Yunzhi Yao
Yunzhi Yao
 Maybe you can use this code to adapt to llama. @JX-Zhang98
> I tried running the `solvers/textsearch.py` on the 4_options/dev.jsonl: > > * for `topn=10` the accuracy was 28.7% > * for `topn=5` the accuracy was 27.7% Me too. I also...
Thank you for your attention, we will trace the error and fix it soon.
same error
It currently does not support multi-gpu and I will add this feature recently. Will let you know when I tackle it. You can try to move the tensor to the...
Thanks for your suggestion, we would support these method in the future.
We also welcome pull request since we're busy at the time and it would take a while for us the do this.
Hi, we think it is practical, but we are too busy now to implement this. We would support this later and welcome a pull request.
Sorry for being late, Here are some solutions for the time being. 1. Use `nnsight` or `transformer_lens` to conduct causal tracing to decide the layer for updating. These are empirical...
Quick update: 1. I updated the requirements, and the current code can support the Qwen3 series code. 2. For the Qwen3‘s suggested layer. I write a code for casual tracing...