Vatsal Mishra comments

Results 5 comments of


                                            Vatsal Mishra

RuntimeError with DirectML backend (torch.cat concat in tokenizer)

@J4R3LL could you please test PR #51 on your AMD + torch-directml setup and confirm if it fixes the concat error? Maintainers don’t have AMD hardware to reproduce.

RuntimeError with DirectML backend (torch.cat concat in tokenizer)

**Update:** #51 fixes the DirectML concat crash **Follow-up:** Opened #54 to address a separate torch-directml issue from Transformers (attention_mask dtype overflow). It forces a boolean `attention_mask` before `model.generate()`/`forward()`. Separate scope;...

Floating point exception (core dumped)

**Quick tester note** (for @zoushipeng and anyone who can repro) **Context.** Issue #71 hits a “Floating point exception (core dumped)” on CUDA with `torch.bfloat16`, crashing at the `lm_head` call on...

add `regr_r2` group by function, and funcs for RMSE and t-stats

Assign this issue to me

Image Redactor - Support Multiframe Image Redaction

Hi — I'd like to work on this issue. Plan: - Investigate current multiframe handling in the image redactor. - Add support to iterate over all frames in a multiframe...