Nicolas Patry

Small differences in numbers can be explained by hardware, torch version etc... Nothing can be done about it. For the difference in output the API uses a different default from...

Hi @Rishabh-Choudhry . This is impossible to do with `whisper`. Whisper simply doesn't work in such a way, it output "timestamp" tokens, roughly when it feels like. And that's all...

There's a PR opened for it: https://github.com/huggingface/transformers/pull/21427 If you look at it, it actually uncovered some issues with Whisper itself (in non timestamp mode, the default in `transformers`, not the...

This change was necessary to get some tests running. Namely testing that oneformer and the like are actually working. These models **do not** have a feature extractor, only a `ImageProcessor`....

> Also cc @Narsil - any ideas here? Yes, by default added tokens always use `lstrip/rstrip=True` which swallows prefix/suffix spaces (it's a convenience for so you don't have to worry...

Hi, why do you want that option ? Sorry, but we try to limit the number of parameters available (for simplicity). This is also not available in `transformers` 's pipeline...

Special tokens are meant to be non-readable, if you want to use readable tokens, couldn't you use regular added tokens ? (`tokenizer.add_tokens` vs `tokenizer.add_special_tokens` IIRC) Special tokens are special mostly...

Hi @Matthieu-Tinycoaching This is linked to: huggingface/api-inference-community#26 Community images do not implement: - private models - GPU inference - Acceleration So what you are seeing is quite normal and is...

Hi, You can test `transformers` based models with all the API features, not `sentence-transformers` at the moment. Also `feature-extraction` even in transformers does not have every optimizations enabled by default....

Anything specific on the machines you are running regarding AVX/SIMD ? What are the CPUs ? https://github.com/BurntSushi/memchr/issues/57 Might be showing some insights.