Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Training Efficiency on Qwen2-57B-A14B Fine-grained

39MFU

MCore w/ Folding

8.73616.59324.4532.307Apr 21, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.04
39
2025.04
35.3
2025.04
25.4
2025.04
23.1
2025.04
9.9