Minor fixes (#285)

.github/workflows/bump_deps.yaml

@@ -9,9 +9,6 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - repository: "go-skynet/go-gpt4all-j.cpp"
-            variable: "GOGPT4ALLJ_VERSION"
-            branch: "master"
           - repository: "go-skynet/go-llama.cpp"
             variable: "GOLLAMA_VERSION"
             branch: "master"
@@ -30,7 +27,7 @@ jobs:
           - repository: "go-skynet/bloomz.cpp"
             variable: "BLOOMZ_VERSION"
             branch: "main"
-          - repository: "go-skynet/gpt4all"
+          - repository: "nomic-ai/gpt4all"
             variable: "GPT4ALL_VERSION"
             branch: "main"
     runs-on: ubuntu-latest

README.md

@@ -696,6 +696,8 @@ curl http://localhost:8080/v1/models
 
 ### Embeddings
 
+OpenAI docs: https://platform.openai.com/docs/api-reference/embeddings
+
 <details>
 
 The embedding endpoint is experimental and enabled only if the model is configured with `embeddings: true` in its `yaml` file, for example:
@@ -742,6 +744,8 @@ curl http://localhost:8080/v1/audio/transcriptions -H "Content-Type: multipart/f
   
 ### Image generation
 
+OpenAI docs: https://platform.openai.com/docs/api-reference/images/create
+
 LocalAI supports generating images with Stable diffusion, running on CPU.
 
 | mode=0                                                                                                                | mode=1 (winograd/sgemm)                                                                                                                |
@@ -773,6 +777,8 @@ curl http://localhost:8080/v1/images/generations -H "Content-Type: application/j
           }'
 ```
 
+Note: image generator supports images up to 512x512. You can use other tools however to upscale the image, for instance: https://github.com/upscayl/upscayl.
+
 #### Setup
 
 Note: In order to use the `images/generation` endpoint, you need to build LocalAI with `GO_TAGS=stablediffusion`.
@@ -847,7 +853,7 @@ Yes! If the client uses OpenAI and supports setting a different base URL to send
 
 <details>
 
-Not currently, as ggml doesn't support GPUs yet: https://github.com/ggerganov/llama.cpp/discussions/915.
+There is partial GPU support, see build instructions above.
 
 </details>
 

api/openai.go

@@ -289,12 +289,14 @@ func chatEndpoint(cm ConfigMerger, debug bool, loader *model.ModelLoader, thread
 
 		mess := []string{}
 		for _, i := range input.Messages {
+			var content string
 			r := config.Roles[i.Role]
-			if r == "" {
+			if r != "" {
-				r = i.Role
+				content = fmt.Sprint(r, " ", i.Content)
+			} else {
+				content = i.Content
 			}
 
-			content := fmt.Sprint(r, " ", i.Content)
 			mess = append(mess, content)
 		}
 
@@ -491,6 +493,11 @@ func imageEndpoint(cm ConfigMerger, debug bool, loader *model.ModelLoader, image
 
 		var result []Item
 		for _, i := range config.PromptStrings {
+			n := input.N
+			if input.N == 0 {
+				n = 1
+			}
+			for j := 0; j < n; j++ {
 				prompts := strings.Split(i, "|")
 				positive_prompt := prompts[0]
 				negative_prompt := ""
@@ -552,6 +559,7 @@ func imageEndpoint(cm ConfigMerger, debug bool, loader *model.ModelLoader, image
 
 				result = append(result, *item)
 			}
+		}
 
 		resp := &OpenAIResponse{
 			Data: result,