From 9504f917e219857dc3e0ab2a8c4c18cd7d8b5c10 Mon Sep 17 00:00:00 2001 From: CRlaptop Date: Mon, 25 Nov 2024 14:32:10 +0800 Subject: [PATCH] =?UTF-8?q?=E6=8E=A5=E5=85=A5Hugging=20Face=20Playground?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- config.py | 7 +- request_llms/bridge_all.py | 34 +++++++++ request_llms/bridge_hf.py | 79 +++++++++++++++++++++ request_llms/com_hfplayground.py | 117 +++++++++++++++++++++++++++++++ tests/test_llms.py | 6 +- 5 files changed, 238 insertions(+), 5 deletions(-) create mode 100644 request_llms/bridge_hf.py create mode 100644 request_llms/com_hfplayground.py diff --git a/config.py b/config.py index 49c0e3872b..cfe2745186 100644 --- a/config.py +++ b/config.py @@ -36,7 +36,8 @@ "gpt-4o", "gpt-4o-mini", "gpt-4-turbo", "gpt-4-turbo-2024-04-09", "gpt-3.5-turbo-1106", "gpt-3.5-turbo-16k", "gpt-3.5-turbo", "azure-gpt-3.5", "gpt-4", "gpt-4-32k", "azure-gpt-4", "glm-4", "glm-4v", "glm-3-turbo", - "gemini-1.5-pro", "chatglm3" + "gemini-1.5-pro", "chatglm3", "HF:Qwen/Qwen2.5-72B-Instruct", + "HF:meta-llama/Meta-Llama-3-70B-Instruct" ] EMBEDDING_MODEL = "text-embedding-3-small" @@ -55,6 +56,7 @@ # "deepseek-chat" ,"deepseek-coder", # "gemini-1.5-flash", # "yi-34b-chat-0205","yi-34b-chat-200k","yi-large","yi-medium","yi-spark","yi-large-turbo","yi-large-preview", +# Hugging Face Playground中的模型，可自行添加，模型名字前需加"HF:" # ] # --- --- --- --- # 此外，您还可以在接入one-api/vllm/ollama/Openroute时， @@ -259,8 +261,7 @@ # Google Gemini API-Key GEMINI_API_KEY = '' - -# HUGGINGFACE的TOKEN，下载LLAMA时起作用 https://huggingface.co/docs/hub/security-tokens +# HUGGINGFACE的TOKEN，使用Playground及下载LLAMA时起作用 https://huggingface.co/docs/hub/security-tokens HUGGINGFACE_ACCESS_TOKEN = "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" diff --git a/request_llms/bridge_all.py b/request_llms/bridge_all.py index 34d05cac92..75ea7b17d2 100644 --- a/request_llms/bridge_all.py +++ b/request_llms/bridge_all.py @@ -35,6 +35,9 @@ from .bridge_zhipu import predict_no_ui_long_connection as zhipu_noui from .bridge_zhipu import predict as zhipu_ui +from .bridge_hf import predict_no_ui_long_connection as hf_noui +from .bridge_hf import predict as hf_ui + from .bridge_taichu import predict_no_ui_long_connection as taichu_noui from .bridge_taichu import predict as taichu_ui @@ -987,6 +990,37 @@ def decode(self, *args, **kwargs): }) except: logger.error(trimmed_format_exc()) +# -=-=-=-=-=-=- Hugging Face Playground -=-=-=-=-=-=- +if any("HF:" in x for x in AVAIL_LLM_MODELS): + try: + for x in AVAIL_LLM_MODELS: + if x == "HF:Qwen/Qwen2.5-72B-Instruct" or x == 'HF:meta-llama/Meta-Llama-3-70B-Instruct': + model_info.update({ + x: { + "fn_with_ui": hf_ui, + "fn_without_ui": hf_noui, + "endpoint": None, + "max_token": 8192, + 'can_multi_thread': True, + "tokenizer": tokenizer_gpt35, + "token_cnt": get_token_num_gpt35, + }, + }) + elif "HF:" in x: # 默认max_token=8192 + model_info.update({ + x: { + "fn_with_ui": hf_ui, + "fn_without_ui": hf_noui, + "endpoint": None, + "max_token": 8192, + 'can_multi_thread': True, + "tokenizer": tokenizer_gpt35, + "token_cnt": get_token_num_gpt35, + }, + }) + except: + logger.error(trimmed_format_exc()) + # -=-=-=-=-=-=- 幻方-深度求索大模型 -=-=-=-=-=-=- if "deepseekcoder" in AVAIL_LLM_MODELS: # deepseekcoder try: diff --git a/request_llms/bridge_hf.py b/request_llms/bridge_hf.py new file mode 100644 index 0000000000..68ac414801 --- /dev/null +++ b/request_llms/bridge_hf.py @@ -0,0 +1,79 @@ +import time +import os +from toolbox import update_ui, get_conf, update_ui_lastest_msg, log_chat +from toolbox import check_packages, report_exception, have_any_recent_upload_image_files +from toolbox import ChatBotWithCookies + +model_name = 'Hugging Face Playground' + + +def validate_key(): + HUGGINGFACE_ACCESS_TOKEN = get_conf("HUGGINGFACE_ACCESS_TOKEN") + if HUGGINGFACE_ACCESS_TOKEN == '': return False + return True + + +def predict_no_ui_long_connection(inputs: str, llm_kwargs: dict, history: list = [], sys_prompt: str = "", + observe_window: list = [], console_slience: bool = False): + """ + ⭐多线程方法 + 函数的说明请见 request_llms/bridge_all.py + """ + watch_dog_patience = 5 + response = "" + + + if validate_key() is False: + raise RuntimeError('请配置HUGGINGFACE_ACCESS_TOKEN') + + # 开始接收回复 + from .com_hfplayground import HFPlaygroundInit + hfp_init = HFPlaygroundInit() + for chunk, response in hfp_init.generate_chat(inputs, llm_kwargs, history, sys_prompt): + if len(observe_window) >= 1: + observe_window[0] = response + if len(observe_window) >= 2: + if (time.time() - observe_window[1]) > watch_dog_patience: + raise RuntimeError("程序终止。") + return response + + +def predict(inputs: str, llm_kwargs: dict, plugin_kwargs: dict, chatbot: ChatBotWithCookies, + history: list = [], system_prompt: str = '', stream: bool = True, additional_fn: str = None): + """ + ⭐单线程方法 + 函数的说明请见 request_llms/bridge_all.py + """ + chatbot.append([inputs, ""]) + yield from update_ui(chatbot=chatbot, history=history) + + # 尝试导入依赖，如果缺少依赖，则给出安装建议 + try: + check_packages(["openai"]) + except: + yield from update_ui_lastest_msg( + f"导入软件依赖失败。使用该模型需要额外依赖，安装方法```pip install --upgrade openai```。", + chatbot=chatbot, history=history, delay=0) + return + + if validate_key() is False: + yield from update_ui_lastest_msg(lastmsg="[Local Message] 请配置HUGGINGFACE_ACCESS_TOKEN", chatbot=chatbot, + history=history, delay=0) + return + + if additional_fn is not None: + from core_functional import handle_core_functionality + inputs, history = handle_core_functionality(additional_fn, inputs, history, chatbot) + chatbot[-1] = [inputs, ""] + yield from update_ui(chatbot=chatbot, history=history) + + + # 开始接收回复 + from .com_hfplayground import HFPlaygroundInit + hfp_init = HFPlaygroundInit() + for chunk, response in hfp_init.generate_chat(inputs, llm_kwargs, history, system_prompt): + chatbot[-1] = [inputs, response] + yield from update_ui(chatbot=chatbot, history=history) + history.extend([inputs, response]) + log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response) + yield from update_ui(chatbot=chatbot, history=history) diff --git a/request_llms/com_hfplayground.py b/request_llms/com_hfplayground.py new file mode 100644 index 0000000000..ba361c8e49 --- /dev/null +++ b/request_llms/com_hfplayground.py @@ -0,0 +1,117 @@ +import httpx +from huggingface_hub import InferenceClient +from toolbox import get_conf, encode_image, get_pictures_list +from loguru import logger +import os + +proxies = get_conf("proxies") + +def input_encode_handler(inputs: str, llm_kwargs: dict): + if llm_kwargs["most_recent_uploaded"].get("path"): + image_paths = get_pictures_list(llm_kwargs["most_recent_uploaded"]["path"]) + md_encode = [] + for md_path in image_paths: + type_ = os.path.splitext(md_path)[1].replace(".", "") + type_ = "jpeg" if type_ == "jpg" else type_ + md_encode.append({"data": encode_image(md_path), "type": type_}) + return inputs, md_encode + + +class HFPlaygroundInit: + + def __init__(self): + HUGGINGFACE_ACCESS_TOKEN = get_conf("HUGGINGFACE_ACCESS_TOKEN") + self.client = InferenceClient(api_key=HUGGINGFACE_ACCESS_TOKEN, proxies=proxies) + self.model = '' + + def __conversation_user(self, user_input: str, llm_kwargs: dict): + return {"role": "user", "content": user_input} + + def __conversation_history(self, history: list, llm_kwargs: dict): + messages = [] + conversation_cnt = len(history) // 2 + if conversation_cnt: + for index in range(0, 2 * conversation_cnt, 2): + what_i_have_asked = self.__conversation_user(history[index], llm_kwargs) + what_gpt_answer = { + "role": "assistant", + "content": history[index + 1] + } + messages.append(what_i_have_asked) + messages.append(what_gpt_answer) + return messages + + @staticmethod + def preprocess_param(param, default=0.95, min_val=0.01, max_val=0.99): + """预处理参数，保证其在允许范围内，并处理精度问题""" + try: + param = float(param) + except ValueError: + return default + + if param <= min_val: + return min_val + elif param >= max_val: + return max_val + else: + return round(param, 2) # 可挑选精度，目前是两位小数 + + def __conversation_message_payload(self, inputs: str, llm_kwargs: dict, history: list, system_prompt: str): + messages = [] + if system_prompt: + messages.append({"role": "system", "content": system_prompt}) + self.model = llm_kwargs['llm_model'].replace("HF:", "").strip() + messages.extend(self.__conversation_history(history, llm_kwargs)) # 处理 history + if inputs.strip() == "": # 处理空输入导致报错的问题 https://github.com/binary-husky/gpt_academic/issues/1640 提示 {"error":{"code":"1214","message":"messages[1]:content和tool_calls 字段不能同时为空"} + inputs = "." # 空格、换行、空字符串都会报错，所以用最没有意义的一个点代替 + messages.append(self.__conversation_user(inputs, llm_kwargs)) # 处理用户对话 + """ + 采样温度，控制输出的随机性，必须为正数 + 取值范围是：(0.0, 1.0)，不能等于 0，默认值为 0.95， + 值越大，会使输出更随机，更具创造性； + 值越小，输出会更加稳定或确定 + 建议您根据应用场景调整 top_p 或 temperature 参数，但不要同时调整两个参数 + """ + temperature = self.preprocess_param( + param=llm_kwargs.get('temperature', 0.95), + default=0.95, + min_val=0.01, + max_val=0.99 + ) + """ + 用温度取样的另一种方法，称为核取样 + 取值范围是：(0.0, 1.0) 开区间， + 不能等于 0 或 1，默认值为 0.7 + 模型考虑具有 top_p 概率质量 tokens 的结果 + 例如：0.1 意味着模型解码器只考虑从前 10% 的概率的候选集中取 tokens + 建议您根据应用场景调整 top_p 或 temperature 参数， + 但不要同时调整两个参数 + """ + top_p = self.preprocess_param( + param=llm_kwargs.get('top_p', 0.70), + default=0.70, + min_val=0.01, + max_val=0.99 + ) + response = self.client.chat.completions.create( + model=self.model, messages=messages, stream=True, + temperature=temperature, + top_p=top_p, + max_tokens=llm_kwargs.get('max_tokens', 1024 * 4), + ) + return response + + def generate_chat(self, inputs: str, llm_kwargs: dict, history: list, system_prompt: str): + self.model = llm_kwargs['llm_model'].replace("HF:", "").strip() + response = self.__conversation_message_payload(inputs, llm_kwargs, history, system_prompt) + bro_results = '' + for chunk in response: + bro_results += chunk.choices[0].delta.content + yield chunk.choices[0].delta.content, bro_results + + +if __name__ == '__main__': + HFP = HFPlaygroundInit() + r = HFP.generate_chat('你好', {'llm_model': 'Qwen/Qwen2.5-72B-Instruct'}, [], '你是WPSAi') + for i in r: + print(i) diff --git a/tests/test_llms.py b/tests/test_llms.py index 20d257875e..6970bd0ed7 100644 --- a/tests/test_llms.py +++ b/tests/test_llms.py @@ -14,13 +14,15 @@ def validate_path(): if "在线模型": if __name__ == "__main__": - from request_llms.bridge_taichu import predict_no_ui_long_connection + from request_llms.bridge_hf import predict_no_ui_long_connection + + # from request_llms.bridge_taichu import predict_no_ui_long_connection # from request_llms.bridge_cohere import predict_no_ui_long_connection # from request_llms.bridge_spark import predict_no_ui_long_connection # from request_llms.bridge_zhipu import predict_no_ui_long_connection # from request_llms.bridge_chatglm3 import predict_no_ui_long_connection llm_kwargs = { - "llm_model": "taichu", + "llm_model": "Qwen/Qwen2.5-72B-Instruct", "max_length": 4096, "top_p": 1, "temperature": 1,