Buckets:
| device=cuda params=25.34M bf16=True compile=False | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/68131e5bd685023a95e0b8ea1ebe089a721deb99/train | |
| downloading patrickfrank1/chess-positions split=test (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=test to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/68131e5bd685023a95e0b8ea1ebe089a721deb99/test | |
| unigram pass: 83 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/68131e5bd685023a95e0b8ea1ebe089a721deb99/train | |
| unigram pass: shard 1/83 done | |
| unigram pass: shard 25/83 done | |
| unigram pass: shard 50/83 done | |
| unigram pass: shard 75/83 done | |
| unigram pass: shard 83/83 done | |
| unigram pass: done, 3513966085 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/68131e5bd685023a95e0b8ea1ebe089a721deb99/test ... | |
| loaded 34396 val games | |
| train games=651466 val games=34396 | |
| step=50 loss=3.3652 lr=3.00e-05 tokens/s=124290 | |
| step=100 loss=2.7005 lr=6.00e-05 tokens/s=126410 | |
| step=150 loss=2.6164 lr=9.00e-05 tokens/s=126195 | |
| step=200 loss=2.5966 lr=1.20e-04 tokens/s=124980 | |
| step=250 loss=2.5905 lr=1.50e-04 tokens/s=126930 | |
| step=300 loss=2.5646 lr=1.80e-04 tokens/s=126094 | |
| device=cuda params=25.34M bf16=True compile=False | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/2a28ed489bc1146afc87ad40d400c2e483f6e55d/train | |
| downloading patrickfrank1/chess-positions split=test (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=test to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/2a28ed489bc1146afc87ad40d400c2e483f6e55d/test | |
| unigram pass: 84 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/2a28ed489bc1146afc87ad40d400c2e483f6e55d/train | |
| unigram pass: shard 1/84 done | |
| unigram pass: shard 25/84 done | |
| unigram pass: shard 50/84 done | |
| unigram pass: shard 75/84 done | |
| unigram pass: shard 84/84 done | |
| unigram pass: done, 4036950397 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/2a28ed489bc1146afc87ad40d400c2e483f6e55d/test ... | |
| loaded 34396 val games | |
| train games=751466 val games=34396 | |
| step=50 loss=3.3662 lr=3.00e-05 tokens/s=87110 | |
| step=50 system rss_gb=2.16 cpu_load=15.31 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=100 loss=2.6980 lr=6.00e-05 tokens/s=126084 | |
| step=100 system rss_gb=2.16 cpu_load=21.48 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=150 loss=2.6154 lr=9.00e-05 tokens/s=125962 | |
| step=150 system rss_gb=2.16 cpu_load=59.85 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=200 loss=2.5893 lr=1.20e-04 tokens/s=125905 | |
| step=200 system rss_gb=2.16 cpu_load=77.19 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=250 loss=2.5775 lr=1.50e-04 tokens/s=125865 | |
| step=250 system rss_gb=2.16 cpu_load=62.15 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=300 loss=2.5626 lr=1.80e-04 tokens/s=125940 | |
| step=300 system rss_gb=2.17 cpu_load=36.41 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=350 loss=2.5510 lr=2.10e-04 tokens/s=126011 | |
| step=350 system rss_gb=2.17 cpu_load=25.72 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=400 loss=2.5203 lr=2.40e-04 tokens/s=126137 | |
| step=400 system rss_gb=2.17 cpu_load=19.42 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=450 loss=2.5042 lr=2.70e-04 tokens/s=125999 | |
| step=450 system rss_gb=2.17 cpu_load=16.59 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.53 vram_total_gb=47.38 | |
| device=cuda params=25.34M bf16=True compile=False | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/06640a4c065c67065c568e0a18eed09e0e78930c/train | |
| downloading patrickfrank1/chess-positions split=test (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=test to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/06640a4c065c67065c568e0a18eed09e0e78930c/test | |
| unigram pass: 85 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/06640a4c065c67065c568e0a18eed09e0e78930c/train | |
| unigram pass: shard 1/85 done | |
| unigram pass: shard 25/85 done | |
| unigram pass: shard 50/85 done | |
| unigram pass: shard 75/85 done | |
| unigram pass: shard 85/85 done | |
| unigram pass: done, 4560982857 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/06640a4c065c67065c568e0a18eed09e0e78930c/test ... | |
| loaded 34396 val games | |
| train games=851466 val games=34396 | |
| step=50 loss=3.3627 lr=3.00e-05 tokens/s=124311 | |
| step=50 system rss_gb=2.25 cpu_load=59.02 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=100 loss=2.7047 lr=6.00e-05 tokens/s=126867 | |
| step=100 system rss_gb=2.25 cpu_load=78.61 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=31.53 vram_total_gb=47.38 | |
| step=150 loss=2.6206 lr=9.00e-05 tokens/s=126857 | |
| step=150 system rss_gb=2.25 cpu_load=70.50 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=200 loss=2.5855 lr=1.20e-04 tokens/s=126797 | |
| step=200 system rss_gb=2.25 cpu_load=42.35 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=250 loss=2.5783 lr=1.50e-04 tokens/s=126741 | |
| step=250 system rss_gb=2.25 cpu_load=25.58 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=300 loss=2.5582 lr=1.80e-04 tokens/s=126596 | |
| step=300 system rss_gb=2.26 cpu_load=19.74 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=350 loss=2.5462 lr=2.10e-04 tokens/s=126675 | |
| step=350 system rss_gb=2.26 cpu_load=19.26 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=400 loss=2.5146 lr=2.40e-04 tokens/s=126608 | |
| step=400 system rss_gb=2.26 cpu_load=18.12 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=450 loss=2.5093 lr=2.70e-04 tokens/s=126560 | |
| step=450 system rss_gb=2.26 cpu_load=28.27 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=500 loss=2.5013 lr=3.00e-04 tokens/s=126526 | |
| step=500 system rss_gb=2.26 cpu_load=27.87 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=550 loss=2.5009 lr=3.00e-04 tokens/s=126574 | |
| step=550 system rss_gb=2.26 cpu_load=60.51 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=600 loss=2.4822 lr=3.00e-04 tokens/s=126608 | |
| step=600 system rss_gb=2.26 cpu_load=85.48 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.03 vram_total_gb=47.38 | |
| step=650 loss=2.4957 lr=3.00e-04 tokens/s=126637 | |
| step=650 system rss_gb=2.26 cpu_load=62.95 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=700 loss=2.4661 lr=3.00e-04 tokens/s=126600 | |
| step=700 system rss_gb=2.26 cpu_load=35.46 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=750 loss=2.4467 lr=3.00e-04 tokens/s=126613 | |
| step=750 system rss_gb=2.26 cpu_load=24.64 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=800 loss=2.4196 lr=2.99e-04 tokens/s=126543 | |
| step=800 system rss_gb=2.26 cpu_load=24.43 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=850 loss=2.4230 lr=2.99e-04 tokens/s=126528 | |
| step=850 system rss_gb=2.26 cpu_load=20.77 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=900 loss=2.4115 lr=2.99e-04 tokens/s=126537 | |
| step=900 system rss_gb=2.26 cpu_load=16.17 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=950 loss=2.4042 lr=2.99e-04 tokens/s=125541 | |
| step=950 system rss_gb=2.26 cpu_load=13.50 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1000 loss=2.4111 lr=2.99e-04 tokens/s=124925 | |
| step=1000 system rss_gb=2.26 cpu_load=32.13 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d512/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d512_streaming | |
| step=1050 loss=2.4146 lr=2.98e-04 tokens/s=38784 | |
| step=1050 system rss_gb=2.44 cpu_load=40.46 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1100 loss=2.4069 lr=2.98e-04 tokens/s=126647 | |
| step=1100 system rss_gb=2.44 cpu_load=28.04 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1150 loss=2.4041 lr=2.98e-04 tokens/s=126541 | |
| step=1150 system rss_gb=2.44 cpu_load=21.82 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1200 loss=2.3911 lr=2.97e-04 tokens/s=126486 | |
| step=1200 system rss_gb=2.44 cpu_load=19.13 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1250 loss=2.3969 lr=2.97e-04 tokens/s=126515 | |
| step=1250 system rss_gb=2.44 cpu_load=17.52 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1300 loss=2.3788 lr=2.96e-04 tokens/s=126429 | |
| step=1300 system rss_gb=2.44 cpu_load=27.08 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1350 loss=2.3935 lr=2.96e-04 tokens/s=126462 | |
| step=1350 system rss_gb=2.44 cpu_load=25.33 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1400 loss=2.3877 lr=2.95e-04 tokens/s=126454 | |
| step=1400 system rss_gb=2.44 cpu_load=68.30 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1450 loss=2.3963 lr=2.95e-04 tokens/s=126480 | |
| step=1450 system rss_gb=2.44 cpu_load=74.18 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1500 loss=2.3874 lr=2.94e-04 tokens/s=126803 | |
| step=1500 system rss_gb=2.44 cpu_load=59.07 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1550 loss=2.3915 lr=2.94e-04 tokens/s=126647 | |
| step=1550 system rss_gb=2.44 cpu_load=33.63 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1600 loss=2.3857 lr=2.93e-04 tokens/s=126461 | |
| step=1600 system rss_gb=2.44 cpu_load=21.84 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1650 loss=2.3902 lr=2.93e-04 tokens/s=126441 | |
| step=1650 system rss_gb=2.44 cpu_load=19.66 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1700 loss=2.3959 lr=2.92e-04 tokens/s=126386 | |
| step=1700 system rss_gb=2.44 cpu_load=18.59 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1750 loss=2.3887 lr=2.91e-04 tokens/s=126477 | |
| step=1750 system rss_gb=2.44 cpu_load=21.34 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1800 loss=2.3857 lr=2.91e-04 tokens/s=126620 | |
| step=1800 system rss_gb=2.44 cpu_load=21.40 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1850 loss=2.3793 lr=2.90e-04 tokens/s=126560 | |
| step=1850 system rss_gb=2.44 cpu_load=39.13 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1900 loss=2.3831 lr=2.89e-04 tokens/s=126474 | |
| step=1900 system rss_gb=2.44 cpu_load=61.35 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=1950 loss=2.3775 lr=2.88e-04 tokens/s=126551 | |
| step=1950 system rss_gb=2.44 cpu_load=89.68 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=2000 loss=2.3704 lr=2.88e-04 tokens/s=126548 | |
| step=2000 system rss_gb=2.44 cpu_load=47.36 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=32.53 vram_total_gb=47.38 | |
| step=2000 val_loss=2.4746 val_acc=0.1361 | |
| step=2000 neighbor query material=-1 8/p4p2/4k2p/1pp1P1p1/3n1PP1/3BK2P/P7/8 w - - 0 1 | |
| top sim=1.000 material=+0 8/1p2kpn1/2p5/2P2pPp/1PB2P1P/4K3/8/8 b - - 0 1 | |
| step=2000 neighbor query material=+0 rn1q1rk1/pbp2pp1/1p1ppn1p/8/2PP3B/2P1PN2/P3BPPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+0 1r1r1nk1/pBq1bpp1/Pp1ppn1p/2p5/QP1P4/2P1PN1P/5PPB/3R1RK1 b - - 0 1 | |
| step=2000 neighbor query material=-7 8/6pk/7p/8/n3P3/4K3/1r3P1P/8 b - - 0 1 | |
| top sim=0.995 material=-14 8/2r5/8/1k4P1/nP6/K4P2/2q5/8 w - - 0 1 | |
| step=2000 neighbor board report saved to data/models/d512/neighbor_examples.html | |
| step=2000 eval plot saved to data/models/d512/pca_phase.png | |
| step=2000 board_turn_acc=0.8000 board_castle_acc=0.3000 board_piece_acc=0.1432 board_prior_baseline=0.5525 random_piece_acc=0.1182 span_piece_acc=0.1461 cos_to_move_child=0.9999 cos_to_corrupted=0.9986 test_r2_material=0.5029 pc1_piece_corr=-0.9893 | |
| saved best checkpoint to data/models/d512/best.pt (val_loss=2.4746) | |
| uploaded best_step2000 to patrickfrank1/chesspos-checkpoints/run_d512_streaming | |
| saved full checkpoint to data/models/d512/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d512_streaming | |
| step=2050 loss=2.3810 lr=2.87e-04 tokens/s=52785 | |
| step=2050 system rss_gb=2.77 cpu_load=21.08 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2100 loss=2.3812 lr=2.86e-04 tokens/s=126556 | |
| step=2100 system rss_gb=2.77 cpu_load=17.46 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2150 loss=2.3683 lr=2.85e-04 tokens/s=126838 | |
| step=2150 system rss_gb=2.77 cpu_load=15.90 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2200 loss=2.3680 lr=2.84e-04 tokens/s=126288 | |
| step=2200 system rss_gb=2.77 cpu_load=17.93 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2250 loss=2.3714 lr=2.83e-04 tokens/s=126483 | |
| step=2250 system rss_gb=2.77 cpu_load=23.57 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2300 loss=2.3700 lr=2.82e-04 tokens/s=126701 | |
| step=2300 system rss_gb=2.77 cpu_load=60.68 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2350 loss=2.3727 lr=2.81e-04 tokens/s=126352 | |
| step=2350 system rss_gb=2.77 cpu_load=74.93 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2400 loss=2.3702 lr=2.80e-04 tokens/s=126572 | |
| step=2400 system rss_gb=2.77 cpu_load=53.38 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2450 loss=2.3649 lr=2.79e-04 tokens/s=126619 | |
| step=2450 system rss_gb=2.77 cpu_load=31.27 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| step=2500 loss=2.3786 lr=2.78e-04 tokens/s=94074 | |
| step=2500 system rss_gb=2.77 cpu_load=19.89 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=35.02 vram_total_gb=47.38 | |
| resuming from data/models/d512/last.pt step=2000 best_val_loss=2.4746 | |
| device=cuda params=25.34M bf16=True compile=False | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/d9351dbe2c947adb6b85d68e99643ef0e4bca799/train | |
| downloading patrickfrank1/chess-positions split=test (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=test to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/d9351dbe2c947adb6b85d68e99643ef0e4bca799/test | |
| unigram pass: 90 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/d9351dbe2c947adb6b85d68e99643ef0e4bca799/train | |
| unigram pass: shard 1/90 done | |
| unigram pass: shard 25/90 done | |
| unigram pass: shard 50/90 done | |
| unigram pass: shard 75/90 done | |
| unigram pass: shard 90/90 done | |
| unigram pass: done, 6835147978 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/d9351dbe2c947adb6b85d68e99643ef0e4bca799/test ... | |
| loaded 62440 val games | |
| train games=1281459 val games=62440 | |
| step=2050 loss=2.3775 lr=2.87e-04 tokens/s=468208 | |
| step=2050 system rss_gb=3.54 cpu_load=22.24 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=45.20 vram_total_gb=47.38 | |
| step=2100 loss=2.3859 lr=2.86e-04 tokens/s=479840 | |
| step=2100 system rss_gb=3.54 cpu_load=37.26 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=17.23 vram_total_gb=47.38 | |
| step=2150 loss=2.3892 lr=2.85e-04 tokens/s=493243 | |
| step=2150 system rss_gb=3.54 cpu_load=45.14 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=18.14 vram_total_gb=47.38 | |
| step=2200 loss=2.3781 lr=2.84e-04 tokens/s=480512 | |
| step=2200 system rss_gb=3.55 cpu_load=52.01 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=18.48 vram_total_gb=47.38 | |
| step=2250 loss=2.3799 lr=2.83e-04 tokens/s=486504 | |
| step=2250 system rss_gb=3.55 cpu_load=60.65 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=18.48 vram_total_gb=47.38 | |
| step=2300 loss=2.3763 lr=2.82e-04 tokens/s=480161 | |
| step=2300 system rss_gb=3.56 cpu_load=64.92 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2350 loss=2.3736 lr=2.81e-04 tokens/s=491557 | |
| step=2350 system rss_gb=3.57 cpu_load=74.60 n_cpus=128.00 vram_alloc_gb=0.66 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2400 loss=2.3666 lr=2.80e-04 tokens/s=481798 | |
| step=2400 system rss_gb=3.58 cpu_load=77.84 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2450 loss=2.3786 lr=2.79e-04 tokens/s=493922 | |
| step=2450 system rss_gb=3.58 cpu_load=83.45 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2500 loss=2.3695 lr=2.78e-04 tokens/s=486712 | |
| step=2500 system rss_gb=3.59 cpu_load=86.29 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2550 loss=2.3636 lr=2.77e-04 tokens/s=498172 | |
| step=2550 system rss_gb=3.59 cpu_load=70.09 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2600 loss=2.3618 lr=2.76e-04 tokens/s=483144 | |
| step=2600 system rss_gb=3.59 cpu_load=59.92 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=25.16 vram_total_gb=47.38 | |
| step=2650 loss=2.3532 lr=2.75e-04 tokens/s=483761 | |
| step=2650 system rss_gb=3.60 cpu_load=52.90 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2700 loss=2.3484 lr=2.74e-04 tokens/s=479702 | |
| step=2700 system rss_gb=3.60 cpu_load=51.59 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2750 loss=2.3424 lr=2.73e-04 tokens/s=491890 | |
| step=2750 system rss_gb=3.60 cpu_load=44.81 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2800 loss=2.3445 lr=2.71e-04 tokens/s=481034 | |
| step=2800 system rss_gb=3.60 cpu_load=39.59 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2850 loss=2.3367 lr=2.70e-04 tokens/s=484313 | |
| step=2850 system rss_gb=3.60 cpu_load=35.43 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2900 loss=2.3318 lr=2.69e-04 tokens/s=480591 | |
| step=2900 system rss_gb=3.60 cpu_load=32.00 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=2950 loss=2.3308 lr=2.68e-04 tokens/s=470193 | |
| step=2950 system rss_gb=3.60 cpu_load=28.32 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| step=3000 loss=2.3226 lr=2.66e-04 tokens/s=474728 | |
| step=3000 system rss_gb=3.61 cpu_load=26.40 n_cpus=128.00 vram_alloc_gb=0.67 vram_used_gb=31.84 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d512/last.pt | |
Xet Storage Details
- Size:
- 19.6 kB
- Xet hash:
- a26e711fd385b3885842eee69efbbd9809d1034cd5466ca9afc1248f9e78d46a
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.