When using LM Studio or viewing model cards, it is difficult to find the model context length… That’s the first thing I’m interested in. I have no use for a model with 2048 context length, I want DOCUMENTS. I want to chat with a library.
Is context length not considered? To me its one of the most important variables. It should be in the model name
Usually in model cards there is information provided on the context length of the model. Qwen, Mistral, Llama, for those they always write context lengths. Fine-tuned versions rarely differ in context length, so finding the base version of the model and its card should be enough
@entfane Yeah not good enough. You probably mean in the model description. That’s volunteered by the model owner. I’m talking about having context length part of the basic embedded information right next to model size. It is that important
I agree with you that It would be good to have, though I assume it is perceived less important due to the ability to scale the context. For example Qwen3 Coder 480B model card tells the context size that was used for pre-training but also they mention how it could be extended and up to what size.
The code related to the display of something on the Hub is mostly in this repository for the frontend part.
However, I think it would be faster to submit your requests in the Community Feedback channel of the HF Discord.
@John6666 I don’t use discord. Help a brother out and submit the feedback. kthx