issues
search
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
Write a deterministic Model Generation Pipeline
#1
Closed
msakarvadia
closed
2 months ago
msakarvadia
commented
3 months ago
[ ] Model must take input embeddings
[ ] Need to be able to add mask to embeddings prior to inference
[ ] Need inference generation pipeline that is deterministic