Shubhashis Roy Dipta

Results 11 issues of Shubhashis Roy Dipta

If I misspell the template file or maybe delete the template file at some point, the extension does not work. It opens a folder with no files in it. Better...

bug

## ❓ Questions and Help **Description** For a large corpus, I couldn't find any way to use an iterator in the dataset like the PyTorch dataset. Is it possible to...

I am getting this error ``` /tmp/tmpfk2hveif/main.c:4:10: fatal error: Python.h: No such file or directory 4 | #include | ^~~~~~~~~~ ``` I have used the following command from terminal: ```...

Do you have any idea how we can use the pre-trained model to make inferences for text-to-video retrieval, i.e., the MSRVTT-1kA dataset? Thanks for the help.

Hello, Is there any approximate time when the 6b model will be available that had been used to get the rank 1 result on the [MSR-VTT leaderboard](https://paperswithcode.com/sota/zero-shot-video-retrieval-on-msr-vtt)?

### Feature request Can we have a utility function that will use `load_from_disk` when given the local path and `load_dataset` if given an HF dataset? It can be something as...

enhancement

Hi, any plan to support NLLB Mixture of Expert model?

I am getting very low IFeval score compared to their [paper](https://arxiv.org/abs/2505.09388) Following is the cmd I have used (followed [this](https://github.com/EleutherAI/lm-evaluation-harness/issues/3118#issuecomment-3069732745) and [this](https://github.com/EleutherAI/lm-evaluation-harness/issues/2941#issuecomment-2870733949) with no luck) ``` python -m lm_eval --model...

Sorry if I have missed the feature, I want to train a multi-turn grpo model with reward on each step and also final reward.

feature request