From ad9112196a2612badd21c5431c2506dbb91ea003 Mon Sep 17 00:00:00 2001 From: gfhdhytghd <3391146750@qq.com> Date: Fri, 20 Sep 2024 17:55:41 +0800 Subject: [PATCH] more --- getmsgserv/LM_work/sendtoLM-MTP-stream.py | 28 ++++++++++++----------- oqqwall.config | 6 +++-- 2 files changed, 19 insertions(+), 15 deletions(-) diff --git a/getmsgserv/LM_work/sendtoLM-MTP-stream.py b/getmsgserv/LM_work/sendtoLM-MTP-stream.py index 5207fe5..9e0be31 100644 --- a/getmsgserv/LM_work/sendtoLM-MTP-stream.py +++ b/getmsgserv/LM_work/sendtoLM-MTP-stream.py @@ -8,7 +8,7 @@ from dashscope.api_entities.dashscope_response import Role import re - +print('using MPT stream') def read_config(file_path): config = {} with open(file_path, 'r') as f: @@ -57,7 +57,7 @@ def fetch_response_in_parts(prompt, max_rounds=5): previous_output = "" while not is_complete and round_count < max_rounds: - seed = random.randint(1, 10000) + seed = 1354 print(f"Round {round_count + 1} - Using seed: {seed}") # 使用流式输出方式调用生成模型 @@ -68,7 +68,7 @@ def fetch_response_in_parts(prompt, max_rounds=5): result_format='message', stream=True, incremental_output=True, - max_tokens=6000, + max_tokens=8192, temperature=0.50, repetition_penalty=1.0 ) @@ -84,31 +84,33 @@ def fetch_response_in_parts(prompt, max_rounds=5): else: print(f"Error in API call: {response.status_code}, {response.message}") break - + #print(output_content) if previous_output: - # 获取前一输出的最后100个字符 + # Get the last 100 characters of the previous output overlap_content = previous_output[-100:] - # 在当前输出的前500个字符中搜索这些字符 + # Search for these 100 characters within the first 500 characters of the current output start_index = output_content[:500].find(overlap_content) if start_index != -1: - # 如果找到，移除之前的重复内容 + # If found, remove everything before this occurrence output_content = output_content[start_index + len(overlap_content):] + # Update the full response full_response += output_content previous_output = output_content - # 检查响应是否包含结束标志 - if output_content.endswith('\n```'): + # Check if the response contains the ending indicator '```' + if output_content.endswith('```'): + print("complete!") is_complete = True else: - # 将输出的最后100个字符截断后添加到消息中 + # Truncate the last 100 characters before adding to messages truncated_output = output_content[:-100] if len(output_content) > 100 else output_content messages.append({ 'role': Role.ASSISTANT, 'content': truncated_output }) - messages.append({'role': Role.USER, 'content': '接着上次停下的地方继续输出，不要重复之前的内容，不要重复sender和needpriv等内容，不要在开头重复一遍```json {"time": },{"message": [{"type": ,"data": {，不要在开头重复任何格式内容，直接接着上次结束的那个字继续'}) - + # Prompt the model to continue without repeating content + messages.append({'role': Role.USER, 'content': '接着上次停下的地方继续输出，不要重复之前的内容，不要重复sender和needpriv等内容，不要在开头重复一遍```json {"time": },{"message": [{"type": ,"data": {，不要在开头重复任何格式内容，直接接着上次结束的那个字继续,但是如果json已经到达末尾，请用\n```结束输出'}) round_count += 1 return full_response @@ -165,7 +167,7 @@ def main(): " },\n" " \"needpriv\": \"true\"/\"false\",\n" " # 判断这条信息是否需要匿名\n" - " # 有时匿名意思会通过“匿”或者”码”的谐音字传达（比如逆，腻，拟或者马，吗，嘛），有时也会通过“🐎”“🐴”之类的emoji传达\n" + " # 有时匿名意思会通过“匿”或者”码”的谐音字传达（比如逆，腻，拟或者马，吗，嘛），有时也会通过“🐎”“🐴”之类的emojy传达\n" " # 凡遇到只有一字意义不明的消息组，就要考虑一下这个字是否传达了匿名意思" " \"safemsg\": \"true\"/\"false\",\n" " # 判断这条信息是否可以过审（是否含有攻击性信息或者政治信息）\n" diff --git a/oqqwall.config b/oqqwall.config index de9a5d1..ad1484d 100644 --- a/oqqwall.config +++ b/oqqwall.config @@ -1,7 +1,9 @@ -http-serv-port=xxx -apikey="sk-xxx" +http-serv-port= +apikey="" disable_qzone_autologin=false enable_selenium_autocorrecttag_onstartup=false use_selenium_to_generate_qzone_cookies=false use_LLOnebot=false +LMarg="-MTP-stream" +process_waittime=120 max_attempts_qzone_autologin=3 \ No newline at end of file