Pedro Cuenca
Pedro Cuenca
Hello @goldingdamien! For transformers.js, I think you need to follow these steps (server environments): https://huggingface.co/docs/transformers.js/en/guides/private
The documentation builds locally for me, is there a way to re-run the CI job @merveenoyan? Or a hint as to why it might be failing @mishig25? This is how...
cc @merveenoyan
Working on Idefics 3 here: https://github.com/Blaizzy/mlx-vlm/pull/124
I agree: ``` >>> torch.finfo(torch.float16) finfo(resolution=0.001, min=-65504, max=65504, eps=0.000976562, smallest_normal=6.10352e-05, tiny=6.10352e-05, dtype=float16) ``` Would you like to open a PR to fix, @alankashkash?
A couple of repos for testing: - [This is a Qwen model](https://huggingface.co/pcuenq/Qwen2.5-0.5B-Instruct-with-new-merges-serialization) that was exported from transformers 4.45 and therefore uses the new tokenizer serialization format. - This one is...
In my opinion, I think upgrading `transformers` is easier.
Hi @CAD1986! Sounds like a very interesting use case. This repo is only watched by a few maintainers, I'd recommend you provide a few more details as a post in...
Hi @Blaizzy, welcome to Swift-land :) Incidentally, I started working on this before seeing your PR. I have a preliminary question, did you manage to get _any_ smolvlm running, even...
Hi @wizzard0! Actually, the typo is also present in the official documentation (see Step 3 in the [JSON-based tool calling section](https://llama.meta.com/docs/model-cards-and-prompt-formats/llama3_1/#json-based-tool-calling)). I doubt that the system prompt has to be...