CDial-GPT

CDial-GPT copied to clipboard

支持incremental decoding吗？

3

如题

songmzhang

nll_loss_forward_reduce_cuda_kernel_2d: block: [0,0,0], thread: [18,0,0] Assertion t >= 0 && t < n_classes failed.

4

ERROR:ignite.engine.engine.Engine:Current run is terminating due to exception: unsupported operand type(s) for /: 'str' and 'int'. /opt/conda/conda-bld/pytorch_1634272168290/work/aten/src/ATen/native/cuda/Loss.cu:247: nll_loss_forward_reduce_cuda_kernel_2d: block: [0,0,0], thread: [0,0,0] Assertion `t >= 0 && t < n_classes` failed....

Alexia1994

关于计算评价指标

4

请问已经生成STC_result.txt，想生成像Readme 评价指标表格中的 PPL | BLEU-2 | BLEU-4 | Dist-1 | Dist-2 | Greedy Matching | Embedding Average 这些值，请问你们是怎么计算的？

huanghonggit

embedding average 计算中ground truth中存在英文，英文分词是如何处理的。

2

你还根据https://github.com/thu-coai/CDial-GPT/issues/53 和https://github.com/thu-coai/CDial-GPT/issues/55 两个issue的指导复现embedding average的计算，发现ground truth中有英文语句，中文分词方法不太适用英文的分词，请问你们是怎么处理的呢？直接丢弃还是适用英文分词方法对英文的ground truth进行分词。例如在STC_test.json中存在“"I f o n l y w e c o u l d s e e t h e w o r l d i...

allyouneeds

关于 Greedy Matching Embedding Average问题，AIlab的向量都是字向量

6

AILab的向量都是中文词向量，而在STC数据集生成的都是以字为单位，请问您在测试的时候，是将STC_test和模型生成的文本转换成词的形式计算，还是没经过分词处理，直接计算Embedding Average和Greedy Matching

Ultraman-Orb

字嵌入向量问题

1

请问对话语每个字的字嵌入向量编码的编码是哪个变量？在代码中没找到word_embedding变量，请问有字嵌入向量编码的变量吗？

Ultraman-Orb

您好，我按照您提供的GREEDY MATCHINHG 和EMBEDDING AVERAGE的代码进行评价，发现速度非常慢，想问一下您当时是用什么配置进行评价操作的？

AKkellan

why？

guantao18

出现报错：C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: block: [0,0,0], thread: [0,0,0] Assertion `t >= 0 && t < n_classes` failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: block: [0,0,0], thread: [1,0,0] Assertion `t >= 0 && t < n_classes` failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: block:...

erminga

Bumps [transformers](https://github.com/huggingface/transformers) from 2.1.1 to 4.30.0. Release notes Sourced from transformers's releases. v4.30.0: 100k, Agents improvements, Safetensors core dependency, Swiftformer, Autoformer, MobileViTv2, timm-as-a-backbone 100k Transformers has just reached 100k stars...

dependabot[bot]