Rewrite README with concise model and endpoint docs

README.md

@@ -9,79 +9,60 @@ pinned: true
 
 # Code-Embed-Qwen-rerank-sentiment
 
-Lazy-loading CPU-first code and multimodal retrieval API. This is the only custom Space in the set because it needs code embeddings, a Qwen reranker, sentiment classification, and CLIP image embeddings without keeping every model resident in memory at once.
+Live API: `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space`
 
-## Model Set
+## Models
 
-- Text / code embeddings: `jinaai/jina-code-embeddings-0.5b`
+- Code embeddings: `jinaai/jina-code-embeddings-0.5b`
+  Served name: `code-embed`
+  Vector dimension: `896`
 - Reranker: `Qwen/Qwen3-Reranker-0.6B`
+  Served name: `code-rerank`
 - Classifier: `clapAI/modernBERT-base-multilingual-sentiment`
+  Served name: `code-sentiment`
 - Image embeddings: `sentence-transformers/clip-ViT-B-32`
+  Served name: `clip-image`
+  Vector dimension: `512`
 
 ## Endpoints
 
-- `GET /health`
-- `GET /models`
-- `POST /embeddings`
-- `POST /rerank`
-- `POST /classify`
-- `POST /embeddings_image`
-- `GET /openapi.json`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/health`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/models`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/embeddings`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/embeddings_image`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/rerank`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/classify`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/openapi.json`
 
-## Example Requests
+## OpenAI-Style Aliases
+
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/v1/models`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/v1/embeddings`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/v1/rerank`
+- `https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/v1/classify`
 
-### Code Embeddings
+## Example Requests
 
 ```bash
-curl -X POST "$SPACE_URL/embeddings" \
+curl -X POST "https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/embeddings" \
   -H "Content-Type: application/json" \
-  -d '{
-    "model": "code-embed",
-    "input": ["def quick_sort(arr): return sorted(arr)"]
-  }'
+  -d '{"model":"code-embed","input":["def quick_sort(arr): return sorted(arr)"]}'
 ```
 
-### Image Embeddings
-
 ```bash
-curl -X POST "$SPACE_URL/embeddings" \
+curl -X POST "https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/embeddings_image" \
   -H "Content-Type: application/json" \
-  -d '{
-    "model": "clip-image",
-    "modality": "image",
-    "input": ["https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"]
-  }'
+  -d '{"model":"clip-image","input":["https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"]}'
 ```
 
-### Reranking
-
 ```bash
-curl -X POST "$SPACE_URL/rerank" \
+curl -X POST "https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/rerank" \
   -H "Content-Type: application/json" \
-  -d '{
-    "model": "code-rerank",
-    "query": "python quick sort implementation",
-    "documents": [
-      "def quick_sort(arr): return sorted(arr)",
-      "SELECT * FROM users ORDER BY created_at DESC"
-    ],
-    "return_documents": true
-  }'
+  -d '{"model":"code-rerank","query":"python quick sort implementation","documents":["def quick_sort(arr): return sorted(arr)","SELECT * FROM users ORDER BY created_at DESC"],"return_documents":true}'
 ```
 
-### Classification
-
 ```bash
-curl -X POST "$SPACE_URL/classify" \
+curl -X POST "https://chmielvu-code-embed-qwen-rerank-sentiment.hf.space/classify" \
   -H "Content-Type: application/json" \
-  -d '{
-    "model": "code-sentiment",
-    "input": ["The API is fast and easy to use."]
-  }'
+  -d '{"model":"code-sentiment","input":["The API is fast and easy to use."]}'
 ```
-
-## Notes
-
-- The server lazy-loads one model family at a time to fit `cpu-basic` more safely.
-- The first request after switching tasks will be slower because the model may need to be loaded.
-- Jina embedding and reranking models are under `CC BY-NC 4.0`; verify that license for your use case.