ConvLab-2
ConvLab-2 copied to clipboard
thu-coai
ConvLab-2: An Open-Source Toolkit for Building, Evaluating, and Diagnosing Dialogue Systems
**Describe the bug** 在使用中文数据集时，在policy部分会出现，for intent, domain, slot, value in system_action:时候会出现ValueError: too many values to unpack(expected 4)的报错，查询代码发现，prev_sys_da传入数据不只有["system_action"]部分，在修改代码如 system_action = prev_sys_da['system_action'] for intent, domain, slot, value in system_action: 后，发现不能正常多轮对话，主要表现在policy.predict部分的问题，同时似乎是无法读取"Request“部分
**Describe the bug** Failed to build agent on CoLab, the convlab2 is not successfully imported. **To Reproduce** Steps to reproduce the behavior: 1. Go to colab 2. Click on all...
Hi, the spacy tokenizer in the NLU module (specifically, jointBERT) is downloaded every time the script is launched? Thus, the module downloads the latest version of the tokenizer? Or the...
Hi, it's not clear the difference between the evaluate.py and test.py scripts in the NLU folder. What do they evaluate? Since the result obtained is completely different even if they...
Hi, could you please provide more information on how the Rule DST module is evaluated? Thanks
After I installed ConvLab-2 and run test_DAMD.py in ConvLab-2/tests/ directory, I got a complete and success rate of 34.3 and 29.5, which is different from that in the README.md. Besides,...
**Describe the feature** Provide a dependency resolver with stricter constraints for ```boto3``` and ```botocore```. **Expected behavior** Direct install the expected version. **Additional context** A sample of installing process: ``` ........
Most of the models in ConvLab-2 are basic and early. Are there plans to incorporate more new models (DST / E2E)?
**Describe the feature** I've noticed that there are a few issues (#8, #13, #15, #20, #40) mention that it's hard to train RL policy (PG, PPO, GDPL). Thanks all of...
当我在/content/ConvLab-2/convlab2/dst目录下执行!python evaluate.py MultiWOZ TRADE test时，得到的准确率分别为：{'Joint Acc': 0.27075420510037984, 'Turn Acc': 0.9226351962379891, 'Joint F1': 0.7568646167948684}  但在您的README文件中给出了更高的准确率，不知道是否是我哪里操作有误？ 
