Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and a Fused Chunked KL Loss Paper • 2608.03796 • Published 11 days ago • 14
huggingfacemultiversecomputing/2026-01-16T09-54-20plus00-00_gdpval Viewer • Updated Jan 16 • 220 • 47
huggingfacemultiversecomputing/2026-01-16T09-54-20plus00-00_gdpval Viewer • Updated Jan 16 • 220 • 47
huggingfacemultiversecomputing/2026-01-15T15-06-36plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 56
huggingfacemultiversecomputing/2026-01-15T15-06-36plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 56
huggingfacemultiversecomputing/2026-01-15T14-55-14plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 70
huggingfacemultiversecomputing/2026-01-15T14-55-14plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 70