Closed mianqiu00 closed 3 months ago
Qnli
zero-shot
python inference.py -m Llama-2-7b-hf -d qnli -shots 0 --ranking_type prob
Our results: 48.82.
python inference.py -m Meta-Llama-3-8B -d qnli -shots 0 --ranking_type prob
Our results: 49.37.
one-shot
python inference.py -m Llama-2-7b-hf -d qnli -shots 1 --ranking_type prob
Our results: 49.42.
python inference.py -m Meta-Llama-3-8B -d qnli -shots 1 --ranking_type prob
Our results: 68.30.
32-shot
python inference.py -m Llama-2-7b-hf -d qnli -shots 32 --ranking_type prob
Our results: 55.85.
python inference.py -m Meta-Llama-3-8B -d qnli -shots 32 --ranking_type prob
Our results: 54.48.
Qnli
zero-shot
Our results: 48.82.
Our results: 49.37.
one-shot
Our results: 49.42.
Our results: 68.30.
32-shot
Our results: 55.85.
Our results: 54.48.