Multi-Preference-Alignment

Preference Benchmarking

python benchmarking_preferences/rm_bench/rm_bench_yes_no.py \
  --hf_key "your_huggingface_token_here" \
  --hf_user "your_hf_username" \
  --model_name "model_name"

python benchmarking_preferences/reward_bench/reward_bench_yes_no.py \
  --hf_key "your_huggingface_token_here" \
  --hf_user "your_hf_username" \
  --model_name "model_name"