jhyuckkim/DeepSeek-V2-Lite-Dense-3B-CP-K6
0559
Align naming and results with the published arXiv version
Match Qwen3 table convention, drop PPL column
Link released rows by name instead of a separate column
Show full comparison table and link the paper
Remove internal paths from training info
Update citation, drop internal paths
Add model card
Add config with use_cache enabled
Add distilled dense student weights
initial commit
