#Model compression
Latest news articles tagged with "Model compression". Follow the timeline of events, related topics, and entities.
Articles (2)
-
🇺🇸 Sign Lock-In: Randomly Initialized Weight Signs Persist and Bottleneck Sub-Bit Model Compression
[USA]
arXiv:2602.17063v1 Announce Type: cross Abstract: Sub-bit model compression seeks storage below one bit per weight; as magnitudes are aggressively compressed, the sign bit becomes a fixed-cost bottle...
Related: #Weight sign dynamics in deep learning, #Statistical analysis of stochastic training, #Optimization techniques for sub‑bit quantization, #Theoretical foundations of neural network sparsity -
🇺🇸 ReplaceMe: Network Simplification via Depth Pruning and Transformer Block Linearization
[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: #Transformer architecture, #Depth pruning, #Training‑free methods, #Performance retention