| [ |
| { |
| "optimizer": "adamw", |
| "category": "1. AdamW", |
| "fwd_bwd_ms": 44.51255885263284, |
| "step_ms": 3.5697922110557556, |
| "hessian_amortised_ms": 0.0, |
| "total_ms": 48.0823510636886, |
| "state_mb": 216.584192, |
| "state_multiple": 2.0 |
| }, |
| { |
| "optimizer": "mars", |
| "category": "2. Variance-reduced AdamW variant", |
| "fwd_bwd_ms": 43.19541243215402, |
| "step_ms": 6.120803082982699, |
| "hessian_amortised_ms": 0.0, |
| "total_ms": 49.31621551513672, |
| "state_mb": 324.876288, |
| "state_multiple": 3.0 |
| }, |
| { |
| "optimizer": "lion", |
| "category": "3. Memory-efficient", |
| "fwd_bwd_ms": 44.26518057783445, |
| "step_ms": 3.246958553791046, |
| "hessian_amortised_ms": 0.0, |
| "total_ms": 47.512139131625496, |
| "state_mb": 108.292096, |
| "state_multiple": 1.0 |
| }, |
| { |
| "optimizer": "muon", |
| "category": "4. Matrix-based", |
| "fwd_bwd_ms": 41.5445347627004, |
| "step_ms": 12.41101324558258, |
| "hessian_amortised_ms": 0.0, |
| "total_ms": 53.95554800828298, |
| "state_mb": 141.08672, |
| "state_multiple": 1.3028348809501296 |
| }, |
| { |
| "optimizer": "sophia", |
| "category": "5. Hessian-based (bonus)", |
| "fwd_bwd_ms": 44.653950879971184, |
| "step_ms": 2.7560430268446603, |
| "hessian_amortised_ms": 4.465395087997118, |
| "total_ms": 51.87538899481296, |
| "state_mb": 216.584192, |
| "state_multiple": 2.0 |
| } |
| ] |
