Open NLPSpeechAI opened 1 year ago
请问能否提供以训练好的解析器呢
请问能否提供以训练好的解析器呢
你可以尝试使用这个解析器: https://github.com/seq-to-mind/DDP_parsing
请问能否提供以训练好的解析器呢
你可以尝试使用这个解析器: https://github.com/seq-to-mind/DDP_parsing
Question: What did you do about the parser handling some session length limits?
请问在使用这个解析器时,使用自己的中文数据,遇到这样的问题,该如何解决?RuntimeError: The expanded size of the tensor (2330) must match the existing size (514) at non-singleton dimension 1. Target sizes: [1, 2330]. Tensor sizes: [1, 514]
请问在使用这个解析器时,使用自己的中文数据,遇到这样的问题,该如何解决?RuntimeError: 张量的扩展大小 (2330) 必须与非单例维度 514 处的现有大小 (1) 匹配。目标尺寸:[1, 2330]。张量大小:[1, 514]
我也遇到了这个问题
请问在使用这个解析器时,使用自己的中文数据,遇到这样的问题,该如何解决?RuntimeError: 张量的扩展大小 (2330) 必须与非单例维度 514 处的现有大小 (1) 匹配。目标尺寸:[1, 2330]。张量大小:[1, 514]
我也遇到了这个问题
我实验用的是英文数据,二者应该没有本质区别,你可以单步调试一下,看看是否忽略了一些维度问题。
请问能否提供以训练好的解析器呢
你可以尝试使用这个解析器: https://github.com/seq-to-mind/DDP_parsing
Question: What did you do about the parser handling some session length limits?
For DailyDialog, EmoryNLP, and MELD, the dialogues are relatively short, and there is no issue with length restrictions; a parser based on the transformer ( https://github.com/seq-to-mind/DDP_parsing) can handle them. Only in IEMOCAP are there longer dialogues, and in such cases, a parser based on RNN (https://github.com/shizhouxing/DialogueDiscourseParsing) can be used.
请问能否提供以训练好的解析器呢
你可以尝试使用这个解析器: https://github.com/seq-to-mind/DDP_parsing
Question: What did you do about the parser handling some session length limits?
For DailyDialog, EmoryNLP, and MELD, the dialogues are relatively short, and there is no issue with length restrictions; a parser based on the transformer ( https://github.com/seq-to-mind/DDP_parsing) can handle them. Only in IEMOCAP are there longer dialogues, and in such cases, a parser based on RNN (https://github.com/shizhouxing/DialogueDiscourseParsing) can be used.
Thank you very much for your answer, which is of great help to me. I wish you every success in your scientific research
请问有复现出结果的吗,有好心人分享一下MELD_Feature.pkl和IEMOCAP_Features.pkl的特征文件吗,非常感谢 ![Uploading QQ截图20240402223622.png…]()
请问能否提供以训练好的解析器呢
你可以尝试使用这个解析器: https://github.com/seq-to-mind/DDP_parsing
Question: What did you do about the parser handling some session length limits?
For DailyDialog, EmoryNLP, and MELD, the dialogues are relatively short, and there is no issue with length restrictions; a parser based on the transformer ( https://github.com/seq-to-mind/DDP_parsing) can handle them. Only in IEMOCAP are there longer dialogues, and in such cases, a parser based on RNN (https://github.com/shizhouxing/DialogueDiscourseParsing) can be used.
Thank you very much for your answer, which is of great help to me. I wish you every success in your scientific research
Hello, I also used the DDA_parsing to generate links and relations, may I ask whether your MELD_feature.pkl file and IEMOCAP_feature.pkl file are reproduced? I reproduced a MELD_feature.pkl file, but the final result is not good. Can we exchange the pkl files?
I apologize, but my first time working with features and discourse graphs was over a year ago. Due to the considerable time that has passed, there was an incident of server failure at our company during that period, resulting in data loss from the hard drive. Currently, only the code submitted at the time of the paper submission is available. As I am usually occupied with work, I will make an effort to process and upload a new set as soon as possible. Alternatively, if time is limited, you could follow the extraction methods mentioned in the Readme and process it on your own.