From bfacdb9c3b0038c074895f56b53fa65d9ff460da Mon Sep 17 00:00:00 2001 From: Han Fangyuan Date: Sat, 9 Dec 2023 12:39:09 +0800 Subject: [PATCH 1/4] feat: support character description of ali qwen model --- bot/tongyi/ali_qwen_session.py | 62 ++++++++++++++++++++++++++++++++++ bot/tongyi/tongyi_qwen_bot.py | 15 ++++++-- 2 files changed, 74 insertions(+), 3 deletions(-) create mode 100644 bot/tongyi/ali_qwen_session.py diff --git a/bot/tongyi/ali_qwen_session.py b/bot/tongyi/ali_qwen_session.py new file mode 100644 index 0000000..0eb1c4a --- /dev/null +++ b/bot/tongyi/ali_qwen_session.py @@ -0,0 +1,62 @@ +from bot.session_manager import Session +from common.log import logger + +""" + e.g. + [ + {"role": "system", "content": "You are a helpful assistant."}, + {"role": "user", "content": "Who won the world series in 2020?"}, + {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."}, + {"role": "user", "content": "Where was it played?"} + ] +""" + +class AliQwenSession(Session): + def __init__(self, session_id, system_prompt=None, model="qianwen"): + super().__init__(session_id, system_prompt) + self.model = model + self.reset() + + def discard_exceeding(self, max_tokens, cur_tokens=None): + precise = True + try: + cur_tokens = self.calc_tokens() + except Exception as e: + precise = False + if cur_tokens is None: + raise e + logger.debug("Exception when counting tokens precisely for query: {}".format(e)) + while cur_tokens > max_tokens: + if len(self.messages) > 2: + self.messages.pop(1) + elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant": + self.messages.pop(1) + if precise: + cur_tokens = self.calc_tokens() + else: + cur_tokens = cur_tokens - max_tokens + break + elif len(self.messages) == 2 and self.messages[1]["role"] == "user": + logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens)) + break + else: + logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens, len(self.messages))) + break + if precise: + cur_tokens = self.calc_tokens() + else: + cur_tokens = cur_tokens - max_tokens + return cur_tokens + + def calc_tokens(self): + return num_tokens_from_messages(self.messages, self.model) + +def num_tokens_from_messages(messages, model): + """Returns the number of tokens used by a list of messages.""" + # 官方token计算规则："对于中文文本来说，1个token通常对应一个汉字；对于英文文本来说，1个token通常对应3至4个字母或1个单词" + # 详情请产看文档：https://help.aliyun.com/document_detail/2586397.html + # 目前根据字符串长度粗略估计token数，不影响正常使用 + tokens = 0 + for msg in messages: + tokens += len(msg["content"]) + return tokens diff --git a/bot/tongyi/tongyi_qwen_bot.py b/bot/tongyi/tongyi_qwen_bot.py index 585cb47..df99564 100644 --- a/bot/tongyi/tongyi_qwen_bot.py +++ b/bot/tongyi/tongyi_qwen_bot.py @@ -10,7 +10,7 @@ import broadscope_bailian from broadscope_bailian import ChatQaMessage from bot.bot import Bot -from bot.baidu.baidu_wenxin_session import BaiduWenxinSession +from bot.tongyi.ali_qwen_session import AliQwenSession from bot.session_manager import SessionManager from bridge.context import ContextType from bridge.reply import Reply, ReplyType @@ -27,7 +27,7 @@ class TongyiQwenBot(Bot): self.node_id = conf().get("qwen_node_id") or "" self.api_key_client = broadscope_bailian.AccessTokenClient(access_key_id=self.access_key_id, access_key_secret=self.access_key_secret) self.api_key_expired_time = self.set_api_key() - self.sessions = SessionManager(BaiduWenxinSession, model=conf().get("model") or "qwen") + self.sessions = SessionManager(AliQwenSession, model=conf().get("model") or "qwen") self.temperature = conf().get("temperature", 0.2) # 值在[0,1]之间，越大表示回复越具有不确定性 self.top_p = conf().get("top_p", 1) @@ -76,7 +76,7 @@ class TongyiQwenBot(Bot): reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type)) return reply - def reply_text(self, session: BaiduWenxinSession, retry_count=0) -> dict: + def reply_text(self, session: AliQwenSession, retry_count=0) -> dict: """ call bailian's ChatCompletion to get the answer :param session: a conversation session @@ -140,6 +140,7 @@ class TongyiQwenBot(Bot): history = [] user_content = '' assistant_content = '' + system_content = '' for message in messages: role = message.get('role') if role == 'user': @@ -149,8 +150,16 @@ class TongyiQwenBot(Bot): history.append(ChatQaMessage(user_content, assistant_content)) user_content = '' assistant_content = '' + elif role =='system': + system_content += message.get('content') if user_content == '': raise Exception('no user message') + if system_content != '': + # NOTE 模拟系统消息，测试发现人格描述以"你需要扮演ChatGPT"开头能够起作用，而以"你是ChatGPT"开头模型会直接否认 + system_qa = ChatQaMessage(system_content, '好的，我会严格按照你的设定回答问题') + history.insert(0, system_qa) + logger.debug("[TONGYI] converted qa messages: {}".format([item.to_dict() for item in history])) + logger.debug("[TONGYI] user content as prompt: {}".format(user_content)) return user_content, history def get_completion_content(self, response, node_id): From 207fa1d019fa9cae74fe356aaaae2c6c135969d5 Mon Sep 17 00:00:00 2001 From: Han Fangyuan Date: Sat, 9 Dec 2023 18:40:17 +0800 Subject: [PATCH 2/4] feat: hot reload conf of ali qwen model --- bot/tongyi/tongyi_qwen_bot.py | 42 +++++++++++++++++++++++++---------- plugins/godcmd/godcmd.py | 4 ++-- 2 files changed, 32 insertions(+), 14 deletions(-) diff --git a/bot/tongyi/tongyi_qwen_bot.py b/bot/tongyi/tongyi_qwen_bot.py index df99564..21c50ca 100644 --- a/bot/tongyi/tongyi_qwen_bot.py +++ b/bot/tongyi/tongyi_qwen_bot.py @@ -15,21 +15,38 @@ from bot.session_manager import SessionManager from bridge.context import ContextType from bridge.reply import Reply, ReplyType from common.log import logger +from common import const from config import conf, load_config class TongyiQwenBot(Bot): def __init__(self): super().__init__() - self.access_key_id = conf().get("qwen_access_key_id") - self.access_key_secret = conf().get("qwen_access_key_secret") - self.agent_key = conf().get("qwen_agent_key") - self.app_id = conf().get("qwen_app_id") - self.node_id = conf().get("qwen_node_id") or "" - self.api_key_client = broadscope_bailian.AccessTokenClient(access_key_id=self.access_key_id, access_key_secret=self.access_key_secret) self.api_key_expired_time = self.set_api_key() - self.sessions = SessionManager(AliQwenSession, model=conf().get("model") or "qwen") - self.temperature = conf().get("temperature", 0.2) # 值在[0,1]之间，越大表示回复越具有不确定性 - self.top_p = conf().get("top_p", 1) + self.sessions = SessionManager(AliQwenSession, model=conf().get("model", const.QWEN)) + + def api_key_client(self): + return broadscope_bailian.AccessTokenClient(access_key_id=self.access_key_id(), access_key_secret=self.access_key_secret()) + + def access_key_id(self): + return conf().get("qwen_access_key_id") + + def access_key_secret(self): + return conf().get("qwen_access_key_secret") + + def agent_key(self): + return conf().get("qwen_agent_key") + + def app_id(self): + return conf().get("qwen_app_id") + + def node_id(self): + return conf().get("qwen_node_id", "") + + def temperature(self): + return conf().get("temperature", 0.2 ) + + def top_p(self): + return conf().get("top_p", 1) def reply(self, query, context=None): # acquire reply content @@ -87,8 +104,8 @@ class TongyiQwenBot(Bot): prompt, history = self.convert_messages_format(session.messages) self.update_api_key_if_expired() # NOTE 阿里百炼的call()函数参数比较奇怪, top_k参数表示top_p, top_p参数表示temperature, 可以参考文档 https://help.aliyun.com/document_detail/2587502.htm - response = broadscope_bailian.Completions().call(app_id=self.app_id, prompt=prompt, history=history, top_k=self.top_p, top_p=self.temperature) - completion_content = self.get_completion_content(response, self.node_id) + response = broadscope_bailian.Completions().call(app_id=self.app_id(), prompt=prompt, history=history, top_k=self.top_p(), top_p=self.temperature()) + completion_content = self.get_completion_content(response, self.node_id()) completion_tokens, total_tokens = self.calc_tokens(session.messages, completion_content) return { "total_tokens": total_tokens, @@ -129,9 +146,10 @@ class TongyiQwenBot(Bot): return result def set_api_key(self): - api_key, expired_time = self.api_key_client.create_token(agent_key=self.agent_key) + api_key, expired_time = self.api_key_client().create_token(agent_key=self.agent_key()) broadscope_bailian.api_key = api_key return expired_time + def update_api_key_if_expired(self): if time.time() > self.api_key_expired_time: self.api_key_expired_time = self.set_api_key() diff --git a/plugins/godcmd/godcmd.py b/plugins/godcmd/godcmd.py index 03a96bd..15c05a1 100644 --- a/plugins/godcmd/godcmd.py +++ b/plugins/godcmd/godcmd.py @@ -313,7 +313,7 @@ class Godcmd(Plugin): except Exception as e: ok, result = False, "你没有设置私有GPT模型" elif cmd == "reset": - if bottype in [const.OPEN_AI, const.CHATGPT, const.CHATGPTONAZURE, const.LINKAI, const.BAIDU, const.XUNFEI]: + if bottype in [const.OPEN_AI, const.CHATGPT, const.CHATGPTONAZURE, const.LINKAI, const.BAIDU, const.XUNFEI, const.QWEN]: bot.sessions.clear_session(session_id) if Bridge().chat_bots.get(bottype): Bridge().chat_bots.get(bottype).sessions.clear_session(session_id) @@ -339,7 +339,7 @@ class Godcmd(Plugin): ok, result = True, "配置已重载" elif cmd == "resetall": if bottype in [const.OPEN_AI, const.CHATGPT, const.CHATGPTONAZURE, const.LINKAI, - const.BAIDU, const.XUNFEI]: + const.BAIDU, const.XUNFEI, const.QWEN]: channel.cancel_all_session() bot.sessions.clear_all_session() ok, result = True, "重置所有会话成功" From c8910b8e148e1a87a74d270ded487bd21bb8d975 Mon Sep 17 00:00:00 2001 From: Han Fangyuan Date: Sat, 9 Dec 2023 19:26:11 +0800 Subject: [PATCH 3/4] fix: set correct top_p params of ali qwen model --- bot/tongyi/tongyi_qwen_bot.py | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/bot/tongyi/tongyi_qwen_bot.py b/bot/tongyi/tongyi_qwen_bot.py index 21c50ca..19bbc82 100644 --- a/bot/tongyi/tongyi_qwen_bot.py +++ b/bot/tongyi/tongyi_qwen_bot.py @@ -103,8 +103,8 @@ class TongyiQwenBot(Bot): try: prompt, history = self.convert_messages_format(session.messages) self.update_api_key_if_expired() - # NOTE 阿里百炼的call()函数参数比较奇怪, top_k参数表示top_p, top_p参数表示temperature, 可以参考文档 https://help.aliyun.com/document_detail/2587502.htm - response = broadscope_bailian.Completions().call(app_id=self.app_id(), prompt=prompt, history=history, top_k=self.top_p(), top_p=self.temperature()) + # NOTE 阿里百炼的call()函数未提供temperature参数，考虑到temperature和top_p参数作用相同，取两者较小的值作为top_p参数传入，详情见文档 https://help.aliyun.com/document_detail/2587502.htm + response = broadscope_bailian.Completions().call(app_id=self.app_id(), prompt=prompt, history=history, top_p=min(self.temperature(), self.top_p())) completion_content = self.get_completion_content(response, self.node_id()) completion_tokens, total_tokens = self.calc_tokens(session.messages, completion_content) return { @@ -181,6 +181,8 @@ class TongyiQwenBot(Bot): return user_content, history def get_completion_content(self, response, node_id): + if not response['Success']: + return f"[ERROR]\n{response['Code']}:{response['Message']}" text = response['Data']['Text'] if node_id == '': return text From 9838979bbdcfce9527fb216fc0cc4a1d8735a14c Mon Sep 17 00:00:00 2001 From: Han Fangyuan Date: Sat, 9 Dec 2023 19:40:07 +0800 Subject: [PATCH 4/4] refactor: update class name of qwen bot --- .../ali_qwen_bot.py} | 28 +++++++++---------- bot/{tongyi => ali}/ali_qwen_session.py | 0 bot/bot_factory.py | 4 +-- 3 files changed, 16 insertions(+), 16 deletions(-) rename bot/{tongyi/tongyi_qwen_bot.py => ali/ali_qwen_bot.py} (89%) rename bot/{tongyi => ali}/ali_qwen_session.py (100%) diff --git a/bot/tongyi/tongyi_qwen_bot.py b/bot/ali/ali_qwen_bot.py similarity index 89% rename from bot/tongyi/tongyi_qwen_bot.py rename to bot/ali/ali_qwen_bot.py index 19bbc82..ae9d767 100644 --- a/bot/tongyi/tongyi_qwen_bot.py +++ b/bot/ali/ali_qwen_bot.py @@ -10,7 +10,7 @@ import broadscope_bailian from broadscope_bailian import ChatQaMessage from bot.bot import Bot -from bot.tongyi.ali_qwen_session import AliQwenSession +from bot.ali.ali_qwen_session import AliQwenSession from bot.session_manager import SessionManager from bridge.context import ContextType from bridge.reply import Reply, ReplyType @@ -18,7 +18,7 @@ from common.log import logger from common import const from config import conf, load_config -class TongyiQwenBot(Bot): +class AliQwenBot(Bot): def __init__(self): super().__init__() self.api_key_expired_time = self.set_api_key() @@ -51,7 +51,7 @@ class TongyiQwenBot(Bot): def reply(self, query, context=None): # acquire reply content if context.type == ContextType.TEXT: - logger.info("[TONGYI] query={}".format(query)) + logger.info("[QWEN] query={}".format(query)) session_id = context["session_id"] reply = None @@ -68,11 +68,11 @@ class TongyiQwenBot(Bot): if reply: return reply session = self.sessions.session_query(query, session_id) - logger.debug("[TONGYI] session query={}".format(session.messages)) + logger.debug("[QWEN] session query={}".format(session.messages)) reply_content = self.reply_text(session) logger.debug( - "[TONGYI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format( + "[QWEN] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format( session.messages, session_id, reply_content["content"], @@ -86,7 +86,7 @@ class TongyiQwenBot(Bot): reply = Reply(ReplyType.TEXT, reply_content["content"]) else: reply = Reply(ReplyType.ERROR, reply_content["content"]) - logger.debug("[TONGYI] reply {} used 0 tokens.".format(reply_content)) + logger.debug("[QWEN] reply {} used 0 tokens.".format(reply_content)) return reply else: @@ -116,31 +116,31 @@ class TongyiQwenBot(Bot): need_retry = retry_count < 2 result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"} if isinstance(e, openai.error.RateLimitError): - logger.warn("[TONGYI] RateLimitError: {}".format(e)) + logger.warn("[QWEN] RateLimitError: {}".format(e)) result["content"] = "提问太快啦，请休息一下再问我吧" if need_retry: time.sleep(20) elif isinstance(e, openai.error.Timeout): - logger.warn("[TONGYI] Timeout: {}".format(e)) + logger.warn("[QWEN] Timeout: {}".format(e)) result["content"] = "我没有收到你的消息" if need_retry: time.sleep(5) elif isinstance(e, openai.error.APIError): - logger.warn("[TONGYI] Bad Gateway: {}".format(e)) + logger.warn("[QWEN] Bad Gateway: {}".format(e)) result["content"] = "请再问我一次" if need_retry: time.sleep(10) elif isinstance(e, openai.error.APIConnectionError): - logger.warn("[TONGYI] APIConnectionError: {}".format(e)) + logger.warn("[QWEN] APIConnectionError: {}".format(e)) need_retry = False result["content"] = "我连接不到你的网络" else: - logger.exception("[TONGYI] Exception: {}".format(e)) + logger.exception("[QWEN] Exception: {}".format(e)) need_retry = False self.sessions.clear_session(session.session_id) if need_retry: - logger.warn("[TONGYI] 第{}次重试".format(retry_count + 1)) + logger.warn("[QWEN] 第{}次重试".format(retry_count + 1)) return self.reply_text(session, retry_count + 1) else: return result @@ -176,8 +176,8 @@ class TongyiQwenBot(Bot): # NOTE 模拟系统消息，测试发现人格描述以"你需要扮演ChatGPT"开头能够起作用，而以"你是ChatGPT"开头模型会直接否认 system_qa = ChatQaMessage(system_content, '好的，我会严格按照你的设定回答问题') history.insert(0, system_qa) - logger.debug("[TONGYI] converted qa messages: {}".format([item.to_dict() for item in history])) - logger.debug("[TONGYI] user content as prompt: {}".format(user_content)) + logger.debug("[QWEN] converted qa messages: {}".format([item.to_dict() for item in history])) + logger.debug("[QWEN] user content as prompt: {}".format(user_content)) return user_content, history def get_completion_content(self, response, node_id): diff --git a/bot/tongyi/ali_qwen_session.py b/bot/ali/ali_qwen_session.py similarity index 100% rename from bot/tongyi/ali_qwen_session.py rename to bot/ali/ali_qwen_session.py diff --git a/bot/bot_factory.py b/bot/bot_factory.py index a0edde1..a54f706 100644 --- a/bot/bot_factory.py +++ b/bot/bot_factory.py @@ -45,6 +45,6 @@ def create_bot(bot_type): return ClaudeAIBot() elif bot_type == const.QWEN: - from bot.tongyi.tongyi_qwen_bot import TongyiQwenBot - return TongyiQwenBot() + from bot.ali.ali_qwen_bot import AliQwenBot + return AliQwenBot() raise RuntimeError