Announcement_trim_icml
Thrilled to share that our paper TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning has been accepted to ICML 2026! 🎉
Thrilled to share that our paper TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning has been accepted to ICML 2026! 🎉