Hiroaki Hayashi comments

Results 5 comments of


                                            Hiroaki Hayashi

The current neovim build in vimr doesn't support python3

@diogovieira You might have already resolved the issue, but setting the relevant python path might help: ``` let g:python3_host_prog = "/path/to/your/python" ```

AttributeError: 'CodeGen25Tokenizer' object has no attribute 'encoder'

Correct, there was a breaking change in transformers that changed the order of variable declarations within PretrainedTokenizer. On it.

Can't this model be trained using multi nodes and multiple cards ?

16GB graphics card may be somewhat limiting. Perhaps you could try CPU-offloading or offloading to NVMe to save more memory, as well as multi-node training.

It seems the result we get is not the same as the repo shows

Hey @FranxYao , thanks for the great work on this paper. I was wondering, did you use the prompt in the code or the modified prompt above for the figure...

It seems the result we get is not the same as the repo shows

Great find! I totally missed that. File order being different after every cloning but consistent within each clone is probably because files are copied and registered by the file system...