TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning Paper • 2510.07118 • Published Oct 8, 2025 • 9
amogh98/Llama-3.2-1B-Instruct-beam_search_64-completions-a100 Viewer • Updated May 6, 2025 • 1.01k • 5
amogh98/Llama-3.2-1B-Instruct-beam_search_16-completions-a100 Viewer • Updated May 6, 2025 • 2.01k • 8