fastnlp/cpt issues and pull requests

#88 - A detailed pre-train process experiment

Issue - State: open - Opened by shivanraptor 3 months ago

#87 - Does the model have the ability to summarize text?

Issue - State: open - Opened by YANCC139 4 months ago

#86 - 为什么没有总结摘要功能？

Issue - State: closed - Opened by YANCC139 4 months ago

#85 - Pre-trained model raises IndexError: index out of range in self

Issue - State: open - Opened by shivanraptor 4 months ago

#84 - run_gen.py returns UnicodeDecodeError

Issue - State: closed - Opened by shivanraptor 4 months ago - 1 comment

#83 - run_gen.py shows error: module 'tensorflow._api.v2.compat.v2.internal' has no attribute 'register_load_context_function'

Issue - State: closed - Opened by shivanraptor 4 months ago - 2 comments

#82 - run_pretrain_bart.sh returns IndexError

Issue - State: open - Opened by shivanraptor 5 months ago - 13 comments

#81 - What are the data formats of `dataset` and `vocab` folder?

Issue - State: open - Opened by shivanraptor 5 months ago - 2 comments

#80 - 直接运行python run_gen.py，GPU 利用率不到50%

Issue - State: open - Opened by runningabcd about 1 year ago - 3 comments

#79 - 使用 CPT 进行微调后，生成任务推理很慢，一条数据都要好几秒，如何提升推理速度？

Issue - State: closed - Opened by runningabcd about 1 year ago - 1 comment

#78 - 如何基于 CPT 进行 Pretrain

Issue - State: closed - Opened by runningabcd about 1 year ago

#77 - TypeError: forward() got an unexpected keyword argument 'labels'

Issue - State: closed - Opened by runningabcd about 1 year ago - 3 comments

#76 - Pytorch model to ONNX, embedding layer parameters are misalignment?

Issue - State: open - Opened by smygreat about 1 year ago

#75 - Why will there be blank between chinese characters? How to fix it?

Issue - State: open - Opened by SupetZYK about 1 year ago - 1 comment

#74 - from modeling_cpt import CPTForConditionalGeneration 出错

Issue - State: open - Opened by sumafuture over 1 year ago - 2 comments

#73 - bug : got an unexpected keyword argument 'tokenizer_class'

Issue - State: closed - Opened by xuefeich over 1 year ago - 1 comment

#72 - 看起来cpt和bart效果差不太多？

Issue - State: open - Opened by xuefeich over 1 year ago - 1 comment

#71 - 关于pretraining时denoising的问题

Issue - State: open - Opened by tu2022 over 1 year ago - 3 comments

#70 - finetune时，怎么能每训练一个epoch，保存一次模型

Issue - State: closed - Opened by deepkun over 1 year ago

#69 - sh run_pretrain_bart.sh 报错

Issue - State: open - Opened by jackychancjcjcj over 1 year ago - 1 comment

#68 - modeling_cpt.py问题请教

Issue - State: open - Opened by LKAMING97 over 1 year ago - 1 comment

#67 - Can I finetune CPTForMaskedLM?

Issue - State: open - Opened by bmkor almost 2 years ago - 4 comments

#66 - CPT pretrain problem

Issue - State: open - Opened by SunyanGu almost 2 years ago - 3 comments

#65 - unrecognized arguments: --local_rank=1

Issue - State: closed - Opened by SunyanGu almost 2 years ago - 1 comment

#64 - token_type_ids报错

Issue - State: closed - Opened by longborn almost 2 years ago - 1 comment

#64 - token_type_ids报错

Issue - State: closed - Opened by longborn almost 2 years ago - 1 comment

#63 - can I change the max length of csl

Issue - State: closed - Opened by Dlinuigh almost 2 years ago - 2 comments

#62 - TypeError: add_code_sample_docstrings() got an unexpected keyword argument 'tokenizer_class'

Issue - State: closed - Opened by HanZijie almost 2 years ago

#61 - How to fine-tune the Chinese BART model?

Issue - State: open - Opened by AlexNLP almost 2 years ago - 2 comments

#60 - bart-base模型ft之前preprocess的方法是什么呢？

Issue - State: closed - Opened by fubuki75 about 2 years ago - 1 comment

#60 - bart-base模型ft之前preprocess的方法是什么呢？

Issue - State: closed - Opened by fubuki75 about 2 years ago - 1 comment

#59 - How much corpus have you used?

Issue - State: closed - Opened by ghost about 2 years ago - 1 comment

#58 - CPT多GPU卡finetuning训练报错

Issue - State: closed - Opened by aidejieceng about 2 years ago - 5 comments

#57 - ModuleNotFoundError: No module named 'amp_C'

Issue - State: closed - Opened by aidejieceng about 2 years ago - 1 comment

#56 - 推理时的token_type_ids

Issue - State: closed - Opened by 1245244103 about 2 years ago - 1 comment

#55 - 有关fnlp/bart-base-chinese模型加载问题

Issue - State: closed - Opened by yedongyu1996 about 2 years ago - 2 comments

#54 - 重载模型时vocab.txt的问题

Issue - State: closed - Opened by Guochry over 2 years ago

#53 - 生成模型预测出的结果跟输入一模一样？

Issue - State: closed - Opened by zhihao-chen over 2 years ago

#52 - 你好，想请教下微调中dev.json数据的结构问题？

Issue - State: closed - Opened by ImmortalZ over 2 years ago - 2 comments

#51 - bart-base-chinese支持 lightseq等推理加速不

Issue - State: closed - Opened by ImmortalZ over 2 years ago - 2 comments

#50 - 想咨询下，fnlp/bart-base-chinese 推理时使用gpu该怎么调用吗

Issue - State: closed - Opened by ImmortalZ over 2 years ago - 1 comment

#49 - 请问做句子复述和文章摘要要调用哪些方法？

Issue - State: closed - Opened by estuday over 2 years ago - 2 comments

#48 - index out of range in self

Issue - State: closed - Opened by OuYanXi over 2 years ago - 1 comment

#47 - 这个支持多机多卡训练吗？可以提供训练脚步👀一下吗？

Issue - State: closed - Opened by lanse-sir over 2 years ago - 1 comment

#46 - CPTForConditionalGeneration使用多GPU报错

Issue - State: closed - Opened by Biaocsu over 2 years ago - 5 comments

#45 - RuntimeError:add_code_sample_docstrings() got an unexpected keyword argument 'tokenizer_class'RuntimeError: Failed to import roformer.modeling_roformer because of the following error (look up to see its traceback): add_code_sample_docstrings() got an unexpected keyword argument 'tokenizer_class'

Issue - State: closed - Opened by Cyanash over 2 years ago - 1 comment

#44 - CSL摘要数据集没达到效果。

Issue - State: closed - Opened by josephcui over 2 years ago - 6 comments

#43 - 中文分词的数据问题

Issue - State: closed - Opened by Richar-Du over 2 years ago

#42 - 基于BART做机器翻译finetune

Issue - State: closed - Opened by codejeam over 2 years ago

#41 - CSL数据集

Issue - State: closed - Opened by josephcui over 2 years ago - 1 comment

#40 - 生成相关问题

Issue - State: closed - Opened by TheHonestBob over 2 years ago

#39 - 不支持token_type_embedding吗

Issue - State: closed - Opened by songmzhang over 2 years ago - 1 comment

#38 - pretrain任务的环境版本

Issue - State: closed - Opened by 1190303311 almost 3 years ago - 1 comment

#37 - generation子任务咨询

Issue - State: closed - Opened by Biaocsu almost 3 years ago

#36 - CPT模型是如何预训练的？

Issue - State: closed - Opened by qingquanli-tech almost 3 years ago

#35 - Why g-dec will inflect the performance of u-dec?

Issue - State: closed - Opened by Chen-Wang-CUHK almost 3 years ago - 1 comment

#34 - classification/CLUEWSC 2020数据集上效果没达到

Issue - State: closed - Opened by zisang0210 almost 3 years ago - 2 comments

#33 - 用huggingface代码直接进行BART large fineturning出现繁体字

Issue - State: closed - Opened by yht4work almost 3 years ago - 7 comments

#32 - 生成任务的数据

Issue - State: closed - Opened by Richar-Du almost 3 years ago - 1 comment

#31 - 请问如果只需要基于特定文本做问题生成的话，是不是只需要自己构建三元组就可以了呢

Issue - State: closed - Opened by ghost almost 3 years ago - 1 comment

#30 - 使用自定义数据集在bart-base-chinese的继续pretrain

Issue - State: closed - Opened by Aureole-1210 almost 3 years ago - 2 comments

#29 - Hardware and computing cost

Issue - State: closed - Opened by Richar-Du almost 3 years ago - 2 comments

#28 - 想咨询run_gen.py如何设置GPU运行？

Issue - State: closed - Opened by PolarisRisingWar almost 3 years ago - 3 comments

#27 - max_position_embeddings是1024吗

Issue - State: closed - Opened by awdrgyjilplij almost 3 years ago - 2 comments

#26 - pretrain_bert 貌似不能用

Issue - State: closed - Opened by AlfredTheBest about 3 years ago - 1 comment

#25 - bart-large-chinese在自定义的数据集上进行进一步的预训练

Issue - State: closed - Opened by maojinyang about 3 years ago - 10 comments

#24 - 请问您这边提供的bart模型能支持类似于文章生成或者是标题补全吗？

Issue - State: closed - Opened by YINGPENGZH about 3 years ago - 5 comments

#23 - 忽略

Issue - State: closed - Opened by AlfredTheBest about 3 years ago

#22 - 模型预测输出

Issue - State: closed - Opened by Ai-Sherry about 3 years ago - 1 comment

#21 - BART 模型在 pretrain 和 finetune 阶段的数据预处理一致性

Issue - State: closed - Opened by wlhgtc about 3 years ago - 5 comments

#20 - 有关bart-base-chinese的vocab.txt文件

Issue - State: closed - Opened by Edward-Joker about 3 years ago - 2 comments

#19 - 请问summarization 句子text最长是不是不能超过512，我看model的posional embedding只有514，想要摘要更长的句子应该怎么做呢？

Issue - State: closed - Opened by hj5167 about 3 years ago - 2 comments

#18 - encoder decoder最大位置编码不一致

Issue - State: closed - Opened by JaheimLee about 3 years ago - 1 comment

#17 - ner模型的问题

Issue - State: closed - Opened by suhejian about 3 years ago - 6 comments

#16 - why BertTokenizer is used instead of BartTokenizer?

Issue - State: closed - Opened by Chen-Wang-CUHK about 3 years ago - 2 comments

#15 - 有关Decoder的层数问题

Issue - State: closed - Opened by AQA6666 about 3 years ago - 2 comments

#14 - G-DEC与U-DEC的区分是如何实现的？

Issue - State: closed - Opened by AQA6666 about 3 years ago

#13 - generation/LCSTS数据集上效果没达到

Issue - State: closed - Opened by zhoucz97 about 3 years ago - 17 comments

#12 - run_mrc.py中只有model_type==bert时的代码

Issue - State: closed - Opened by zhoucz97 about 3 years ago - 1 comment

#11 - update

Pull Request - State: closed - Opened by thorseraq about 3 years ago - 1 comment

GitHub / fastnlp/cpt issues and pull requests