stream.py

from flask import Flask, request, jsonify, Response, stream_with_context
from huggingface_hub import InferenceClient
import time

# Initialize Flask app
app = Flask(__name__)

print("\nHello welcome to Sema AI\n", flush=True)  # Flush to ensure immediate output

# Initialize InferenceClient
client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.1")

def format_prompt(message, history):
    prompt = "<s>"
    for user_prompt, bot_response in history:
        prompt += f"[INST] {user_prompt} [/INST]"
        prompt += f" {bot_response}</s> "
    prompt += f"[INST] {message} [/INST]"
    return prompt

def generate(prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0):
    print(f"\nUser: {prompt}\n", flush=True)

    temperature = float(temperature)
    if temperature < 1e-2:
        temperature = 1e-2
    top_p = float(top_p)

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    formatted_prompt = format_prompt(prompt, history)

    try:
        # Get response from Mistral model
        response = client.text_generation(
            formatted_prompt,
            **generate_kwargs,
            stream=True,
            details=True,
            return_full_text=False
        )

        output = ""
        buffer = []
        buffer_size = 5  # Adjust the buffer size as needed

        for token in response:
            buffer.append(token.token.text)
            if len(buffer) >= buffer_size:
                chunk = ''.join(buffer)
                yield chunk
                buffer.clear()
                time.sleep(0.1)  # Introduce a delay to manage the flow of data

        if buffer:
            yield ''.join(buffer)

        # Print AI response
        print(f"\nSema AI: {output}\n, flush=True")
    except Exception as e:
        print(f"Exception during generation: {str(e)}")
        yield "Error occurred"

@app.route("/generate", methods=["POST"])
def generate_text():
    data = request.json
    prompt = data.get("prompt", "")
    history = data.get("history", [])
    temperature = data.get("temperature", 0.9)
    max_new_tokens = data.get("max_new_tokens", 256)
    top_p = data.get("top_p", 0.95)
    repetition_penalty = data.get("repetition_penalty", 1.0)

    try:
        return Response(stream_with_context(generate(
            prompt,
            history,
            temperature=temperature,
            max_new_tokens=max_new_tokens,
            top_p=top_p,
            repetition_penalty=repetition_penalty
        )), content_type='text/plain')
    except Exception as e:
        print(f"Error: {str(e)}")
        return jsonify({"error": str(e)}), 500

if __name__ == "__main__":
    app.run(debug=True, port=5000)