Training Masks - Githubissues

msakarvadia / llm_bias

Investigating if we can find circuits in LLMs that reinforce human-biases found in training data

MIT License

0 stars 0 forks source link

Open msakarvadia opened 4 months ago

msakarvadia commented 4 months ago

msakarvadia commented 4 months ago

[ ] Need to ensure that the LR is lower
[ ] Need to ensure that the mask is appended to comment rather than entire prompt
[ ] Need to Check what non-optimized model performance is with the initialized mask