rome
rome copied to clipboard
kmeng01
Locating and editing factual associations in GPT (NeurIPS 2022)
Thank you very much for making this awesome work publicly available. I'm working on extending ROME to understanding and editing the "values" representations that the model knows (as in, human...
Hi, I was wondering if you guys can tell me how I can generate weights for distilgpt2 for the mend/efk baselines, similar to what you have for gpt2-xl here: https://rome.baulab.info/data/weights/....
Hello, Due to memory limitations, I need to store ROME's cache somewhere other than my home directory. What would be the best way to specify the new cache dir's location...
Here's my function to look at the importances of individual neurons using your causal trace methodology, and then plot them. I don't know if you'll actually want to merge this...
working on some code to expand your causal trace demo to look at the prob contributions by individual neuron rather than layer. It's pretty messy still but if you're interested...
Hi, When running experiments.evaluate on ROME on gpt2-xl, I get an OOM after 4 cases on an 11 GB RTX 2080 Ti GPU. Given that each case runs sequentially, is...
I've found what I believe to be a bug in the implementation of the fine-tuning baseline which would yield incorrect results when the target is longer than one token. Looking...
I tried to replay the causal tracing on gpt-xl and llama-7b, but it failed in some ways 1. Failed on `predict_token` ``` def predict_token(mt, prompts, return_p=False): inp = make_inputs(mt.tokenizer, prompts)...
Hello, can i ask some questions for you? why do I still have problems with the environment configuration now? Is the project no longer updated? Is it necessary to continue...
