chore(stablediffusion-ncn): drop in favor of ggml implementation (#4652)

* chore(stablediffusion-ncn): drop in favor of ggml implementation

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(ci): drop stablediffusion build

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(tests): add

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(tests): try to fixup current tests

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* Try to fix tests

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* Tests improvements

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(tests): use quality to specify step

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(tests): switch to sd-1.5

also increase prep time for downloading models

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

core/config/backend_config.go

@@ -515,7 +515,7 @@ func (c *BackendConfig) GuessUsecases(u BackendConfigUsecases) bool {
 		}
 	}
 	if (u & FLAG_IMAGE) == FLAG_IMAGE {
-		imageBackends := []string{"diffusers", "stablediffusion"}
+		imageBackends := []string{"diffusers", "stablediffusion", "stablediffusion-ggml"}
 		if !slices.Contains(imageBackends, c.Backend) {
 			return false
 		}

core/config/config_test.go

@@ -48,5 +48,66 @@ var _ = Describe("Test cases for config related functions", func() {
 			// config should includes whisper-1 models's api.config
 			Expect(loadedModelNames).To(ContainElements("whisper-1"))
 		})
+
+		It("Test new loadconfig", func() {
+
+			bcl := NewBackendConfigLoader(os.Getenv("MODELS_PATH"))
+			err := bcl.LoadBackendConfigsFromPath(os.Getenv("MODELS_PATH"))
+			Expect(err).To(BeNil())
+			configs := bcl.GetAllBackendConfigs()
+			loadedModelNames := []string{}
+			for _, v := range configs {
+				loadedModelNames = append(loadedModelNames, v.Name)
+			}
+			Expect(configs).ToNot(BeNil())
+			totalModels := len(loadedModelNames)
+
+			Expect(loadedModelNames).To(ContainElements("code-search-ada-code-001"))
+
+			// config should includes text-embedding-ada-002 models's api.config
+			Expect(loadedModelNames).To(ContainElements("text-embedding-ada-002"))
+
+			// config should includes rwkv_test models's api.config
+			Expect(loadedModelNames).To(ContainElements("rwkv_test"))
+
+			// config should includes whisper-1 models's api.config
+			Expect(loadedModelNames).To(ContainElements("whisper-1"))
+
+			// create a temp directory and store a temporary model
+			tmpdir, err := os.MkdirTemp("", "test")
+			Expect(err).ToNot(HaveOccurred())
+			defer os.RemoveAll(tmpdir)
+
+			// create a temporary model
+			model := `name: "test-model"
+description: "test model"
+options:
+- foo
+- bar
+- baz
+`
+			modelFile := tmpdir + "/test-model.yaml"
+			err = os.WriteFile(modelFile, []byte(model), 0644)
+			Expect(err).ToNot(HaveOccurred())
+
+			err = bcl.LoadBackendConfigsFromPath(tmpdir)
+			Expect(err).ToNot(HaveOccurred())
+
+			configs = bcl.GetAllBackendConfigs()
+			Expect(len(configs)).ToNot(Equal(totalModels))
+
+			loadedModelNames = []string{}
+			var testModel BackendConfig
+			for _, v := range configs {
+				loadedModelNames = append(loadedModelNames, v.Name)
+				if v.Name == "test-model" {
+					testModel = v
+				}
+			}
+			Expect(loadedModelNames).To(ContainElements("test-model"))
+			Expect(testModel.Description).To(Equal("test model"))
+			Expect(testModel.Options).To(ContainElements("foo", "bar", "baz"))
+
+		})
 	})
 })

core/http/app_test.go

@@ -687,6 +687,10 @@ var _ = Describe("API test", func() {
 					Name: "model-gallery",
 					URL:  "https://raw.githubusercontent.com/go-skynet/model-gallery/main/index.yaml",
 				},
+				{
+					Name: "localai",
+					URL:  "https://raw.githubusercontent.com/mudler/LocalAI/refs/heads/master/gallery/index.yaml",
+				},
 			}
 
 			application, err := application.New(
@@ -764,10 +768,8 @@ var _ = Describe("API test", func() {
 			}
 
 			response := postModelApplyRequest("http://127.0.0.1:9090/models/apply", modelApplyRequest{
-				ID: "model-gallery@stablediffusion",
-				Overrides: map[string]interface{}{
-					"parameters": map[string]interface{}{"model": "stablediffusion_assets"},
-				},
+				ID:   "localai@sd-1.5-ggml",
+				Name: "stablediffusion",
 			})
 
 			Expect(response["uuid"]).ToNot(BeEmpty(), fmt.Sprint(response))
@@ -778,14 +780,14 @@ var _ = Describe("API test", func() {
 				response := getModelStatus("http://127.0.0.1:9090/models/jobs/" + uuid)
 				fmt.Println(response)
 				return response["processed"].(bool)
-			}, "360s", "10s").Should(Equal(true))
+			}, "1200s", "10s").Should(Equal(true))
 
 			resp, err := http.Post(
 				"http://127.0.0.1:9090/v1/images/generations",
 				"application/json",
 				bytes.NewBuffer([]byte(`{
-					 			"prompt": "floating hair, portrait, ((loli)), ((one girl)), cute face, hidden hands, asymmetrical bangs, beautiful detailed eyes, eye shadow, hair ornament, ribbons, bowties, buttons, pleated skirt, (((masterpiece))), ((best quality)), colorful|((part of the head)), ((((mutated hands and fingers)))), deformed, blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, extra limb, ugly, poorly drawn hands, missing limb, blurry, floating limbs, disconnected limbs, malformed hands, blur, out of focus, long neck, long body, Octane renderer, lowres, bad anatomy, bad hands, text",
-								"mode": 2,  "seed":9000,
+					 			"prompt": "a lovely cat",
+								"step": 1,  "seed":9000,
 					 			"size": "256x256", "n":2}`)))
 			// The response should contain an URL
 			Expect(err).ToNot(HaveOccurred(), fmt.Sprint(resp))
@@ -794,6 +796,7 @@ var _ = Describe("API test", func() {
 
 			imgUrlResp := &schema.OpenAIResponse{}
 			err = json.Unmarshal(dat, imgUrlResp)
+			Expect(err).ToNot(HaveOccurred(), fmt.Sprint(dat))
 			Expect(imgUrlResp.Data).ToNot(Or(BeNil(), BeZero()))
 			imgUrl := imgUrlResp.Data[0].URL
 			Expect(imgUrl).To(ContainSubstring("http://127.0.0.1:9090/"), imgUrl)

core/http/endpoints/openai/image.go

@@ -72,7 +72,7 @@ func ImageEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appCon
 		}
 
 		if m == "" {
-			m = model.StableDiffusionBackend
+			m = "stablediffusion"
 		}
 		log.Debug().Msgf("Loading model: %+v", m)
 
@@ -129,9 +129,9 @@ func ImageEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appCon
 
 		switch config.Backend {
 		case "stablediffusion":
-			config.Backend = model.StableDiffusionBackend
+			config.Backend = model.StableDiffusionGGMLBackend
 		case "":
-			config.Backend = model.StableDiffusionBackend
+			config.Backend = model.StableDiffusionGGMLBackend
 		}
 
 		if !strings.Contains(input.Size, "x") {

core/http/endpoints/openai/request.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"strconv"
 
 	"github.com/gofiber/fiber/v2"
 	"github.com/google/uuid"
@@ -296,6 +297,14 @@ func updateRequestConfig(config *config.BackendConfig, input *schema.OpenAIReque
 			}
 		}
 	}
+
+	// If a quality was defined as number, convert it to step
+	if input.Quality != "" {
+		q, err := strconv.Atoi(input.Quality)
+		if err == nil {
+			config.Step = q
+		}
+	}
 }
 
 func mergeRequestWithConfig(modelFile string, input *schema.OpenAIRequest, cm *config.BackendConfigLoader, loader *model.ModelLoader, debug bool, threads, ctx int, f16 bool) (*config.BackendConfig, *schema.OpenAIRequest, error) {

core/schema/openai.go

@@ -191,8 +191,9 @@ type OpenAIRequest struct {
 	Stream bool `json:"stream"`
 
 	// Image (not supported by OpenAI)
-	Mode int `json:"mode"`
-	Step int `json:"step"`
+	Mode    int    `json:"mode"`
+	Quality string `json:"quality"`
+	Step    int    `json:"step"`
 
 	// A grammar to constrain the LLM output
 	Grammar string `json:"grammar" yaml:"grammar"`