Bryce Meyer
Bryce Meyer
# Description Everything that was previously in `transformer_lens/components` has been refactored so that every class has a separate file. I also simplified and improved a couple things here and there...
When the plugin site is in use in conjecture with amazon s3 the plugin does not request the image properly after first cropping. It could be possible that the plugin...
### Description Whenever MPS is used in a runner the following error is generated... > RuntimeError: MPS backend out of memory (MPS allocated: 0 bytes, other allocations: 0 bytes, max...
# Description Right now this is a draft in order to discuss and experiment with. We still need to add Baichuan 1, and there is currently an issue with compatibility...
### Proposal The model configuration needs to be tested to make sure that they are able to loaded by the hooked transformer. ### Motivation To test model configuration in isolation....
### Proposal Right now there are two modules in the code that are organized to basically do the same thing. The `utils.py` file should be broken up and organized into...
# Description @fakerybakery We have to reopen this to address an issue that was found when the build docs was ran. The model has `window_size` set to None, but it...
# Description Changes the dependencies to allow more flexibility with bear type. Fixes #665 ## Type of change Please delete options that are not relevant. - [ ] Bug fix...
**Describe the bug** Originally from @ArthurConmy via slack > We seem to use batch_addmm stuff ([link](https://github.com/TransformerLensOrg/TransformerLens/blob/cb5017ad0f30cde0d3ac0b0f863c27fbec964c28/transformer_lens/utilities/addmm.py#L21)) as GPT-2 uses Conv1D ([link](https://github.com/huggingface/transformers/e68ec18/main/src/transformers/models/gpt2/modeling_gpt2.py#L156)), but this doesn't really fix the problem as Pythia...
The whole Qwen model family seems to be pretty inaccurate. I have not done complete benchmarks to determine where the issue is yet. That still needs to be done to...