AnodicConsulting commited on
Commit
bb54aff
·
verified ·
1 Parent(s): d032441

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +36 -8
app.py CHANGED
@@ -1,10 +1,38 @@
1
- from fastapi import FastAPI
2
- from transformers import pipeline
 
3
 
4
- app = FastAPI()
5
- generator = pipeline("text-generation", model="gpt2")
6
 
7
- @app.post("/chat")
8
- def chat(prompt: str):
9
- result = generator(prompt, max_new_tokens=100)
10
- return {"response": result[0]["generated_text"]}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+ from transformers import AutoTokenizer, AutoModelForCausalLM
3
+ import torch
4
 
5
+ MODEL_ID = "distilgpt2" # fast + small + reliable on CPU
 
6
 
7
+ tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
8
+ model = AutoModelForCausalLM.from_pretrained(MODEL_ID)
9
+
10
+ def respond(message, history):
11
+ # Build a simple prompt from chat history
12
+ prompt = ""
13
+ for user, bot in history:
14
+ prompt += f"User: {user}\nBot: {bot}\n"
15
+ prompt += f"User: {message}\nBot:"
16
+
17
+ inputs = tokenizer(prompt, return_tensors="pt")
18
+ with torch.no_grad():
19
+ output_ids = model.generate(
20
+ **inputs,
21
+ max_new_tokens=120,
22
+ do_sample=True,
23
+ temperature=0.8,
24
+ top_p=0.95,
25
+ pad_token_id=tokenizer.eos_token_id,
26
+ )
27
+
28
+ decoded = tokenizer.decode(output_ids[0], skip_special_tokens=True)
29
+ reply = decoded.split("Bot:")[-1].strip()
30
+ return reply
31
+
32
+ demo = gr.ChatInterface(
33
+ fn=respond,
34
+ title="🤖 Fast Chat",
35
+ description="Lightweight CPU-friendly chatbot running on Hugging Face Spaces."
36
+ )
37
+
38
+ demo.launch()