Spaces:
Running
on
Zero
Running
on
Zero
| import os | |
| import torch | |
| import gradio as gr | |
| from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer | |
| # وصف التطبيق | |
| DESCRIPTION = """\ | |
| # Llama 3.2 3B Instruct (CPU-Only) | |
| هذا نموذج توضيحي لـ [`meta-llama/Llama-3.2-3B-Instruct`](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct) يعمل باستخدام الـ CPU فقط. | |
| """ | |
| # إعداد الثوابت | |
| MAX_MAX_NEW_TOKENS = 2048 | |
| DEFAULT_MAX_NEW_TOKENS = 512 | |
| MAX_INPUT_TOKEN_LENGTH = 4096 # الحد الأقصى لعدد التوكنات في المدخلات | |
| # تحديد الجهاز: استخدام CPU فقط | |
| device = torch.device("cpu") | |
| # تحديد معرف النموذج وتحميله | |
| model_id = "meta-llama/Llama-3.2-3B-Instruct" | |
| # تحميل التوكن الخاص بالنموذج | |
| tokenizer = AutoTokenizer.from_pretrained(model_id) | |
| # تحميل النموذج على CPU مع استخدام torch.float32 | |
| model = AutoModelForCausalLM.from_pretrained( | |
| model_id, | |
| device_map=None, # عدم استخدام GPU | |
| torch_dtype=torch.float32 | |
| ) | |
| model.eval() | |
| model.to(device) | |
| def generate( | |
| message: str, | |
| chat_history: list[dict], | |
| max_new_tokens: int = DEFAULT_MAX_NEW_TOKENS, | |
| temperature: float = 0.6, | |
| top_p: float = 0.9, | |
| top_k: int = 50, | |
| repetition_penalty: float = 1.2, | |
| ): | |
| # دمج سجل المحادثة مع الرسالة الجديدة | |
| conversation = [*chat_history, {"role": "user", "content": message}] | |
| # تحويل المحادثة إلى مدخلات للنموذج | |
| inputs = tokenizer.apply_chat_template( | |
| conversation, | |
| add_generation_prompt=True, | |
| return_tensors="pt" | |
| ) | |
| input_ids = inputs["input_ids"] | |
| # قص التوكنز إذا تجاوز طولها الحد المسموح | |
| if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH: | |
| input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:] | |
| input_ids = input_ids.to(device) | |
| # إعداد البث التدريجي للنص باستخدام TextIteratorStreamer | |
| streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True) | |
| generate_kwargs = dict( | |
| input_ids=input_ids, | |
| streamer=streamer, | |
| max_new_tokens=max_new_tokens, | |
| do_sample=True, | |
| top_p=top_p, | |
| top_k=top_k, | |
| temperature=temperature, | |
| num_beams=1, | |
| repetition_penalty=repetition_penalty, | |
| ) | |
| # تشغيل عملية التوليد على نفس الخيط (CPU) | |
| model.generate(**generate_kwargs) | |
| outputs = [] | |
| # بث النص تدريجيًا أثناء توليد النموذج | |
| for text in streamer: | |
| outputs.append(text) | |
| yield "".join(outputs) | |
| # إنشاء واجهة الدردشة باستخدام Gradio | |
| demo = gr.ChatInterface( | |
| fn=generate, | |
| additional_inputs=[ | |
| 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, | |
| step=0.05, | |
| 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, | |
| examples=[ | |
| ["Hello there! How are you doing?"], | |
| ["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, | |
| type="messages", | |
| description=DESCRIPTION, | |
| css_paths="style.css", # تأكدي من رفع ملف style.css إذا كان موجوداً | |
| fill_height=True, | |
| ) | |
| if __name__ == "__main__": | |
| # استخدام queue() لإدارة الطلبات المتزامنة | |
| demo.queue(max_size=20).launch() | |