initial version

app.py

@@ -1,63 +1,92 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import os
+from typing import Iterator
+import sambanova
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
+def generate(
+    message: str,
+    chat_history: list[tuple[str, str]],
+    max_new_tokens: int = 1024,
+    temperature: float = 0.6,
+    top_p: float = 0.9,
+    top_k: int = 50,
+    repetition_penalty: float = 1.2,
+) -> Iterator[str]:
+    conversation = []
+    for user, assistant in chat_history:
+        conversation.extend(
+            [
+                {"role": "user", "content": user},
+                {"role": "assistant", "content": assistant},
+            ]
+        )
+    conversation.append({"role": "user", "content": message})
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+    outputs = []
+    for text in sambanova.Streamer(conversation, new_tokens=max_new_tokens,
+                                   temperature=temperature, top_k=top_k, top_p=top_p):
+        outputs.append(text)
+        yield "".join(outputs)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+MAX_MAX_NEW_TOKENS = 2048
+DEFAULT_MAX_NEW_TOKENS = 1024
+MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
 
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
+chat_interface = gr.ChatInterface(
+    fn=generate,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
+            label="Max new tokens",
+            minimum=1,
+            maximum=MAX_MAX_NEW_TOKENS,
+            step=1,
+            value=DEFAULT_MAX_NEW_TOKENS,
+        ),
+        gr.Slider(
+            label="Temperature",
             minimum=0.1,
+            maximum=4.0,
+            step=0.1,
+            value=0.6,
+        ),
+        gr.Slider(
+            label="Top-p (nucleus sampling)",
+            minimum=0.05,
             maximum=1.0,
-            value=0.95,
             step=0.05,
-            label="Top-p (nucleus sampling)",
+            value=0.9,
+        ),
+        gr.Slider(
+            label="Top-k",
+            minimum=1,
+            maximum=1000,
+            step=1,
+            value=50,
+        ),
+        gr.Slider(
+            label="Repetition penalty",
+            minimum=1.0,
+            maximum=2.0,
+            step=0.05,
+            value=1.2,
         ),
     ],
+    stop_btn=None,
+    fill_height=True,
+    examples=[
+        ["Which one is bigger? 4.9 or 4.11"],
+        ["Can you explain briefly to me what is the Python programming language?"],
+        ["Explain the plot of Cinderella in a sentence."],
+        ["How many hours does it take a man to eat a Helicopter?"],
+        ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
+    ],
+    cache_examples=False,
 )
+with gr.Blocks() as demo:
+    gr.Markdown('# Sambanova model inference LLAMA 405B')
 
-
+    chat_interface.render()
+    
 if __name__ == "__main__":
-    demo.launch()
+    demo.queue(max_size=20).launch()

requirements.txt

@@ -1 +1,2 @@
-huggingface_hub==0.22.2
+huggingface_hub==0.22.2
+gradio

sambanova.py

@@ -0,0 +1,89 @@
+import requests
+import json
+import os
+
+def _stream_chat_response(url, headers, payload):
+    """
+    Streams the chat response from the given URL with the specified headers and payload.
+    
+    Args:
+        url (str): The URL to send the POST request to.
+        headers (dict): The headers for the POST request.
+        payload (dict): The payload for the POST request.
+    
+    Raises:
+        InvalidArgument: If the payload does not have the 'stream' key.
+        ConnectionError: If the request fails.
+    
+    Yields:
+        str: The content of the streamed response.
+    """
+    if not payload.get('stream'):
+        raise ValueError('This method can only handle stream payload')
+    
+    try:
+        # Make the POST request
+        response = requests.post(url, headers=headers, json=payload, stream=True)
+        response.raise_for_status()  # Raise an error for bad status codes
+
+        # Process the streamed response
+        for line in response.iter_lines():
+            if line:
+                decoded_line = line.decode('utf-8')
+                DATA_PREFIX = "data: "
+                if decoded_line.startswith(DATA_PREFIX):
+                    decoded_line = decoded_line[len(DATA_PREFIX):]  # Remove the "data: " prefix
+                    if decoded_line.strip() == "[DONE]":
+                        break
+                    try:
+                        json_data = json.loads(decoded_line)
+                        content = json_data.get('choices', [{}])[0].get('delta', {}).get('content', '')
+                        if content:
+                            yield content
+                    except json.JSONDecodeError as e:
+                        print(f"Warning: Error decoding JSON: {decoded_line}. Skipping this line.")
+    except requests.RequestException as e:
+        raise ConnectionError(f"Request failed: {e}") from e
+
+def Streamer(history, **kwargs):
+    """
+    Streams the chat response based on the provided history and additional kwargs.
+    
+    Args:
+        history (dict): The chat history.
+        **kwargs: Additional parameters to update the payload.
+    
+    Yields:
+        str: The content of the streamed response.
+    """
+    url = os.getenv('URL')
+    token = os.getenv('TOKEN')
+    
+    if not url or not token:
+        raise EnvironmentError("URL or TOKEN environment variable is not set.")
+    
+    headers = {
+        "Authorization": f"Basic {token}",
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "messages": history,
+        "max_tokens": 1000,
+        "stop": ["<|eot_id|>"],
+        "model": "llama3-405b",
+        "stream": True
+    }
+    payload.update(kwargs)
+
+    for update in _stream_chat_response(url, headers, payload):
+        yield update
+
+# Example usage
+if __name__ == "__main__":
+    try:
+        history = [{"role": "user", "content": "Tell me a joke"}]
+
+        for content in Streamer(history):
+            print(content, end='')
+    except Exception as e:
+        print(f"An error occurred: {e}")