4 Aug 2026
Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and a Fused Chunked KL Loss
4 Aug 2026 · Preprint · arXiv · Computer Vision and Pattern Recognition
BR
Bakbergen Ryskulov
@bakbergen
- publications
- 2
- citations
- 4
- h-index
- 1
Publications2
21 Aug 2026
Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMs
21 Aug 2026 · Preprint · arXiv · Computer Vision and Pattern Recognition