Karandeep singh comments

Results 35 comments of


                                            Karandeep singh

Output tokens logits

Any update on this?

Error(s) in loading state_dict for VideoLLAMA

This weight works in my case - https://huggingface.co/wangrongsheng/MiniGPT4-7B/tree/main

After completing the fine-tuning of 'visionbranch stage2', what should be its loss value to be considered as having correctly converged?

Any updates on this?

Allow downloading just some columns of a dataset

Hi @lhoestq, I have an audio dataset of 250GB on the huggingface hub in parquet format. I only wanted to load the text column. It is taking a lot of...

Discussion on training the AR model with reference audio for dubbing.

Hi @Plachtaa, M4T has the potential to be used for dubbing, but it would require architectural changes to incorporate both text and speech as controllable parameters. The VALL-E-X architecture already...

Discussion on training the AR model with reference audio for dubbing.

Thank you, @Plachtaa and @UniversalTechno, for your suggestion. I'll take a look at the duration predictor component of M4T. I have a quick question, and I apologize if it sounds...

Discussion on training the AR model with reference audio for dubbing.

Hi @Plachtaa, I'm currently facing a challenge and would appreciate your expertise in helping me understand why my loss is not converging. I'd like to provide some additional information so...

push_to_hub doesn't push checkpoint folder while training

Any updates? Why is it happening?

Missing documentation on how to train a model

Is there an estimated timeline for when we will have documentation and training recipes for fariseq2 models specifically **w2v-bert**?

Write w2v-BERT pretraining recipe

Hi @cbalioglu, Any update on this... We want to do the continual training of w2v-bert on specific Indic low-resource languages on audio-only data. Any suggestion on how should we approach...