ResearchDatasetsQwen2Qwen2FollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsConstrained LLM DecodingQwen2-14B INT816.52Inference Time (ms)14SteganographyQwen21.673Entropy (bit/token)9Training EfficiencyQwen2-57B-A14B Fine-grained39MFU5AutotuningQwen2-1.5B on vLLM concurrent harness10.2Fast Cluster Trials2