,11 @@
 intel-extension-for-pytorch==2.3.110+xpu
 torch==2.3.1+cxx11.abi
 torchaudio==2.3.1+cxx11.abi
-transformers==4.46.3
-chatterbox-tts
+transformers
+numpy>=1.24.0,<1.26.0
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate
 oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]
-setuptools
-accelerate
+setuptools

backend/python/chatterbox/requirements-l4t.txt

@@ -0,0 +1,7 @@
+--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu126/
+torch
+torchaudio
+transformers
+numpy>=1.24.0,<1.26.0
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
+accelerate

backend/python/chatterbox/requirements.txt

@@ -2,4 +2,5 @@ grpcio==1.71.0
 protobuf
 certifi
 packaging
-setuptools
+setuptools
+poetry

backend/python/common/template/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.74.0
+grpcio==1.76.0
 protobuf
 grpcio-tools

backend/python/coqui/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.74.0
+grpcio==1.76.0
 protobuf
 certifi
 packaging==24.1

backend/python/diffusers/backend.py

@@ -66,11 +66,20 @@ from diffusers.schedulers import (
 )
 
 def is_float(s):
+    """Check if a string can be converted to float."""
     try:
         float(s)
         return True
     except ValueError:
         return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
 
 # The scheduler list mapping was taken from here: https://github.com/neggles/animatediff-cli/blob/6f336f5f4b5e38e85d7f06f1744ef42d0a45f2a7/src/animatediff/schedulers.py#L39
 # Credits to https://github.com/neggles
@@ -177,10 +186,11 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                 key, value = opt.split(":")
                 # if value is a number, convert it to the appropriate type
                 if is_float(value):
-                    if value.is_integer():
-                        value = int(value)
-                    else:
-                        value = float(value)
+                    value = float(value)
+                elif is_int(value):
+                    value = int(value)
+                elif value.lower() in ["true", "false"]:
+                    value = value.lower() == "true"
                 self.options[key] = value
 
             # From options, extract if present "torch_dtype" and set it to the appropriate type

backend/python/diffusers/requirements.txt

@@ -1,5 +1,5 @@
 setuptools
-grpcio==1.74.0
+grpcio==1.76.0
 pillow
 protobuf
 certifi

backend/python/diffusers/test.py

@@ -31,7 +31,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method tests if the server starts up successfully
         """
-        time.sleep(10)
+        time.sleep(20)
         try:
             self.setUp()
             with grpc.insecure_channel("localhost:50051") as channel:
@@ -48,7 +48,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method tests if the model is loaded successfully
         """
-        time.sleep(10)
+        time.sleep(20)
         try:
             self.setUp()
             with grpc.insecure_channel("localhost:50051") as channel:
@@ -66,7 +66,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method tests if the backend can generate images
         """
-        time.sleep(10)
+        time.sleep(20)
         try:
             self.setUp()
             with grpc.insecure_channel("localhost:50051") as channel:

backend/python/exllama2/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.74.0
+grpcio==1.76.0
 protobuf
 certifi
 wheel

backend/python/kokoro/backend.py

@@ -64,15 +64,15 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             # Generate audio using Kokoro pipeline
             generator = self.pipeline(request.text, voice=voice)
             
-            # Get the first (and typically only) audio segment
+            speechs = []
+            # Get all the audio segment
             for i, (gs, ps, audio) in enumerate(generator):
-                # Save audio to the destination file
-                sf.write(request.dst, audio, 24000)
+                speechs.append(audio)
                 print(f"Generated audio segment {i}: gs={gs}, ps={ps}", file=sys.stderr)
-                # For now, we only process the first segment
-                # If you need to handle multiple segments, you might want to modify this
-                break
-                
+            # Merges the audio segments and writes them to the destination
+            speech = torch.cat(speechs, dim=0)
+            sf.write(request.dst, speech, 24000)
+
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         

backend/python/kokoro/requirements-l4t.txt

@@ -0,0 +1,7 @@
+--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu126/
+torch
+torchaudio
+transformers
+accelerate
+kokoro
+soundfile

backend/python/mlx-audio/backend.py

@@ -20,6 +20,21 @@ import soundfile as sf
 import numpy as np
 import uuid
 
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
@@ -32,14 +47,6 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
     This backend provides TTS (Text-to-Speech) functionality using MLX-Audio.
     """
 
-    def _is_float(self, s):
-        """Check if a string can be converted to float."""
-        try:
-            float(s)
-            return True
-        except ValueError:
-            return False
-
     def Health(self, request, context):
         """
         Returns a health check message.
@@ -80,11 +87,10 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                 key, value = opt.split(":", 1)  # Split only on first colon to handle values with colons
                 
                 # Convert numeric values to appropriate types
-                if self._is_float(value):
-                    if float(value).is_integer():
-                        value = int(value)
-                    else:
-                        value = float(value)
+                if is_float(value):
+                    value = float(value)
+                elif is_int(value):
+                    value = int(value)
                 elif value.lower() in ["true", "false"]:
                     value = value.lower() == "true"
                     

backend/python/mlx-vlm/backend.py

@@ -21,6 +21,21 @@ import io
 from PIL import Image
 import tempfile
 
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
@@ -32,14 +47,6 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
     A gRPC servicer that implements the Backend service defined in backend.proto.
     """
 
-    def _is_float(self, s):
-        """Check if a string can be converted to float."""
-        try:
-            float(s)
-            return True
-        except ValueError:
-            return False
-
     def Health(self, request, context):
         """
         Returns a health check message.
@@ -79,12 +86,10 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                     continue
                 key, value = opt.split(":", 1)  # Split only on first colon to handle values with colons
                 
-                # Convert numeric values to appropriate types
-                if self._is_float(value):
-                    if float(value).is_integer():
-                        value = int(value)
-                    else:
-                        value = float(value)
+                if is_float(value):
+                    value = float(value)
+                elif is_int(value):
+                    value = int(value)
                 elif value.lower() in ["true", "false"]:
                     value = value.lower() == "true"
                     

backend/python/mlx/backend.py

@@ -24,20 +24,27 @@ _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
 MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
 
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
     """
     A gRPC servicer that implements the Backend service defined in backend.proto.
     """
 
-    def _is_float(self, s):
-        """Check if a string can be converted to float."""
-        try:
-            float(s)
-            return True
-        except ValueError:
-            return False
-
     def Health(self, request, context):
         """
         Returns a health check message.
@@ -78,11 +85,10 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                 key, value = opt.split(":", 1)  # Split only on first colon to handle values with colons
                 
                 # Convert numeric values to appropriate types
-                if self._is_float(value):
-                    if float(value).is_integer():
-                        value = int(value)
-                    else:
-                        value = float(value)
+                if is_float(value):
+                    value = float(value)
+                elif is_int(value):
+                    value = int(value)
                 elif value.lower() in ["true", "false"]:
                     value = value.lower() == "true"
                     

backend/python/neutts/Makefile

@@ -0,0 +1,23 @@
+.PHONY: neutts
+neutts:
+	bash install.sh
+
+.PHONY: run
+run: neutts
+	@echo "Running neutts..."
+	bash run.sh
+	@echo "neutts run."
+
+.PHONY: test
+test: neutts
+	@echo "Testing neutts..."
+	bash test.sh
+	@echo "neutts tested."
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: clean
+clean: protogen-clean
+	rm -rf venv __pycache__

backend/python/neutts/backend.py

@@ -0,0 +1,162 @@
+#!/usr/bin/env python3
+"""
+This is an extra gRPC server of LocalAI for NeuTTSAir
+"""
+from concurrent import futures
+import time
+import argparse
+import signal
+import sys
+import os
+import backend_pb2
+import backend_pb2_grpc
+import torch
+from neuttsair.neutts import NeuTTSAir
+import soundfile as sf
+
+import grpc
+
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    """
+    BackendServicer is the class that implements the gRPC service
+    """
+    def Health(self, request, context):
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+    def LoadModel(self, request, context):
+
+        # Get device
+        # device = "cuda" if request.CUDA else "cpu"
+        if torch.cuda.is_available():
+            print("CUDA is available", file=sys.stderr)
+            device = "cuda"
+        else:
+            print("CUDA is not available", file=sys.stderr)
+            device = "cpu"
+        mps_available = hasattr(torch.backends, "mps") and torch.backends.mps.is_available()
+        if mps_available:
+            device = "mps"
+        if not torch.cuda.is_available() and request.CUDA:
+            return backend_pb2.Result(success=False, message="CUDA is not available")
+
+
+        options = request.Options
+
+        # empty dict
+        self.options = {}
+        self.ref_text = None
+
+        # The options are a list of strings in this form optname:optvalue
+        # We are storing all the options in a dict so we can use it later when
+        # generating the images
+        for opt in options:
+            if ":" not in opt:
+                continue
+            key, value = opt.split(":")
+            # if value is a number, convert it to the appropriate type
+            if is_float(value):
+                value = float(value)
+            elif is_int(value):
+                value = int(value)
+            elif value.lower() in ["true", "false"]:
+                value = value.lower() == "true"
+            self.options[key] = value
+
+        codec_repo = "neuphonic/neucodec"
+        if "codec_repo" in self.options:
+            codec_repo = self.options["codec_repo"]
+            del self.options["codec_repo"]
+        if "ref_text" in self.options:
+            self.ref_text = self.options["ref_text"]
+            del self.options["ref_text"]
+
+        self.AudioPath = None
+
+        if os.path.isabs(request.AudioPath):
+            self.AudioPath = request.AudioPath
+        elif request.AudioPath and request.ModelFile != "" and not os.path.isabs(request.AudioPath):
+            # get base path of modelFile
+            modelFileBase = os.path.dirname(request.ModelFile)
+            # modify LoraAdapter to be relative to modelFileBase
+            self.AudioPath = os.path.join(modelFileBase, request.AudioPath)
+        try:
+            print("Preparing models, please wait", file=sys.stderr)
+            self.model = NeuTTSAir(backbone_repo=request.Model, backbone_device=device, codec_repo=codec_repo, codec_device=device)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        # Implement your logic here for the LoadModel service
+        # Replace this with your desired response
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def TTS(self, request, context):
+        try:
+            kwargs = {}
+
+            # add options to kwargs
+            kwargs.update(self.options)
+
+            ref_codes = self.model.encode_reference(self.AudioPath)
+
+            wav = self.model.infer(request.text, ref_codes, self.ref_text)
+
+            sf.write(request.dst, wav, 24000)            
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        return backend_pb2.Result(success=True)
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS),
+        options=[
+            ('grpc.max_message_length', 50 * 1024 * 1024),  # 50MB
+            ('grpc.max_send_message_length', 50 * 1024 * 1024),  # 50MB
+            ('grpc.max_receive_message_length', 50 * 1024 * 1024),  # 50MB
+        ])
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+
+    serve(args.addr)

backend/python/neutts/install.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+set -e
+
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
+# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
+# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
+# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
+if [ "x${BUILD_PROFILE}" == "xintel" ]; then
+    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
+fi
+
+if [ "x${BUILD_TYPE}" == "xcublas" ] || [ "x${BUILD_TYPE}" == "xl4t" ]; then
+    export CMAKE_ARGS="-DGGML_CUDA=on"
+fi
+
+if [ "x${BUILD_TYPE}" == "xhipblas" ]; then
+    export CMAKE_ARGS="-DGGML_HIPBLAS=on"
+fi
+
+EXTRA_PIP_INSTALL_FLAGS+=" --no-build-isolation"
+
+git clone https://github.com/neuphonic/neutts-air neutts-air
+
+cp -rfv neutts-air/neuttsair ./
+
+installRequirements

backend/python/neutts/requirements-after.txt

@@ -0,0 +1,2 @@
+datasets==4.1.1
+torchtune==0.6.1

backend/python/neutts/requirements-cpu.txt

@@ -0,0 +1,10 @@
+--extra-index-url https://download.pytorch.org/whl/cpu
+accelerate
+torch==2.8.0
+transformers==4.56.1
+librosa==0.11.0
+neucodec>=0.0.4
+phonemizer==3.3.0
+soundfile==0.13.1
+resemble-perth==1.0.1
+llama-cpp-python

backend/python/neutts/requirements-cublas12.txt

@@ -0,0 +1,8 @@
+librosa==0.11.0
+neucodec>=0.0.4
+phonemizer==3.3.0
+soundfile==0.13.1
+torch==2.8.0
+transformers==4.56.1
+resemble-perth==1.0.1
+accelerate

backend/python/neutts/requirements-hipblas.txt

@@ -0,0 +1,10 @@
+--extra-index-url https://download.pytorch.org/whl/rocm6.3
+torch==2.8.0+rocm6.3
+transformers==4.56.1
+accelerate
+librosa==0.11.0
+neucodec>=0.0.4
+phonemizer==3.3.0
+soundfile==0.13.1
+resemble-perth==1.0.1
+llama-cpp-python

backend/python/neutts/requirements-l4t.txt

@@ -0,0 +1,10 @@
+--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu126/
+torch
+transformers
+accelerate
+librosa==0.11.0
+neucodec>=0.0.4
+phonemizer==3.3.0
+soundfile==0.13.1
+resemble-perth==1.0.1
+llama-cpp-python

backend/python/neutts/requirements.txt

@@ -0,0 +1,7 @@
+grpcio==1.71.0
+protobuf
+certifi
+packaging
+setuptools
+numpy==2.2.6
+scikit_build_core

backend/python/neutts/run.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+
+startBackend $@

backend/python/neutts/test.py

@@ -0,0 +1,82 @@
+"""
+A test script to test the gRPC service
+"""
+import unittest
+import subprocess
+import time
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+
+class TestBackendServicer(unittest.TestCase):
+    """
+    TestBackendServicer is the class that tests the gRPC service
+    """
+    def setUp(self):
+        """
+        This method sets up the gRPC service by starting the server
+        """
+        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
+        time.sleep(30)
+
+    def tearDown(self) -> None:
+        """
+        This method tears down the gRPC service by terminating the server
+        """
+        self.service.terminate()
+        self.service.wait()
+
+    def test_server_startup(self):
+        """
+        This method tests if the server starts up successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.Health(backend_pb2.HealthMessage())
+                self.assertEqual(response.message, b'OK')
+        except Exception as err:
+            print(err)
+            self.fail("Server failed to start")
+        finally:
+            self.tearDown()
+
+    def test_load_model(self):
+        """
+        This method tests if the model is loaded successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions())
+                print(response)
+                self.assertTrue(response.success)
+                self.assertEqual(response.message, "Model loaded successfully")
+        except Exception as err:
+            print(err)
+            self.fail("LoadModel service failed")
+        finally:
+            self.tearDown()
+
+    def test_tts(self):
+        """
+        This method tests if the embeddings are generated successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions())
+                self.assertTrue(response.success)
+                tts_request = backend_pb2.TTSRequest(text="80s TV news production music hit for tonight's biggest story")
+                tts_response = stub.TTS(tts_request)
+                self.assertIsNotNone(tts_response)
+        except Exception as err:
+            print(err)
+            self.fail("TTS service failed")
+        finally:
+            self.tearDown()

backend/python/neutts/test.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+set -e
+
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+runUnittests

backend/python/rerankers/backend.py

@@ -61,7 +61,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             if request.PipelineType != "": # Reuse the PipelineType field for language
                 kwargs['lang'] = request.PipelineType
             self.model_name = model_name
-            self.model = Reranker(model_name, **kwargs)  
+            self.model = Reranker(model_name, **kwargs)
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
 
@@ -75,12 +75,13 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             documents.append(doc)
         ranked_results=self.model.rank(query=request.query, docs=documents, doc_ids=list(range(len(request.documents))))
         # Prepare results to return
+        cropped_results = ranked_results.top_k(request.top_n) if request.top_n > 0 else ranked_results
         results = [
             backend_pb2.DocumentResult(
                 index=res.doc_id,
                 text=res.text,
                 relevance_score=res.score
-            ) for res in ranked_results.results
+            ) for res in (cropped_results)
         ]
 
         # Calculate the usage and total tokens

backend/python/rerankers/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.74.0
+grpcio==1.76.0
 protobuf
 certifi

backend/python/rerankers/test.py

@@ -76,7 +76,7 @@ class TestBackendServicer(unittest.TestCase):
                 )
                 response = stub.LoadModel(backend_pb2.ModelOptions(Model="cross-encoder"))
                 self.assertTrue(response.success)
-               
+
                 rerank_response = stub.Rerank(request)
                 print(rerank_response.results[0])
                 self.assertIsNotNone(rerank_response.results)
@@ -87,4 +87,60 @@ class TestBackendServicer(unittest.TestCase):
             print(err)
             self.fail("Reranker service failed")
         finally:
-            self.tearDown()
+            self.tearDown()
+
+    def test_rerank_omit_top_n(self):
+        """
+        This method tests if the embeddings are generated successfully even top_n is omitted
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                request = backend_pb2.RerankRequest(
+                    query="I love you",
+                    documents=["I hate you", "I really like you"],
+                    top_n=0 # 
+                )
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="cross-encoder"))
+                self.assertTrue(response.success)
+
+                rerank_response = stub.Rerank(request)
+                print(rerank_response.results[0])
+                self.assertIsNotNone(rerank_response.results)
+                self.assertEqual(len(rerank_response.results), 2)
+                self.assertEqual(rerank_response.results[0].text, "I really like you")
+                self.assertEqual(rerank_response.results[1].text, "I hate you")
+        except Exception as err:
+            print(err)
+            self.fail("Reranker service failed")
+        finally:
+            self.tearDown()
+
+    def test_rerank_crop(self):
+        """
+        This method tests top_n cropping
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                request = backend_pb2.RerankRequest(
+                    query="I love you",
+                    documents=["I hate you", "I really like you", "I hate ignoring top_n"],
+                    top_n=2
+                )
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="cross-encoder"))
+                self.assertTrue(response.success)
+
+                rerank_response = stub.Rerank(request)
+                print(rerank_response.results[0])
+                self.assertIsNotNone(rerank_response.results)
+                self.assertEqual(len(rerank_response.results), 2)
+                self.assertEqual(rerank_response.results[0].text, "I really like you")
+                self.assertEqual(rerank_response.results[1].text, "I hate you")
+        except Exception as err:
+            print(err)
+            self.fail("Reranker service failed")
+        finally:
+            self.tearDown()

backend/python/transformers/requirements-cpu.txt

@@ -6,4 +6,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.1.0
-protobuf==6.32.0
+protobuf==6.33.1

backend/python/transformers/requirements-cublas11.txt

@@ -7,4 +7,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.1.0
-protobuf==6.32.0
+protobuf==6.33.1

backend/python/transformers/requirements-cublas12.txt

@@ -6,4 +6,4 @@ transformers
 bitsandbytes
 outetts
 sentence-transformers==5.1.0
-protobuf==6.32.0
+protobuf==6.33.1

backend/python/transformers/requirements-hipblas.txt

@@ -8,4 +8,4 @@ bitsandbytes
 outetts
 bitsandbytes
 sentence-transformers==5.1.0
-protobuf==6.32.0
+protobuf==6.33.1

backend/python/transformers/requirements-intel.txt

@@ -10,4 +10,4 @@ intel-extension-for-transformers
 bitsandbytes
 outetts
 sentence-transformers==5.1.0
-protobuf==6.32.0
+protobuf==6.33.1