paulcx

> What kind of multilingual support are you interested in? The nodes' names are mathlib source filenames, it is impossible to replace them in current version. Can file names be...

ok, I have an initial proposal to expand MathlibExplorer, and I'm wondering if I could get your contact information?

I translated the mathlib pacakge name to Chinese. [mathlib_zh.json](https://github.com/user-attachments/files/15950416/mathlib_zh.json) [import_graph.txt](https://github.com/user-attachments/files/15950418/import_graph.txt)

Btw, the translation is done with the help of an LLM, so it is bound to have some errors, I just fix some simple ones.

feel free to use and share.

> Can you dump your `nvidia-smi`? NVIDIA-SMI 525.105.17Driver Version: 525.105.17CUDA Version: 12.0 | GPU | Name | Persistence-M | Bus-Id | Disp.A | Volatile Uncorr. ECC | GPU-Util | Compute...

After removing `--max-total-tokens 4096` from the docker command line and getting it to work again, what's going on with new update? @OlivierDehaene Additionally, I noticed that in the recently released...

@PoTaTo-Mika Is there a solution to this problem?

> Hi @paulcx 👋 thanks for opening the issue! > > I'm not a 100% sure I understand your question but if it's "can the output be truncated even if...

> Okay so there's actually a good reason for it defaulting to 100. The reason is that the `max_new_tokens` takes memory in the scheduling. So we have to pick some...