#Training‑free methods

Articles (2)

• 🇺🇸 ReplaceMe: Network Simplification via Depth Pruning and Transformer Block Linearization — 20/02/2026 [USA]

  arXiv:2505.02819v4 Announce Type: replace-cross Abstract: We introduce ReplaceMe, a generalized training-free depth pruning method that effectively replaces transformer blocks with a linear operation...

  Related: #Model compression, #Transformer architecture, #Depth pruning, #Performance retention
• 🇺🇸 Beyond Learning: A Training-Free Alternative to Model Adaptation — 19/02/2026 [USA]

  arXiv:2602.16189v1 Announce Type: cross Abstract: Despite the continuous research and evolution of language models, they sometimes underperform previous versions. Existing approaches to overcome thes...

  Related: #Model adaptation, #Resource efficiency, #Local modules, #Language models