Add Sst2

[mianqiu00]

• Sst2

• choices: [negative, positive]

• zero-shot

  python inference.py -m Llama-2-7b-hf -d sst2 -shots 0 --ranking_type prob

  Our results: 59.75.

  python inference.py -m Meta-Llama-3-8B -d sst2 -shots 0 --ranking_type prob

  Our results: 74.89.

• one-shot

  python inference.py -m Llama-2-7b-hf -d sst2 -shots 1 --ranking_type prob

  Our results: 49.08.

  python inference.py -m Meta-Llama-3-8B -d sst2 -shots 1 --ranking_type prob

  Our results: 92.43.

• 32-shot

  python inference.py -m Llama-2-7b-hf -d sst2 -shots 32 --ranking_type prob

  Our results: 87.16.

  python inference.py -m Meta-Llama-3-8B -d sst2 -shots 32 --ranking_type prob

  Our results: 94.72.