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

Qwen2

Benchmarks

Task NameDataset NameSOTA ResultTrend
Constrained LLM DecodingQwen2-14B INT8
Inference Time (ms)16.52
14
SteganographyQwen2
Entropy (bit/token)1.6734
9
Training EfficiencyQwen2-57B-A14B Fine-grained
MFU39
5
AutotuningQwen2-1.5B on vLLM concurrent harness
Fast Cluster Trials10.2
2
Showing 4 of 4 rows