JackieWu
JackieWu
hi,我使用100条数据,按照custom的数据阻止方式,同时val,test 和train 数据一样,其他使用默认的参数,发现val打印的rouge 都为0,看到results.json里面的所有的pred也为空,请问可能是什么原因引起的?是否有其他参数需要调整,数据就是随机的cmrc的以query作为source,答案句作为target进行生成的。rouge里面打印的fix_tokenization 返回有结果,但是duplication ration 后的buf就为空了,是因为生成的太多重复了么?
### Is there an existing issue for this? - [X] I have searched the existing issues ### Current Behavior 我想用chatglm来训练摘要任务?但是看模型的效果好像比复旦的cpt模型效果bart/CPT 的base效果差,请问这个问题怎么优化?还是说模型对于这种任务不太适用? 参考:  ### Expected Behavior no ### Steps To...
hi,请教下cmrc数据集上的程序可以正常跑通? 结果怎么样,我按照seq_cmrc.sh 跑的,pred的所有的预测结果都是空呢,请问是不是中文seq2seq上有什么问题?或者哪里设置不对
hi,请问下使用rocketqa训练的模型如何限制gpu大小? 使用FLAGS_gpu_memory_limit_mb 设置并没有生效
hi, 我使用双卡运行模型并行,train_model_all.py 报错: ValueError: DistributedDataParallel device_ids and output_device arguments only work with single-device/Multi-devie gpu modules, but got device_ids[0],output_device 0,and module parameter {device(type='cuda',index=0),device(type='cuda',index=1)
ddp traing multi gpu Expected all tensors to be on the same device, but found at least two devices
File "/opt/conda/lib/python3.10/site-packages/torch/nn/functional.py", line 2199, in embedding return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse) RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:1 and...
### 🐛 Describe the bug When i run the code in summarization-rlhf using GPT2Chinese, the following error occurs . I have checked the "specail_tokens_map.json", it does have the "[PAD]" token....
colab 上执行memorylist 报错如下: frames /usr/local/lib/python3.10/dist-packages/memorag/memorag_lite.py in adapt_batch_size(self) 168 batch_size = bs 169 break --> 170 return batch_size 171 172 def memorize( UnboundLocalError: local variable 'batch_size' referenced before assignment 貌似batch size...
changes: 1. model changed to **canopylabs/3b-es_it-ft-research_release** 2. max lenght : 3200 3. def redistribute_codes(code_list): if len(code_list) == 0: print("Warning: Empty code list, returning silence") return torch.zeros(1, 1, 24000) # 1秒的静音...