Spaces:
Sleeping
Sleeping
| import os | |
| from threading import Thread | |
| from typing import Iterator | |
| import gradio as gr | |
| # from gradio import MultimodalTextbox | |
| # from gradio.data_classes import FileData | |
| import spaces | |
| import torch | |
| from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer | |
| from typing_extensions import NotRequired, TypedDict | |
| MAX_MAX_NEW_TOKENS = 1024 | |
| DEFAULT_MAX_NEW_TOKENS = 256 | |
| MAX_INPUT_TOKEN_LENGTH = 50000 | |
| DESCRIPTION = """\ | |
| # Yam-Peleg's Hebrew-Mistral-7B-200K | |
| Hebrew-Mistral-7B-200K was introduced in [this Facebook post](https://www.facebook.com/groups/MDLI1/posts/2708679492629415/). | |
| Please, check the [original model card](https://huggingface.co/yam-peleg/Hebrew-Mistral-7B-200K) for more details. | |
| You can see the other Hebrew models by Yam [here](https://huggingface.co/collections/yam-peleg/hebrew-models-65e957875324e2b9a4b68f08) | |
| ## While the user interface is of a chatbot for convenience, this is a base model and is not fine-tuned for chatbot tasks or instruction following tasks. | |
| """ | |
| LICENSE = """ | |
| <p/> | |
| --- | |
| A derivative work of [mistral-7b](https://mistral.ai/news/announcing-mistral-7b/) by Mistral-AI. | |
| The model and space are released under the Apache 2.0 license | |
| This demo Space was created by [Doron Adler](https://linktr.ee/Norod78) | |
| """ | |
| if not torch.cuda.is_available(): | |
| DESCRIPTION += "\n<p>Running on CPU ๐ฅถ This demo does not work on CPU.</p>" | |
| if torch.cuda.is_available(): | |
| model_id = "yam-peleg/Hebrew-Mistral-7B-200K" | |
| model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.bfloat16, low_cpu_mem_usage=True) | |
| tokenizer_id = "yam-peleg/Hebrew-Mistral-7B-200K" | |
| tokenizer = AutoTokenizer.from_pretrained(tokenizer_id) | |
| def generate( | |
| message: str, | |
| chat_history: list[tuple[str, str]], | |
| max_new_tokens: int = 1024, | |
| temperature: float = 0.2, | |
| top_p: float = 0.7, | |
| top_k: int = 30, | |
| repetition_penalty: float = 1.0, | |
| ) -> Iterator[str]: | |
| historical_text = "" | |
| #Prepend the entire chat history to the message with new lines between each message | |
| for user, assistant in chat_history: | |
| historical_text += f"\n{user}\n{assistant}" | |
| if len(historical_text) > 0: | |
| message = historical_text + f"\n{message}" | |
| input_ids = tokenizer([message], return_tensors="pt").input_ids | |
| if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH: | |
| input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:] | |
| gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.") | |
| input_ids = input_ids.to(model.device) | |
| streamer = TextIteratorStreamer(tokenizer, timeout=10.0, 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, | |
| pad_token_id = tokenizer.eos_token_id, | |
| repetition_penalty=repetition_penalty, | |
| no_repeat_ngram_size=5, | |
| early_stopping=False, | |
| ) | |
| t = Thread(target=model.generate, kwargs=generate_kwargs) | |
| t.start() | |
| outputs = [] | |
| for text in streamer: | |
| outputs.append(text) | |
| yield "".join(outputs) | |
| chat_interface = gr.ChatInterface( | |
| fn=generate, | |
| chatbot=gr.Chatbot(rtl=True, show_copy_button=True), | |
| textbox=gr.Textbox(text_align = 'right', rtl = True), | |
| 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=0.9, | |
| ), | |
| gr.Slider( | |
| label="Top-p (nucleus sampling)", | |
| minimum=0.05, | |
| maximum=1.0, | |
| step=0.05, | |
| value=0.7, | |
| ), | |
| gr.Slider( | |
| label="Top-k", | |
| minimum=1, | |
| maximum=1000, | |
| step=1, | |
| value=40, | |
| ), | |
| ], | |
| stop_btn=None, | |
| examples=[ | |
| ["ืืชืืื ืืขืืืช ืฉืืงืืื:"], | |
| ["ืฉืคืช ืืชืื ืืช ืคืืืืื ืืื"], | |
| ["ืืืืฉ ืืืืจืื ืืขืืื ืืฉื ืืื ืืืืจื, ืืฉืืคืชืข"], | |
| ["ืฉืืื: ืืื ืขืืจ ืืืืจื ืฉื ืืืื ืช ืืฉืจืื?\nืชืฉืืื:"], | |
| ["ืฉืืื: ืื ื ืืืฉ ืขืืืฃ, ืื ืืืื ืื ืืขืฉืืช?\nืชืฉืืื:"], | |
| ], | |
| ) | |
| with gr.Blocks(css="style.css") as demo: | |
| gr.Markdown(DESCRIPTION) | |
| gr.DuplicateButton(value="Duplicate Space for private use", elem_id="duplicate-button") | |
| chat_interface.render() | |
| gr.Markdown(LICENSE) | |
| if __name__ == "__main__": | |
| demo.queue(max_size=20).launch() |