List your company

Reasoning on SenTSR-Bench

82.7What Happened ScoreTSLM + GRLM (Claude3.7-Vision)
49.9458.44566.9575.455Feb 23, 2026

Evaluation results

MethodLinks
TSLM + GRLM (Claude3.7-Vision)Paradigm=RL-Injection2026.0282.766.159.769.526.112.1
GRLM (Claude3.7-Vision)Paradigm=Few-shot2026.0282.455.255.564.3—3.7
TSLM + GRLM (Claude3.7-Text)Paradigm=RL-Injection2026.0277.962.754.2651822.4
GRLM (Claude3.7-Vision)Paradigm=Zero-shot2026.0276.454.255.562——
TSLM + GRLM (Claude3.7-Vision)Paradigm=SFT-Injection2026.0275.658.864.966.521.17.3
TSLM + GRLM (Claude3.7-Text)Paradigm=SFT-Injection2026.0274.260.355.863.415.519.4
GRLM (Claude3.7-Text)Paradigm=Zero-shot2026.0271.240.947.353.1——
GRLM (Claude3.7-Text)Paradigm=Few-shot2026.0269.156.150.958.7—10.5
TSLM (Qwen-VL-3B)Paradigm=SFT2026.025356.754.854.9——
TSLM (Qwen-VL-3B)Paradigm=RL2026.0251.259.454.655.1——