YonghyunPark

Hi, and thank you for your awesome repository! I recently tried using this library to validate some attribution methods, specifically by utilizing [evaluator.evaluate](https://github.com/AI4LIFE-GROUP/OpenXAI/blob/main/openxai/evaluator.py). However, I discovered that it only uses...

Could you possibly share the pretrained weights?

1

I am trying to reproduce experiments on **nudity and violence** using your code, but I keep failing. If you don't mind, could you possibly share the pretrained weights?

[feature request] specify layer / timesteps

3

Thank you for your awesome work! It is easy to use without much modification. BTW, since attention map often use as a tool for analyze the layer's behavior, specify what...

TRAK on cuda version 11.8 have numerical issue

2

As the title mentioned, TRAK is experiencing numerical issues on the cuda version 11.8. - Minimum reproducible code ```python import torch from trak.projectors import ProjectionType, AbstractProjector, CudaProjector grad_dim = int(1e6)...

- minimal code for reproduce the error: ```python import torch import trak from trak.projectors import ProjectionType, AbstractProjector, CudaProjector print("trak.test_install:", trak.test_install(use_fast_jl=True)) grad_dim = int(1e6) projector = CudaProjector( grad_dim=grad_dim, proj_dim=32768, seed=42, proj_type=ProjectionType.normal,...

Natural language intermediate steps for GSM8K Aug.

1

Hello, I am currently try to reproduce the explicit CoT result from your paper, but realized that I need natural language intermediate steps for training GPT-2 for explicit CoT. Do...

Do you have any plans to share the code you used to create the 4by4 mult, 5by5 mult? Thanks in advance for your response.