Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and a Fused Chunked KL Loss Paper • 2608.03796 • Published 20 days ago • 15
huggingfacemultiversecomputing/2026-01-16T09-54-20plus00-00_gdpval Viewer • Updated Jan 16 • 220 • 201
huggingfacemultiversecomputing/2026-01-16T09-54-20plus00-00_gdpval Viewer • Updated Jan 16 • 220 • 201
huggingfacemultiversecomputing/2026-01-15T15-06-36plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 36
huggingfacemultiversecomputing/2026-01-15T15-06-36plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 36
huggingfacemultiversecomputing/2026-01-15T14-55-14plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 34
huggingfacemultiversecomputing/2026-01-15T14-55-14plus00-00_gdpval Viewer • Updated Jan 15 • 220 • 34