PengWenChen

@tjinjin95 Hi. I've tried to run the online code by myself and it works, but I can't exactly reproduce the MO score written on paper. The paper says the average...

Hi there~ I would also like to ask about adding special tokens. The case is that for some models, such as Qwen1.5, the special tokens are not in the vocab.json...

Hi @knighton. Thanks for your reply! My working environment cannot access as root account. If one of my partner runs your package first and somehow generates some cache files, such...

Hi @knighton. I found another root path here: https://github.com/mosaicml/streaming/blob/main/streaming/base/stream.py#L166 After modifying both `self._filelock_root` in `dataset.py` and `root` in `stream.py`, the scripts can successfully executed! But I still want to confirm...

Hi there! The sharedmemory seems not be cleaned up successfully by the first or other users and then the error occurs: `Permission denied: '/00000_locals'`. I also found another closed issue:...

Hi @Skylion007. Thanks for your reply. I haven't tried this to force destroy the leftover shared memory. If I encounter the same problem in the future, I will give it...

For these path modification: Yes. My path is readable and writeable, and I can change the permission by `chmod` either. https://github.com/mosaicml/streaming/issues/546#issuecomment-1868672842 As for `/000000_locals` issue, I failed to change the...

Hi @xiamengzhou, I also encounter this issue with the original dynamic loading setup in `pruning.sh`. `set_names=[cc,github,book,stackexchange,wiki,arxiv,c4]` `proportion=[0.67,0.045,0.045,0.02,0.045,0.025,0.15]` And NaN happens in the first batch when calculating `metric/train/stackexchange_LanguageCrossEntropy`. The environment I...

Hi @xiamengzhou! Thanks for your reply. However, I can not access google drive where I am working :( Could you please upload the processed data to this repository? It would...

Hi, @xiamengzhou! The proportion updating fails because of NaN loss on evaluation data. And it is because of the missing data of some subdatasets. I solved this issue by increasing...