docs: add documentation for the chat endpoint and include required dependencies in requirements.txt

README.md

@@ -54,3 +54,52 @@ Navigate to the root URL in your browser:
   "Hello": "World!"
 }
 ```
+
+## Chat Endpoint
+
+The `chat` endpoint is located at `/chat` and uses Hugging Face's `InferenceClient` (running `nvidia/Llama-3.1-Nemotron-Nano-8B-v1`) to get chat completions.
+
+> [!NOTE]
+> This endpoint requires the `HF_TOKEN` environment variable to be set with a valid Hugging Face Hub token.
+
+### How to Call
+
+#### 1. Via cURL
+- **Hosted Space**:
+  ```bash
+  curl -X POST https://devserenerocks-claw.hf.space/chat \
+    -H "Content-Type: application/json" \
+    -d '{"message": "Hello, how are you?"}'
+  ```
+- **Locally**:
+  ```bash
+  curl -X POST http://localhost:7860/chat \
+    -H "Content-Type: application/json" \
+    -d '{"message": "Hello, how are you?"}'
+  ```
+
+#### 2. Via Python
+- **Hosted Space**:
+  ```python
+  import requests
+
+  payload = {"message": "Hello, how are you?"}
+  response = requests.post("https://devserenerocks-claw.hf.space/chat", json=payload)
+  print(response.json())
+  ```
+- **Locally**:
+  ```python
+  import requests
+
+  payload = {"message": "Hello, how are you?"}
+  response = requests.post("http://localhost:7860/chat", json=payload)
+  print(response.json())
+  ```
+
+### Expected Response
+```json
+{
+  "response": "Hello! I am doing well, thank you. How can I help you today?"
+}
+```
+

requirements.txt

@@ -1,3 +1,4 @@
 fastapi
 uvicorn[standard]
 huggingface-hub
+pydantic