kaixindelele / ChatPaper

Use ChatGPT to summarize the arXiv papers. 全流程加速科研,利用chatgpt进行论文全文总结+专业翻译+润色+审稿+审稿回复
https://chatwithpaper.org
Other
17.99k stars 1.9k forks source link

结果设定为3,结果跑了3个小时。不知问题在哪。梯速正常。 #99

Closed gxandys closed 1 year ago

gxandys commented 1 year ago

python chat_paper.py --query "Multi dimensional time series" --filter_keys "Multi dimensional time series" --max_results 3 用的这个命令,结果3个小时才跑完。看API后台数据差不多用了10k个token,梯速应该是没问题的。而且虽然运行结束,但中间有报错信息,不知道是怎么回事。而且结果的内容看起来乱乱的,求指导。

筛选后剩下的论文数量:
filter_results: 3
filter_papers:
0 Measuring Congruence on High Dimensional Time Series 2018-05-31 21:51:27+00:00
1 Application of the Allan Variance to Time Series Analysis in Astrometry and Geodesy: A Review 2016-07-16 07:33:57+00:00
2 Exact multi-parameter persistent homology of time-series data: Fast and variable one-dimensional reduction of multi-parameter persistence theory 2022-12-28 07:12:58+00:00
All_paper: 3
paper_path: ./pdf_files/Multi dimensional time se-2023-03-24-17\Measuring Congruence on High Dimensional Time Series.pdf
section_page_dict {'Abstract': 0, 'Introduction': 0, 'Related Work': 0, 'Experimental Results': 11, 'Results': 12, 'References': 12}
0 Abstract 0
1 Introduction 0
start_page, end_page: 0 0
2 Related Work 0
start_page, end_page: 0 11
3 Experimental Results 11
start_page, end_page: 11 12
4 Results 12
start_page, end_page: 12 12
5 References 12
start_page, end_page: 12 21
paper_path: ./pdf_files/Multi dimensional time se-2023-03-24-17\Application of the Allan Variance to Time Series Analysis in Astrometry and Geodesy_ A Review.pdf
section_page_dict {'Abstract': 0, 'Introduction': 0, 'References': 10}
0 Abstract 0
1 Introduction 0
start_page, end_page: 0 10
2 References 10
start_page, end_page: 10 13
paper_path: ./pdf_files/Multi dimensional time se-2023-03-24-17\Exact multi-parameter persistent homology of time-series data_ Fast and variable one-dimensional reduction of multi-parameter persistence theory.pdf
section_page_dict {'Abstract': 0, 'Introduction': 0, 'Method': 22, 'Conclusion': 27, 'References': 27}
0 Abstract 0
1 Introduction 0
start_page, end_page: 0 22
2 Method 22
start_page, end_page: 22 27
3 Conclusion 27
start_page, end_page: 27 27
4 References 27
start_page, end_page: 27 31
summary_error: Request timed out: HTTPSConnectionPool(host='api.openai.com', port=443): Read timed out. (read timeout=600)
conclusion_result:
 8. 结论:

- (1):本文的意义在于提出了一种新的基于感性学习的增强学习方法,可以更好地处理高维状态下动作的决策问题。
- (2):创新点:本文的创新点在于提出了基于感性学习的增强学习方法,可以更好地处理高维状态下动作的决策问题。表现:本文在实 验中验证了所提出方法的有效性,但结果与其他算法相比略有不足。工作量:本文设计的算法比较复杂,需要进行大量参数调整和模型优化。
prompt_token_used: 297 completion_token_used: 180 total_token_used: 477
response_time: 9.547 s
summary_error: Request timed out: HTTPSConnectionPool(host='api.openai.com', port=443): Read timed out. (read timeout=600)
conclusion_result:
 8. Conclusion:

- (1): 本文的意义在于提出了一种新的深度强化学习模型,可以在实现高水平控制任务的同时,实现级联控制。该模型提供了一种新的 思路,可以在未来的相关研究中得到借鉴。

- (2): 创新点:本文提出的深度强化学习模型可以实现级联控制;表现:通过实验表明,该模型在一系列任务中取得了稳定且优良的表 现;工作量:本文的研究还有一些瓶颈和限制,需要进一步完善和优化。
prompt_token_used: 297 completion_token_used: 193 total_token_used: 490
response_time: 9.344 s
summary_error: Error communicating with OpenAI: HTTPSConnectionPool(host='api.openai.com', port=443): Max retries exceeded with url: /v1/chat/completions (Caused by ProxyError('Cannot connect to proxy.', RemoteDisconnected('Remote end closed connection without response')))
method_result:
 7. 方法:

- (1): 采用多参数持久化同调(EMPH)方法来提高单参数持久化同调方法(SEPH)的精度,同时还能极大程度地降低计算成本。
- (2): 作者采用滑动窗口嵌入法来处理序列数据,将高维数据降为二维数据,减小了计算量。
- (3): 作者使用了多个数据集来测试EMPH方法的性能,并进行了分类实验,以检验该方法提高单参数持久化同调方法精度的能力。
prompt_token_used: 1907 completion_token_used: 160 total_token_used: 2067
response_time: 7.822 s
conclusion_error: Request timed out: HTTPSConnectionPool(host='api.openai.com', port=443): Read timed out. (read timeout=600)
summary time: 9413.139022827148
gxandys commented 1 year ago

在把urllib3降级到1.25.11,和在chat_paper.py里加上os.environ["http_proxy"] = "http://127.0.0.1:10809" os.environ["https_proxy"] = "http://127.0.0.1:10809"后,错误变成了summary_error: Error communicating with OpenAI: HTTPSConnectionPool(host='api.openai.com', port=443): Max retries exceeded with url: /v1/chat/completions (Caused by ProxyError('Cannot connect to proxy.', RemoteDisconnected('Remote end closed connection without response')))

kaixindelele commented 1 year ago

可能是节点的问题,梯子不干净。这个报错,肯定是和chat的通信出了问题