File size: 2,456 Bytes
95adc09
885f9ce
4c4dfc8
 
95adc09
4c4dfc8
 
885f9ce
 
 
 
 
 
 
 
3857def
95adc09
4c4dfc8
 
 
 
 
 
 
 
 
 
 
95adc09
4c4dfc8
 
 
 
 
 
 
 
 
 
95adc09
4c4dfc8
 
95adc09
4c4dfc8
 
 
 
95adc09
 
4c4dfc8
 
 
95adc09
4c4dfc8
 
 
 
95adc09
4c4dfc8
95adc09
 
 
4c4dfc8
 
 
 
95adc09
4c4dfc8
95adc09
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
import gradio as gr
import torch
import spaces
from transformers import AutoTokenizer, AutoModelForCausalLM

# Load MedScholar model and tokenizer
model_name = "yasserrmd/MedScholar-1.5B"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",
    torch_dtype=torch.float16,
    trust_remote_code=True
)
model.eval()


# Chat function (streaming style)
@spaces.GPU
def respond(message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p):
    # Prepare the full conversation
    conversation = [{"role": "system", "content": system_message}]
    for user_msg, bot_reply in history:
        if user_msg:
            conversation.append({"role": "user", "content": user_msg})
        if bot_reply:
            conversation.append({"role": "assistant", "content": bot_reply})
    conversation.append({"role": "user", "content": message})

    # Convert conversation into prompt string
    prompt = ""
    for turn in conversation:
        if turn["role"] == "system":
            prompt += f"<|system|>\n{turn['content']}\n"
        elif turn["role"] == "user":
            prompt += f"<|user|>\n{turn['content']}\n"
        elif turn["role"] == "assistant":
            prompt += f"<|assistant|>\n{turn['content']}\n"
    prompt += "<|assistant|>\n"

    # Tokenize
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

    # Generate with streaming-like loop
    output_ids = model.generate(
        **inputs,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
        eos_token_id=tokenizer.eos_token_id,
    )

    # Decode and stream the new content
    decoded = tokenizer.decode(output_ids[0], skip_special_tokens=True)
    response = decoded.split("<|assistant|>\n")[-1].strip()
    yield response

# Build Gradio interface
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a helpful medical assistant.", label="System message"),
        gr.Slider(minimum=1, maximum=1024, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=1.5, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
    ],
    title="🩺 MedScholar-1.5B: Medical Chatbot"
)

if __name__ == "__main__":
    demo.launch()