To whom it may concern, First, here is some context. As part of a research project, I want to distill the Whisper model on a dataset of non-native English speakers. By distillThis text will be hiddening Whisper, the goal is to create a more streamlined and cost-efficient model for speech recognitio…

Hey, I know this is rather old post, but I bumped into it when trying to understand how to perform only 1 Whisper Decoder step (overall how its forward method works), which took me longer than I would’ve liked. Therefore I’ve decided to post my understanding here. I am using WhisperForConditionalGe…

Disparity between output from `forward` and `generate` for greedy search (using Whisper)

🤗Transformers

system Closed August 11, 2024, 4:12pm 4

This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.

Topic		Replies	Views
Decode whisper logits to transcript using forward instead of generate method 🤗Transformers	3	1854	December 20, 2022
How to generate without decoding? 🤗Transformers	1	371	December 13, 2023
Difference in trainer.predict() and model.generate() for LM 🤗Transformers	0	1805	July 5, 2023
Understanding Output of `PreTrainedModel.forward` Beginners	2	2017	February 12, 2024
Whisper fine-tuning without Seq2SeqTrainer Intermediate	0	350	December 15, 2023

Disparity between output from `forward` and `generate` for greedy search (using Whisper)

Related topics