app.py

import os
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# טוען את המודל וה-tokenizer
tokenizer = AutoTokenizer.from_pretrained('dicta-il/dictalm-7b-instruct')
model = AutoModelForCausalLM.from_pretrained('dicta-il/dictalm-7b-instruct', trust_remote_code=True).cuda()

# הגדרת הפונקציה לצ'אט עם המודל
def chat_with_model(prompt):
    model.eval()
    with torch.inference_mode():
        kwargs = dict(
            inputs=tokenizer(prompt, return_tensors='pt').input_ids.to(model.device),
            do_sample=True,
            top_k=50,
            top_p=0.95,
            temperature=0.75,
            max_length=100,
            min_new_tokens=5
        )
        output = model.generate(**kwargs)
        response_text = tokenizer.batch_decode(output, skip_special_tokens=True)[0]
    return response_text

# יצירת ממשק עם Gradio
interface = gr.Interface(fn=chat_with_model, inputs="text", outputs="text", title="Chat with DictaLM Model")
interface.launch()