List your company

Coreference Resolution Calibration Evaluation on WinoQueer

27.7GayDeepSeek-8B
1.5968.37315.1521.927Jan 12, 2026

Evaluation results

MethodLinks
DeepSeek-8BParameters=8B2026.0127.783.825.891
Qwen2.5-7BParameters=7B2026.0118.989.891.978.8
GPT-J-6BParameters=6B2026.0112.17981.670
Llama-3.1-8BParameters=8B2026.018.3737665.7
Falcon3-7BParameters=7B2026.015.971.568.643.2
Gemma-2-9BParameters=9B2026.012.622.158.618.2