{ "metadata": { "width": 1344, "height": 768, "frames": 124, "steps": 12, "sampler_step": 1, "seed": 440420, "text_tokens": 100, "block_index": 24, "attention": "sage2", "hidden_shape": [ 37810, 5376 ], "h_msa_shape": [ 37810, 5376 ], "segments": [ [ 0, 100, 1 ], [ 100, 514, 2 ], [ 514, 37810, 0 ] ] }, "sequence": 37810, "conditioning_stop": 514, "warmup": 3, "iterations": 10, "measurement_policy": { "component_timings": "CUDA-synchronized nested probes; attribution only", "residual_timings": "difference of component p50 values; labeled estimates", "end_to_end": "must be measured separately without component synchronization" }, "sage2": { "version": "2.2.0", "timings": { "attention_forward": { "count": 10, "mean_s": 0.36555117740063, "p50_s": 0.2388111045002006, "p90_s": 0.4044706951128315, "p95_s": 0.921806076535721, "p99_s": 1.3356743816740346, "min_s": 0.2377422249992378, "max_s": 1.4391414579586126 }, "qk_quantize": { "count": 10, "mean_s": 0.008795083011500537, "p50_s": 0.007684325042646378, "p90_s": 0.010721880599157883, "p95_s": 0.013224297293345438, "p99_s": 0.015226230648695492, "min_s": 0.00766948499949649, "max_s": 0.015726713987533003 }, "total": { "count": 10, "mean_s": 0.39030183949507774, "p50_s": 0.2600231445103418, "p90_s": 0.4261628873937293, "p95_s": 0.947522030692197, "p99_s": 1.3646093453309733, "min_s": 0.25837415299611166, "max_s": 1.4688811739906669 }, "v_quantize": { "count": 10, "mean_s": 0.01344956418615766, "p50_s": 0.010653970995917916, "p90_s": 0.018150134984171014, "p95_s": 0.02297045400191563, "p99_s": 0.02682670921611134, "min_s": 0.010557003959547728, "max_s": 0.027790773019660264 }, "framework_and_k_smoothing_estimate": { "count": 10, "mean_s": 0.002873743971576914, "p50_s": 0.002873743971576914, "p90_s": 0.002873743971576914, "p95_s": 0.002873743971576914, "p99_s": 0.002873743971576914, "min_s": 0.002873743971576914, "max_s": 0.002873743971576914 } }, "checksum": 9190064.0 }, "sol": [] }