Buckets:
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/5af6afb2cc1bd7892fa77391bd19eb4ebec4b8ad/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/5af6afb2cc1bd7892fa77391bd19eb4ebec4b8ad/test | |
| unigram pass: 96 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/5af6afb2cc1bd7892fa77391bd19eb4ebec4b8ad/train | |
| unigram pass: done, 9788965521 tokens counted (from sidecars) | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/5af6afb2cc1bd7892fa77391bd19eb4ebec4b8ad/test ... | |
| loaded 91781 val games | |
| train games=1841088 val games=91781 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/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/8b7c44393948b424d81181e7f3dff3e715e15c32/test | |
| unigram pass: 99 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/train | |
| unigram pass: 3 shards without sidecar histograms | |
| unigram pass: shard 1/3 done | |
| unigram pass: shard 3/3 done | |
| unigram pass: done, 11392841470 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/test ... | |
| loaded 91781 val games | |
| train games=2141088 val games=91781 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/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/8b7c44393948b424d81181e7f3dff3e715e15c32/test | |
| unigram pass: 99 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/train | |
| unigram pass: 3 shards without sidecar histograms | |
| unigram pass: shard 1/3 done | |
| unigram pass: shard 3/3 done | |
| unigram pass: done, 11392841470 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/8b7c44393948b424d81181e7f3dff3e715e15c32/test ... | |
| loaded 91781 val games | |
| train games=2141088 val games=91781 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| device=cuda params=2.43M bf16=True compile=True | |
| downloading patrickfrank1/chess-positions split=train (unbounded) ... | |
| downloaded patrickfrank1/chess-positions split=train to /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| device=cuda params=2.43M 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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| step=50 loss=3.7593 lr=3.00e-05 tokens/s=975024 | |
| step=50 system rss_gb=6.27 cpu_load=32.87 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=39.39 vram_total_gb=47.38 | |
| step=100 loss=3.0555 lr=6.00e-05 tokens/s=2860350 | |
| step=100 system rss_gb=6.27 cpu_load=31.92 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=31.19 vram_total_gb=47.38 | |
| step=150 loss=2.7771 lr=9.00e-05 tokens/s=2858169 | |
| step=150 system rss_gb=6.27 cpu_load=31.94 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=31.19 vram_total_gb=47.38 | |
| step=200 loss=2.6536 lr=1.20e-04 tokens/s=2837527 | |
| step=200 system rss_gb=6.27 cpu_load=28.58 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=250 loss=2.6018 lr=1.50e-04 tokens/s=2833497 | |
| step=250 system rss_gb=6.26 cpu_load=27.53 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=300 loss=2.5736 lr=1.80e-04 tokens/s=2890490 | |
| step=300 system rss_gb=6.26 cpu_load=25.46 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=350 loss=2.5597 lr=2.10e-04 tokens/s=2872964 | |
| step=350 system rss_gb=6.27 cpu_load=22.05 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=400 loss=2.5365 lr=2.40e-04 tokens/s=2862996 | |
| step=400 system rss_gb=6.26 cpu_load=20.54 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=450 loss=2.5186 lr=2.70e-04 tokens/s=2896143 | |
| step=450 system rss_gb=6.26 cpu_load=19.85 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=500 loss=2.5018 lr=3.00e-04 tokens/s=2843515 | |
| step=500 system rss_gb=6.26 cpu_load=21.70 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=550 loss=2.4912 lr=3.00e-04 tokens/s=2856995 | |
| step=550 system rss_gb=6.26 cpu_load=25.16 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=600 loss=2.4780 lr=3.00e-04 tokens/s=2881010 | |
| step=600 system rss_gb=6.26 cpu_load=27.27 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=650 loss=2.4753 lr=3.00e-04 tokens/s=2862995 | |
| step=650 system rss_gb=6.26 cpu_load=26.00 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=700 loss=2.4716 lr=3.00e-04 tokens/s=2834671 | |
| step=700 system rss_gb=6.26 cpu_load=24.13 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=750 loss=2.4684 lr=3.00e-04 tokens/s=2836337 | |
| step=750 system rss_gb=6.26 cpu_load=32.40 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=800 loss=2.4631 lr=2.99e-04 tokens/s=2835391 | |
| step=800 system rss_gb=6.26 cpu_load=41.56 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=850 loss=2.4607 lr=2.99e-04 tokens/s=2845619 | |
| step=850 system rss_gb=6.26 cpu_load=51.56 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=900 loss=2.4593 lr=2.99e-04 tokens/s=2843739 | |
| step=900 system rss_gb=6.26 cpu_load=62.67 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=950 loss=2.4521 lr=2.99e-04 tokens/s=2868660 | |
| step=950 system rss_gb=6.26 cpu_load=69.42 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1000 loss=2.4514 lr=2.99e-04 tokens/s=2866258 | |
| step=1000 system rss_gb=6.26 cpu_load=73.54 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=1050 loss=2.4428 lr=2.98e-04 tokens/s=1390427 | |
| step=1050 system rss_gb=6.28 cpu_load=71.00 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1100 loss=2.4186 lr=2.98e-04 tokens/s=2844511 | |
| step=1100 system rss_gb=6.28 cpu_load=63.35 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1150 loss=2.3891 lr=2.98e-04 tokens/s=2829866 | |
| step=1150 system rss_gb=6.28 cpu_load=57.54 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1200 loss=2.3866 lr=2.97e-04 tokens/s=2846306 | |
| step=1200 system rss_gb=6.28 cpu_load=48.09 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1250 loss=2.3819 lr=2.97e-04 tokens/s=2859557 | |
| step=1250 system rss_gb=6.28 cpu_load=39.49 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1300 loss=2.3765 lr=2.96e-04 tokens/s=2815518 | |
| step=1300 system rss_gb=6.28 cpu_load=35.00 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1350 loss=2.3686 lr=2.96e-04 tokens/s=2830573 | |
| step=1350 system rss_gb=6.28 cpu_load=31.25 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1400 loss=2.3623 lr=2.95e-04 tokens/s=2845669 | |
| step=1400 system rss_gb=6.28 cpu_load=29.39 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1450 loss=2.3587 lr=2.95e-04 tokens/s=2834126 | |
| step=1450 system rss_gb=6.28 cpu_load=27.73 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1500 loss=2.3480 lr=2.94e-04 tokens/s=2872222 | |
| step=1500 system rss_gb=6.28 cpu_load=25.80 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1550 loss=2.3390 lr=2.94e-04 tokens/s=2842938 | |
| step=1550 system rss_gb=6.28 cpu_load=22.85 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1600 loss=2.3286 lr=2.93e-04 tokens/s=2828329 | |
| step=1600 system rss_gb=6.28 cpu_load=21.66 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1650 loss=2.3233 lr=2.93e-04 tokens/s=2868984 | |
| step=1650 system rss_gb=6.28 cpu_load=22.01 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1700 loss=2.3182 lr=2.92e-04 tokens/s=2839497 | |
| step=1700 system rss_gb=6.28 cpu_load=22.27 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1750 loss=2.3178 lr=2.91e-04 tokens/s=2835782 | |
| step=1750 system rss_gb=6.28 cpu_load=22.99 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1800 loss=2.3122 lr=2.91e-04 tokens/s=2834153 | |
| step=1800 system rss_gb=6.28 cpu_load=21.66 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1850 loss=2.3104 lr=2.90e-04 tokens/s=2863379 | |
| step=1850 system rss_gb=6.28 cpu_load=24.28 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1900 loss=2.3062 lr=2.89e-04 tokens/s=2863255 | |
| step=1900 system rss_gb=6.28 cpu_load=30.96 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1950 loss=2.3022 lr=2.88e-04 tokens/s=2836582 | |
| step=1950 system rss_gb=6.28 cpu_load=27.28 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=2000 loss=2.3030 lr=2.88e-04 tokens/s=2835529 | |
| step=2000 system rss_gb=6.28 cpu_load=29.65 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=2000 val_loss=2.4964 val_acc=0.1449 | |
| resuming from data/models/d256l3/last.pt step=1000 best_val_loss=inf | |
| device=cuda params=2.43M 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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| step=1050 loss=2.4435 lr=2.98e-04 tokens/s=1215703 | |
| step=1050 system rss_gb=6.60 cpu_load=28.92 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=39.39 vram_total_gb=47.38 | |
| step=1100 loss=2.4190 lr=2.98e-04 tokens/s=2868205 | |
| step=1100 system rss_gb=6.60 cpu_load=27.29 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=31.19 vram_total_gb=47.38 | |
| step=1150 loss=2.3922 lr=2.98e-04 tokens/s=2856521 | |
| step=1150 system rss_gb=6.60 cpu_load=22.98 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=31.19 vram_total_gb=47.38 | |
| step=1200 loss=2.3864 lr=2.97e-04 tokens/s=2838456 | |
| step=1200 system rss_gb=6.60 cpu_load=20.67 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1250 loss=2.3816 lr=2.97e-04 tokens/s=2832160 | |
| step=1250 system rss_gb=6.60 cpu_load=20.62 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1300 loss=2.3729 lr=2.96e-04 tokens/s=2886951 | |
| step=1300 system rss_gb=6.60 cpu_load=19.32 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1350 loss=2.3719 lr=2.96e-04 tokens/s=2865465 | |
| step=1350 system rss_gb=6.60 cpu_load=19.90 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1400 loss=2.3637 lr=2.95e-04 tokens/s=2856871 | |
| step=1400 system rss_gb=6.59 cpu_load=19.24 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1450 loss=2.3596 lr=2.95e-04 tokens/s=2892391 | |
| step=1450 system rss_gb=6.59 cpu_load=17.92 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1500 loss=2.3480 lr=2.94e-04 tokens/s=2844255 | |
| step=1500 system rss_gb=6.59 cpu_load=19.94 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1550 loss=2.3404 lr=2.94e-04 tokens/s=2859029 | |
| step=1550 system rss_gb=6.59 cpu_load=23.97 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1600 loss=2.3267 lr=2.93e-04 tokens/s=2887290 | |
| step=1600 system rss_gb=6.59 cpu_load=21.76 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1650 loss=2.3238 lr=2.93e-04 tokens/s=2872062 | |
| step=1650 system rss_gb=6.59 cpu_load=24.55 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1700 loss=2.3182 lr=2.92e-04 tokens/s=2842897 | |
| step=1700 system rss_gb=6.59 cpu_load=22.94 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1750 loss=2.3148 lr=2.91e-04 tokens/s=2847785 | |
| step=1750 system rss_gb=6.59 cpu_load=44.51 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1800 loss=2.3108 lr=2.91e-04 tokens/s=2846813 | |
| step=1800 system rss_gb=6.60 cpu_load=48.39 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| step=1850 loss=2.3059 lr=2.90e-04 tokens/s=2860793 | |
| step=1850 system rss_gb=6.59 cpu_load=58.86 n_cpus=128.00 vram_alloc_gb=1.08 vram_used_gb=40.57 vram_total_gb=47.38 | |
| resuming from data/models/d256l3/last.pt step=1000 best_val_loss=inf | |
| device=cuda params=2.43M 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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/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/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test | |
| unigram pass: 101 shards in /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/train | |
| unigram pass: 5 shards without sidecar histograms | |
| unigram pass: shard 1/5 done | |
| unigram pass: shard 5/5 done | |
| unigram pass: done, 12388510313 tokens counted | |
| frequency-aware loss weighting enabled (streaming counts) | |
| loading val split from /workspace/chesspos/data/hf-cache/datasets--patrickfrank1--chess-positions/snapshots/db565951a9f46c204f2dc8079e8e4c4b048d9b17/test ... | |
| loaded 122720 val games | |
| train games=2326469 val games=122720 | |
| step=1050 loss=2.4481 lr=2.98e-04 tokens/s=543811 | |
| step=1050 system rss_gb=5.14 cpu_load=54.86 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=20.06 vram_total_gb=47.38 | |
| step=1100 loss=2.4242 lr=2.98e-04 tokens/s=2838745 | |
| step=1100 system rss_gb=5.15 cpu_load=49.02 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1150 loss=2.4019 lr=2.98e-04 tokens/s=2793449 | |
| step=1150 system rss_gb=5.15 cpu_load=48.62 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1200 loss=2.3960 lr=2.97e-04 tokens/s=2842259 | |
| step=1200 system rss_gb=5.15 cpu_load=44.00 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1250 loss=2.3864 lr=2.97e-04 tokens/s=2820632 | |
| step=1250 system rss_gb=5.16 cpu_load=40.40 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1300 loss=2.3887 lr=2.96e-04 tokens/s=2794656 | |
| step=1300 system rss_gb=5.16 cpu_load=38.04 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1350 loss=2.3874 lr=2.96e-04 tokens/s=2774294 | |
| step=1350 system rss_gb=5.16 cpu_load=34.85 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1400 loss=2.3759 lr=2.95e-04 tokens/s=2790041 | |
| step=1400 system rss_gb=5.16 cpu_load=32.34 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1450 loss=2.3756 lr=2.95e-04 tokens/s=2823657 | |
| step=1450 system rss_gb=5.17 cpu_load=30.14 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1500 loss=2.3674 lr=2.94e-04 tokens/s=2788584 | |
| step=1500 system rss_gb=5.17 cpu_load=29.21 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1550 loss=2.3577 lr=2.94e-04 tokens/s=2811584 | |
| step=1550 system rss_gb=5.17 cpu_load=32.35 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1600 loss=2.3556 lr=2.93e-04 tokens/s=2857641 | |
| step=1600 system rss_gb=5.17 cpu_load=30.52 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1650 loss=2.3556 lr=2.93e-04 tokens/s=2815070 | |
| step=1650 system rss_gb=5.17 cpu_load=28.80 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1700 loss=2.3457 lr=2.92e-04 tokens/s=2821209 | |
| step=1700 system rss_gb=5.17 cpu_load=28.41 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1750 loss=2.3316 lr=2.91e-04 tokens/s=2809745 | |
| step=1750 system rss_gb=5.17 cpu_load=26.22 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1800 loss=2.3304 lr=2.91e-04 tokens/s=2811976 | |
| step=1800 system rss_gb=5.17 cpu_load=25.44 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1850 loss=2.3228 lr=2.90e-04 tokens/s=2815147 | |
| step=1850 system rss_gb=5.17 cpu_load=23.74 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1900 loss=2.3255 lr=2.89e-04 tokens/s=2832661 | |
| step=1900 system rss_gb=5.17 cpu_load=21.88 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=1950 loss=2.3166 lr=2.88e-04 tokens/s=2782315 | |
| step=1950 system rss_gb=5.17 cpu_load=20.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=2000 loss=2.3142 lr=2.88e-04 tokens/s=2777604 | |
| step=2000 system rss_gb=5.17 cpu_load=19.97 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.04 vram_total_gb=47.38 | |
| step=2000 val_loss=2.5623 val_acc=0.1373 | |
| step=2000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=2000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.998 material=+5 2R1r1k1/1p2Q1pp/p4p2/1q6/5P2/1P6/P5PP/6RK b - - 0 1 | |
| step=2000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 r1bqkb1r/pp3ppp/2n2n2/2p1p3/3pP3/3P2P1/PPPNNPBP/R1BQ1RK1 b kq - 0 1 | |
| step=2000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=2000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=2000 board_turn_acc=0.6250 board_castle_acc=0.5250 board_piece_acc=0.1186 board_prior_baseline=0.5205 random_piece_acc=0.1268 span_piece_acc=0.1166 cos_to_move_child=0.9852 cos_to_corrupted=0.9983 test_r2_material=0.7380 pc1_piece_corr=-0.9273 | |
| saved best checkpoint to data/models/d256l3/best.pt (val_loss=2.5623) | |
| uploaded best_step2000 to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=2050 loss=2.3121 lr=2.87e-04 tokens/s=139552 | |
| step=2050 system rss_gb=6.27 cpu_load=50.75 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2100 loss=2.3129 lr=2.86e-04 tokens/s=2781735 | |
| step=2100 system rss_gb=6.27 cpu_load=56.71 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2150 loss=2.3095 lr=2.85e-04 tokens/s=2851245 | |
| step=2150 system rss_gb=6.27 cpu_load=65.45 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2200 loss=2.3029 lr=2.84e-04 tokens/s=2815285 | |
| step=2200 system rss_gb=6.27 cpu_load=69.76 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2250 loss=2.2986 lr=2.83e-04 tokens/s=2803894 | |
| step=2250 system rss_gb=6.28 cpu_load=73.11 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2300 loss=2.2933 lr=2.82e-04 tokens/s=2824530 | |
| step=2300 system rss_gb=6.27 cpu_load=74.62 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2350 loss=2.2924 lr=2.81e-04 tokens/s=2798000 | |
| step=2350 system rss_gb=6.28 cpu_load=76.35 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2400 loss=2.2816 lr=2.80e-04 tokens/s=2769572 | |
| step=2400 system rss_gb=6.28 cpu_load=77.57 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2450 loss=2.2827 lr=2.79e-04 tokens/s=2802308 | |
| step=2450 system rss_gb=6.28 cpu_load=79.24 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2500 loss=2.2784 lr=2.78e-04 tokens/s=2786277 | |
| step=2500 system rss_gb=6.27 cpu_load=77.39 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2550 loss=2.2720 lr=2.77e-04 tokens/s=2807481 | |
| step=2550 system rss_gb=6.28 cpu_load=70.94 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2600 loss=2.2658 lr=2.76e-04 tokens/s=2781920 | |
| step=2600 system rss_gb=6.28 cpu_load=63.76 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2650 loss=2.2591 lr=2.75e-04 tokens/s=2819682 | |
| step=2650 system rss_gb=6.28 cpu_load=56.44 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2700 loss=2.2594 lr=2.74e-04 tokens/s=2795621 | |
| step=2700 system rss_gb=6.28 cpu_load=52.88 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2750 loss=2.2578 lr=2.73e-04 tokens/s=2789458 | |
| step=2750 system rss_gb=6.28 cpu_load=53.34 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2800 loss=2.2530 lr=2.71e-04 tokens/s=2811999 | |
| step=2800 system rss_gb=6.28 cpu_load=53.38 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2850 loss=2.2449 lr=2.70e-04 tokens/s=2830816 | |
| step=2850 system rss_gb=6.28 cpu_load=48.66 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2900 loss=2.2416 lr=2.69e-04 tokens/s=2799466 | |
| step=2900 system rss_gb=6.28 cpu_load=44.75 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=2950 loss=2.2385 lr=2.68e-04 tokens/s=2797606 | |
| step=2950 system rss_gb=6.28 cpu_load=41.81 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.05 vram_total_gb=47.38 | |
| step=3000 loss=2.2374 lr=2.66e-04 tokens/s=2809590 | |
| step=3000 system rss_gb=6.28 cpu_load=37.77 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=3050 loss=2.2127 lr=2.65e-04 tokens/s=1040161 | |
| step=3050 system rss_gb=6.30 cpu_load=31.54 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3100 loss=2.2056 lr=2.64e-04 tokens/s=2833542 | |
| step=3100 system rss_gb=6.30 cpu_load=29.85 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3150 loss=2.1938 lr=2.62e-04 tokens/s=2765181 | |
| step=3150 system rss_gb=6.30 cpu_load=29.30 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3200 loss=2.1903 lr=2.61e-04 tokens/s=2785520 | |
| step=3200 system rss_gb=6.30 cpu_load=28.19 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3250 loss=2.1713 lr=2.60e-04 tokens/s=2790385 | |
| step=3250 system rss_gb=6.30 cpu_load=27.28 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3300 loss=2.1637 lr=2.58e-04 tokens/s=2796282 | |
| step=3300 system rss_gb=6.30 cpu_load=26.97 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3350 loss=2.1610 lr=2.57e-04 tokens/s=2810221 | |
| step=3350 system rss_gb=6.30 cpu_load=25.12 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3400 loss=2.1546 lr=2.55e-04 tokens/s=2800333 | |
| step=3400 system rss_gb=6.30 cpu_load=24.29 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3450 loss=2.1481 lr=2.54e-04 tokens/s=2812166 | |
| step=3450 system rss_gb=6.30 cpu_load=24.38 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3500 loss=2.1340 lr=2.52e-04 tokens/s=2812753 | |
| step=3500 system rss_gb=6.30 cpu_load=23.22 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3550 loss=2.1279 lr=2.51e-04 tokens/s=2819133 | |
| step=3550 system rss_gb=6.30 cpu_load=23.76 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3600 loss=2.1257 lr=2.49e-04 tokens/s=2739115 | |
| step=3600 system rss_gb=6.30 cpu_load=23.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3650 loss=2.1062 lr=2.48e-04 tokens/s=2779333 | |
| step=3650 system rss_gb=6.30 cpu_load=23.10 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3700 loss=2.0969 lr=2.46e-04 tokens/s=2815048 | |
| step=3700 system rss_gb=6.30 cpu_load=23.24 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3750 loss=2.0953 lr=2.45e-04 tokens/s=2802349 | |
| step=3750 system rss_gb=6.30 cpu_load=23.18 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3800 loss=2.0903 lr=2.43e-04 tokens/s=2762084 | |
| step=3800 system rss_gb=6.30 cpu_load=22.63 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3850 loss=2.0834 lr=2.41e-04 tokens/s=2819393 | |
| step=3850 system rss_gb=6.30 cpu_load=21.77 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=25.06 vram_total_gb=47.38 | |
| step=3900 loss=2.0679 lr=2.40e-04 tokens/s=2785977 | |
| step=3900 system rss_gb=6.30 cpu_load=22.01 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=3950 loss=2.0803 lr=2.38e-04 tokens/s=2823178 | |
| step=3950 system rss_gb=6.30 cpu_load=21.93 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4000 loss=2.0606 lr=2.37e-04 tokens/s=2808110 | |
| step=4000 system rss_gb=6.30 cpu_load=21.59 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4000 val_loss=2.3564 val_acc=0.1847 | |
| step=4000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=4000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.994 material=-1 8/1p6/pr3p2/4p1p1/3pP1P1/kP1P2K1/P6R/8 b - - 0 1 | |
| step=4000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 rnb1k1nr/pp2bpp1/2pp1q1p/4p3/2B1P3/2NP1N2/PPP2PPP/R1BQR1K1 b kq - 0 1 | |
| step=4000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=4000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=4000 board_turn_acc=0.8000 board_castle_acc=0.4375 board_piece_acc=0.1318 board_prior_baseline=0.5247 random_piece_acc=0.2313 span_piece_acc=0.1320 cos_to_move_child=0.9996 cos_to_corrupted=0.9632 test_r2_material=0.9122 pc1_piece_corr=-0.9786 | |
| saved best checkpoint to data/models/d256l3/best.pt (val_loss=2.3564) | |
| uploaded best_step4000 to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=4050 loss=2.0596 lr=2.35e-04 tokens/s=445907 | |
| step=4050 system rss_gb=6.36 cpu_load=29.51 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4100 loss=2.0496 lr=2.33e-04 tokens/s=2767540 | |
| step=4100 system rss_gb=6.36 cpu_load=28.26 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4150 loss=2.0520 lr=2.31e-04 tokens/s=2802591 | |
| step=4150 system rss_gb=6.36 cpu_load=26.55 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4200 loss=2.0449 lr=2.30e-04 tokens/s=2724553 | |
| step=4200 system rss_gb=6.36 cpu_load=38.29 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4250 loss=2.0362 lr=2.28e-04 tokens/s=2733866 | |
| step=4250 system rss_gb=6.36 cpu_load=50.01 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4300 loss=2.0428 lr=2.26e-04 tokens/s=2782074 | |
| step=4300 system rss_gb=6.36 cpu_load=56.70 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4350 loss=2.0307 lr=2.24e-04 tokens/s=2717501 | |
| step=4350 system rss_gb=6.36 cpu_load=61.21 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4400 loss=2.0276 lr=2.23e-04 tokens/s=2740606 | |
| step=4400 system rss_gb=6.36 cpu_load=64.59 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4450 loss=2.0358 lr=2.21e-04 tokens/s=2789940 | |
| step=4450 system rss_gb=6.37 cpu_load=68.23 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4500 loss=2.0272 lr=2.19e-04 tokens/s=2723581 | |
| step=4500 system rss_gb=6.36 cpu_load=70.79 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4550 loss=2.0272 lr=2.17e-04 tokens/s=2757469 | |
| step=4550 system rss_gb=6.36 cpu_load=78.96 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4600 loss=2.0298 lr=2.15e-04 tokens/s=2713869 | |
| step=4600 system rss_gb=6.36 cpu_load=81.10 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4650 loss=2.0160 lr=2.13e-04 tokens/s=2768606 | |
| step=4650 system rss_gb=6.36 cpu_load=83.00 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4700 loss=2.0231 lr=2.12e-04 tokens/s=2799634 | |
| step=4700 system rss_gb=6.36 cpu_load=79.97 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4750 loss=2.0218 lr=2.10e-04 tokens/s=2765622 | |
| step=4750 system rss_gb=6.36 cpu_load=89.87 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4800 loss=2.0193 lr=2.08e-04 tokens/s=2800673 | |
| step=4800 system rss_gb=6.36 cpu_load=86.92 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4850 loss=2.0131 lr=2.06e-04 tokens/s=2799704 | |
| step=4850 system rss_gb=6.36 cpu_load=82.04 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4900 loss=2.0112 lr=2.04e-04 tokens/s=2771434 | |
| step=4900 system rss_gb=6.36 cpu_load=73.75 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=4950 loss=2.0126 lr=2.02e-04 tokens/s=2767907 | |
| step=4950 system rss_gb=6.36 cpu_load=66.69 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5000 loss=2.0155 lr=2.00e-04 tokens/s=2776120 | |
| step=5000 system rss_gb=6.36 cpu_load=59.34 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=5050 loss=2.0046 lr=1.98e-04 tokens/s=1079364 | |
| step=5050 system rss_gb=6.39 cpu_load=49.09 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5100 loss=2.0064 lr=1.96e-04 tokens/s=2790145 | |
| step=5100 system rss_gb=6.39 cpu_load=45.17 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5150 loss=2.0230 lr=1.94e-04 tokens/s=2790691 | |
| step=5150 system rss_gb=6.39 cpu_load=42.51 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5200 loss=2.0129 lr=1.92e-04 tokens/s=2843059 | |
| step=5200 system rss_gb=6.39 cpu_load=39.37 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5250 loss=2.0077 lr=1.90e-04 tokens/s=2806088 | |
| step=5250 system rss_gb=6.39 cpu_load=37.81 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5300 loss=1.9980 lr=1.88e-04 tokens/s=2780105 | |
| step=5300 system rss_gb=6.39 cpu_load=34.89 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5350 loss=2.0093 lr=1.87e-04 tokens/s=2791375 | |
| step=5350 system rss_gb=6.39 cpu_load=31.83 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5400 loss=2.0106 lr=1.85e-04 tokens/s=2790124 | |
| step=5400 system rss_gb=6.39 cpu_load=29.20 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5450 loss=2.0026 lr=1.83e-04 tokens/s=2798661 | |
| step=5450 system rss_gb=6.38 cpu_load=27.15 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5500 loss=2.0025 lr=1.81e-04 tokens/s=2807870 | |
| step=5500 system rss_gb=6.39 cpu_load=26.86 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5550 loss=2.0000 lr=1.79e-04 tokens/s=2788079 | |
| step=5550 system rss_gb=6.39 cpu_load=25.57 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5600 loss=2.0001 lr=1.76e-04 tokens/s=2779807 | |
| step=5600 system rss_gb=6.39 cpu_load=25.09 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5650 loss=2.0083 lr=1.74e-04 tokens/s=2777343 | |
| step=5650 system rss_gb=6.39 cpu_load=29.73 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5700 loss=1.9951 lr=1.72e-04 tokens/s=2779331 | |
| step=5700 system rss_gb=6.39 cpu_load=28.17 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5750 loss=2.0035 lr=1.70e-04 tokens/s=2813488 | |
| step=5750 system rss_gb=6.39 cpu_load=27.57 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5800 loss=1.9967 lr=1.68e-04 tokens/s=2764147 | |
| step=5800 system rss_gb=6.39 cpu_load=25.91 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5850 loss=1.9868 lr=1.66e-04 tokens/s=2808735 | |
| step=5850 system rss_gb=6.39 cpu_load=23.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5900 loss=1.9880 lr=1.64e-04 tokens/s=2780000 | |
| step=5900 system rss_gb=6.38 cpu_load=23.20 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=5950 loss=1.9929 lr=1.62e-04 tokens/s=2801967 | |
| step=5950 system rss_gb=6.39 cpu_load=22.85 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6000 loss=2.0002 lr=1.60e-04 tokens/s=2827240 | |
| step=6000 system rss_gb=6.39 cpu_load=21.11 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6000 val_loss=2.2905 val_acc=0.1962 | |
| step=6000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=6000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.976 material=+1 3R4/1p1R2p1/p4k1p/7P/1q6/8/P4PP1/6K1 w - - 0 1 | |
| step=6000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 rnb1k1nr/pp2bpp1/2pp1q1p/4p3/2B1P3/2NP1N2/PPP2PPP/R1BQR1K1 b kq - 0 1 | |
| step=6000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=6000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=6000 board_turn_acc=0.8875 board_castle_acc=0.4000 board_piece_acc=0.1354 board_prior_baseline=0.5255 random_piece_acc=0.2450 span_piece_acc=0.1398 cos_to_move_child=0.9990 cos_to_corrupted=0.9173 test_r2_material=0.9725 pc1_piece_corr=-0.9650 | |
| saved best checkpoint to data/models/d256l3/best.pt (val_loss=2.2905) | |
| uploaded best_step6000 to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=6050 loss=1.9987 lr=1.58e-04 tokens/s=409623 | |
| step=6050 system rss_gb=6.44 cpu_load=22.72 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6100 loss=2.0004 lr=1.56e-04 tokens/s=2820406 | |
| step=6100 system rss_gb=6.44 cpu_load=22.75 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6150 loss=1.9936 lr=1.54e-04 tokens/s=2788725 | |
| step=6150 system rss_gb=6.44 cpu_load=21.43 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6200 loss=1.9991 lr=1.52e-04 tokens/s=2788957 | |
| step=6200 system rss_gb=6.44 cpu_load=20.94 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6250 loss=1.9830 lr=1.50e-04 tokens/s=2792740 | |
| step=6250 system rss_gb=6.44 cpu_load=27.15 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6300 loss=1.9923 lr=1.48e-04 tokens/s=2758938 | |
| step=6300 system rss_gb=6.44 cpu_load=25.50 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6350 loss=1.9979 lr=1.46e-04 tokens/s=2751453 | |
| step=6350 system rss_gb=6.44 cpu_load=27.88 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6400 loss=1.9955 lr=1.44e-04 tokens/s=2751491 | |
| step=6400 system rss_gb=6.44 cpu_load=27.17 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6450 loss=1.9944 lr=1.42e-04 tokens/s=2754569 | |
| step=6450 system rss_gb=6.44 cpu_load=26.28 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6500 loss=1.9966 lr=1.40e-04 tokens/s=2753087 | |
| step=6500 system rss_gb=6.44 cpu_load=30.10 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6550 loss=1.9871 lr=1.38e-04 tokens/s=2768377 | |
| step=6550 system rss_gb=6.44 cpu_load=39.00 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6600 loss=2.0036 lr=1.36e-04 tokens/s=2782823 | |
| step=6600 system rss_gb=6.44 cpu_load=45.95 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6650 loss=1.9797 lr=1.34e-04 tokens/s=2787992 | |
| step=6650 system rss_gb=6.44 cpu_load=50.65 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6700 loss=1.9883 lr=1.32e-04 tokens/s=2755065 | |
| step=6700 system rss_gb=6.44 cpu_load=57.08 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6750 loss=1.9885 lr=1.30e-04 tokens/s=2724282 | |
| step=6750 system rss_gb=6.44 cpu_load=61.62 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6800 loss=1.9899 lr=1.28e-04 tokens/s=2783677 | |
| step=6800 system rss_gb=6.44 cpu_load=66.13 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6850 loss=1.9881 lr=1.26e-04 tokens/s=2782364 | |
| step=6850 system rss_gb=6.44 cpu_load=78.55 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6900 loss=1.9863 lr=1.24e-04 tokens/s=2744399 | |
| step=6900 system rss_gb=6.44 cpu_load=80.60 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=6950 loss=1.9965 lr=1.21e-04 tokens/s=2790837 | |
| step=6950 system rss_gb=6.44 cpu_load=79.83 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7000 loss=1.9863 lr=1.19e-04 tokens/s=2754754 | |
| step=7000 system rss_gb=6.44 cpu_load=86.56 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=7050 loss=1.9780 lr=1.17e-04 tokens/s=1062021 | |
| step=7050 system rss_gb=6.47 cpu_load=89.50 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7100 loss=1.9782 lr=1.15e-04 tokens/s=2740715 | |
| step=7100 system rss_gb=6.47 cpu_load=81.11 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7150 loss=1.9832 lr=1.13e-04 tokens/s=2765999 | |
| step=7150 system rss_gb=6.47 cpu_load=77.54 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7200 loss=1.9897 lr=1.12e-04 tokens/s=2774489 | |
| step=7200 system rss_gb=6.47 cpu_load=68.52 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7250 loss=1.9855 lr=1.10e-04 tokens/s=2778308 | |
| step=7250 system rss_gb=6.47 cpu_load=60.51 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7300 loss=1.9839 lr=1.08e-04 tokens/s=2805629 | |
| step=7300 system rss_gb=6.47 cpu_load=53.36 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7350 loss=1.9858 lr=1.06e-04 tokens/s=2798578 | |
| step=7350 system rss_gb=6.47 cpu_load=48.60 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7400 loss=1.9819 lr=1.04e-04 tokens/s=2809349 | |
| step=7400 system rss_gb=6.47 cpu_load=46.63 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7450 loss=1.9778 lr=1.02e-04 tokens/s=2831491 | |
| step=7450 system rss_gb=6.47 cpu_load=42.11 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7500 loss=1.9772 lr=9.98e-05 tokens/s=2776193 | |
| step=7500 system rss_gb=6.47 cpu_load=39.40 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7550 loss=1.9830 lr=9.78e-05 tokens/s=2790666 | |
| step=7550 system rss_gb=6.47 cpu_load=36.61 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7600 loss=1.9869 lr=9.59e-05 tokens/s=2767218 | |
| step=7600 system rss_gb=6.47 cpu_load=34.05 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7650 loss=1.9861 lr=9.40e-05 tokens/s=2852556 | |
| step=7650 system rss_gb=6.47 cpu_load=31.48 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7700 loss=1.9780 lr=9.21e-05 tokens/s=2805505 | |
| step=7700 system rss_gb=6.47 cpu_load=29.78 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7750 loss=1.9815 lr=9.02e-05 tokens/s=2796760 | |
| step=7750 system rss_gb=6.47 cpu_load=28.56 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7800 loss=1.9804 lr=8.84e-05 tokens/s=2768154 | |
| step=7800 system rss_gb=6.47 cpu_load=28.12 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7850 loss=1.9768 lr=8.65e-05 tokens/s=2833047 | |
| step=7850 system rss_gb=6.47 cpu_load=25.35 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7900 loss=1.9714 lr=8.47e-05 tokens/s=2770508 | |
| step=7900 system rss_gb=6.47 cpu_load=24.83 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=7950 loss=1.9801 lr=8.28e-05 tokens/s=2787084 | |
| step=7950 system rss_gb=6.47 cpu_load=25.00 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8000 loss=1.9812 lr=8.10e-05 tokens/s=2785586 | |
| step=8000 system rss_gb=6.47 cpu_load=25.46 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8000 val_loss=2.2952 val_acc=0.1945 | |
| step=8000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=8000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.976 material=+0 8/5pkp/1ppR2p1/2n5/7P/Pp3NP1/1P2qP2/3R2K1 w - - 0 1 | |
| step=8000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 r1b1k1nr/pppp1p2/1bn2q1p/1B2p1p1/PP2P3/2P2N1P/3P1PP1/RNBQ1RK1 b kq - 0 1 | |
| step=8000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=8000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=8000 board_turn_acc=0.8500 board_castle_acc=0.4250 board_piece_acc=0.1381 board_prior_baseline=0.5266 random_piece_acc=0.2338 span_piece_acc=0.1392 cos_to_move_child=0.9991 cos_to_corrupted=0.9170 test_r2_material=0.9798 pc1_piece_corr=-0.9727 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=8050 loss=1.9784 lr=7.92e-05 tokens/s=573299 | |
| step=8050 system rss_gb=6.49 cpu_load=24.06 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8100 loss=1.9814 lr=7.74e-05 tokens/s=2813022 | |
| step=8100 system rss_gb=6.49 cpu_load=22.72 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8150 loss=1.9886 lr=7.56e-05 tokens/s=2741815 | |
| step=8150 system rss_gb=6.48 cpu_load=22.60 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8200 loss=1.9951 lr=7.38e-05 tokens/s=2793647 | |
| step=8200 system rss_gb=6.49 cpu_load=22.98 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8250 loss=1.9776 lr=7.21e-05 tokens/s=2759457 | |
| step=8250 system rss_gb=6.49 cpu_load=22.31 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8300 loss=1.9727 lr=7.03e-05 tokens/s=2760905 | |
| step=8300 system rss_gb=6.49 cpu_load=21.12 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8350 loss=1.9801 lr=6.86e-05 tokens/s=2736108 | |
| step=8350 system rss_gb=6.49 cpu_load=21.91 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8400 loss=1.9737 lr=6.69e-05 tokens/s=2785955 | |
| step=8400 system rss_gb=6.49 cpu_load=21.01 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8450 loss=1.9746 lr=6.52e-05 tokens/s=2740848 | |
| step=8450 system rss_gb=6.49 cpu_load=20.35 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8500 loss=1.9701 lr=6.35e-05 tokens/s=2751847 | |
| step=8500 system rss_gb=6.49 cpu_load=21.23 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8550 loss=1.9815 lr=6.18e-05 tokens/s=2737871 | |
| step=8550 system rss_gb=6.49 cpu_load=20.41 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8600 loss=1.9863 lr=6.02e-05 tokens/s=2747940 | |
| step=8600 system rss_gb=6.49 cpu_load=20.08 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8650 loss=1.9759 lr=5.86e-05 tokens/s=2759595 | |
| step=8650 system rss_gb=6.49 cpu_load=20.26 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8700 loss=1.9737 lr=5.69e-05 tokens/s=2812281 | |
| step=8700 system rss_gb=6.48 cpu_load=20.46 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8750 loss=1.9825 lr=5.53e-05 tokens/s=2724520 | |
| step=8750 system rss_gb=6.48 cpu_load=21.47 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8800 loss=1.9763 lr=5.38e-05 tokens/s=2831159 | |
| step=8800 system rss_gb=6.48 cpu_load=20.86 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8850 loss=1.9788 lr=5.22e-05 tokens/s=2770490 | |
| step=8850 system rss_gb=6.49 cpu_load=20.31 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8900 loss=1.9771 lr=5.06e-05 tokens/s=2754312 | |
| step=8900 system rss_gb=6.48 cpu_load=30.49 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=8950 loss=1.9884 lr=4.91e-05 tokens/s=2791014 | |
| step=8950 system rss_gb=6.48 cpu_load=38.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9000 loss=1.9779 lr=4.76e-05 tokens/s=2723232 | |
| step=9000 system rss_gb=6.48 cpu_load=45.89 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=9050 loss=1.9728 lr=4.61e-05 tokens/s=1075430 | |
| step=9050 system rss_gb=6.50 cpu_load=55.13 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9100 loss=1.9762 lr=4.47e-05 tokens/s=2781626 | |
| step=9100 system rss_gb=6.50 cpu_load=60.03 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9150 loss=1.9751 lr=4.32e-05 tokens/s=2725910 | |
| step=9150 system rss_gb=6.51 cpu_load=68.25 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9200 loss=1.9755 lr=4.18e-05 tokens/s=2732613 | |
| step=9200 system rss_gb=6.50 cpu_load=72.31 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9250 loss=1.9650 lr=4.04e-05 tokens/s=2731307 | |
| step=9250 system rss_gb=6.50 cpu_load=68.80 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9300 loss=1.9775 lr=3.90e-05 tokens/s=2779174 | |
| step=9300 system rss_gb=6.51 cpu_load=71.84 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9350 loss=1.9705 lr=3.76e-05 tokens/s=2746328 | |
| step=9350 system rss_gb=6.51 cpu_load=78.50 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9400 loss=1.9820 lr=3.63e-05 tokens/s=2772620 | |
| step=9400 system rss_gb=6.51 cpu_load=76.51 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9450 loss=1.9834 lr=3.49e-05 tokens/s=2798110 | |
| step=9450 system rss_gb=6.51 cpu_load=74.33 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9500 loss=1.9733 lr=3.36e-05 tokens/s=2757321 | |
| step=9500 system rss_gb=6.51 cpu_load=73.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9550 loss=1.9789 lr=3.24e-05 tokens/s=1333435 | |
| step=9550 system rss_gb=6.51 cpu_load=63.48 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9600 loss=1.9825 lr=3.11e-05 tokens/s=2809913 | |
| step=9600 system rss_gb=6.51 cpu_load=56.59 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9650 loss=1.9665 lr=2.99e-05 tokens/s=2764170 | |
| step=9650 system rss_gb=6.51 cpu_load=50.94 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9700 loss=1.9752 lr=2.86e-05 tokens/s=2755415 | |
| step=9700 system rss_gb=6.51 cpu_load=46.21 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9750 loss=1.9741 lr=2.75e-05 tokens/s=2757103 | |
| step=9750 system rss_gb=6.51 cpu_load=42.79 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9800 loss=1.9789 lr=2.63e-05 tokens/s=2750079 | |
| step=9800 system rss_gb=6.51 cpu_load=40.26 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9850 loss=1.9722 lr=2.51e-05 tokens/s=2813584 | |
| step=9850 system rss_gb=6.51 cpu_load=39.28 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9900 loss=1.9593 lr=2.40e-05 tokens/s=2725660 | |
| step=9900 system rss_gb=6.51 cpu_load=36.62 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=9950 loss=1.9744 lr=2.29e-05 tokens/s=2718414 | |
| step=9950 system rss_gb=6.51 cpu_load=34.91 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10000 loss=1.9730 lr=2.18e-05 tokens/s=2755834 | |
| step=10000 system rss_gb=6.51 cpu_load=34.22 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10000 val_loss=2.2788 val_acc=0.1973 | |
| step=10000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=10000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.978 material=+0 8/5pkp/1ppR2p1/2n5/7P/Pp3NP1/1P2qP2/3R2K1 w - - 0 1 | |
| step=10000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 r1b1kn1r/ppq1bppp/2pp1n2/4p3/2PPP3/2N1BN1P/PP2BPP1/R2Q1RK1 b kq - 0 1 | |
| step=10000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=10000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=10000 board_turn_acc=0.8375 board_castle_acc=0.4250 board_piece_acc=0.1395 board_prior_baseline=0.5271 random_piece_acc=0.2425 span_piece_acc=0.1412 cos_to_move_child=0.9991 cos_to_corrupted=0.9194 test_r2_material=0.9768 pc1_piece_corr=0.9871 | |
| saved best checkpoint to data/models/d256l3/best.pt (val_loss=2.2788) | |
| uploaded best_step10000 to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=10050 loss=1.9808 lr=2.08e-05 tokens/s=350868 | |
| step=10050 system rss_gb=6.51 cpu_load=27.61 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10100 loss=1.9736 lr=1.98e-05 tokens/s=2822846 | |
| step=10100 system rss_gb=6.51 cpu_load=26.96 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10150 loss=1.9748 lr=1.88e-05 tokens/s=2779096 | |
| step=10150 system rss_gb=6.51 cpu_load=26.40 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10200 loss=1.9847 lr=1.78e-05 tokens/s=2747719 | |
| step=10200 system rss_gb=6.51 cpu_load=25.08 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10250 loss=1.9640 lr=1.68e-05 tokens/s=2764721 | |
| step=10250 system rss_gb=6.51 cpu_load=24.63 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10300 loss=1.9768 lr=1.59e-05 tokens/s=2773478 | |
| step=10300 system rss_gb=6.51 cpu_load=23.26 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10350 loss=1.9748 lr=1.50e-05 tokens/s=2750870 | |
| step=10350 system rss_gb=6.51 cpu_load=22.04 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10400 loss=1.9744 lr=1.41e-05 tokens/s=2754199 | |
| step=10400 system rss_gb=6.51 cpu_load=21.63 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10450 loss=1.9722 lr=1.32e-05 tokens/s=2762474 | |
| step=10450 system rss_gb=6.51 cpu_load=22.47 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10500 loss=1.9734 lr=1.24e-05 tokens/s=2749014 | |
| step=10500 system rss_gb=6.51 cpu_load=26.12 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=30.05 vram_total_gb=47.38 | |
| step=10550 loss=1.9736 lr=1.16e-05 tokens/s=2775823 | |
| step=10550 system rss_gb=6.51 cpu_load=25.30 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10600 loss=1.9735 lr=1.08e-05 tokens/s=2729309 | |
| step=10600 system rss_gb=6.51 cpu_load=24.62 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10650 loss=1.9763 lr=1.01e-05 tokens/s=2781653 | |
| step=10650 system rss_gb=6.51 cpu_load=24.68 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10700 loss=1.9701 lr=9.36e-06 tokens/s=2792357 | |
| step=10700 system rss_gb=6.51 cpu_load=29.51 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10750 loss=1.9746 lr=8.66e-06 tokens/s=2740412 | |
| step=10750 system rss_gb=6.51 cpu_load=27.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10800 loss=1.9695 lr=7.99e-06 tokens/s=2752553 | |
| step=10800 system rss_gb=6.51 cpu_load=26.91 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10850 loss=1.9641 lr=7.34e-06 tokens/s=2753008 | |
| step=10850 system rss_gb=6.52 cpu_load=26.76 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10900 loss=1.9771 lr=6.72e-06 tokens/s=2782558 | |
| step=10900 system rss_gb=6.52 cpu_load=26.48 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=10950 loss=1.9785 lr=6.13e-06 tokens/s=2778475 | |
| step=10950 system rss_gb=6.52 cpu_load=25.94 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11000 loss=1.9833 lr=5.56e-06 tokens/s=2796389 | |
| step=11000 system rss_gb=6.52 cpu_load=26.34 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
| uploaded last to patrickfrank1/chesspos-checkpoints/run_d256l3_streaming | |
| step=11050 loss=1.9675 lr=5.02e-06 tokens/s=1094544 | |
| step=11050 system rss_gb=6.52 cpu_load=51.45 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11100 loss=1.9663 lr=4.51e-06 tokens/s=2767338 | |
| step=11100 system rss_gb=6.52 cpu_load=57.83 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11150 loss=1.9726 lr=4.03e-06 tokens/s=2779585 | |
| step=11150 system rss_gb=6.52 cpu_load=66.78 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11200 loss=1.9845 lr=3.57e-06 tokens/s=2795365 | |
| step=11200 system rss_gb=6.52 cpu_load=69.97 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11250 loss=1.9604 lr=3.14e-06 tokens/s=2749169 | |
| step=11250 system rss_gb=6.52 cpu_load=71.73 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11300 loss=1.9732 lr=2.73e-06 tokens/s=2833275 | |
| step=11300 system rss_gb=6.52 cpu_load=74.77 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11350 loss=1.9732 lr=2.36e-06 tokens/s=2758073 | |
| step=11350 system rss_gb=6.52 cpu_load=73.66 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11400 loss=1.9771 lr=2.01e-06 tokens/s=2766614 | |
| step=11400 system rss_gb=6.52 cpu_load=75.86 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11450 loss=1.9706 lr=1.69e-06 tokens/s=2776053 | |
| step=11450 system rss_gb=6.52 cpu_load=77.90 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11500 loss=1.9750 lr=1.40e-06 tokens/s=2766858 | |
| step=11500 system rss_gb=6.52 cpu_load=84.55 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11550 loss=1.9858 lr=1.13e-06 tokens/s=2787122 | |
| step=11550 system rss_gb=6.52 cpu_load=83.33 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11600 loss=1.9721 lr=8.95e-07 tokens/s=2768066 | |
| step=11600 system rss_gb=6.52 cpu_load=75.92 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11650 loss=1.9660 lr=6.85e-07 tokens/s=2750159 | |
| step=11650 system rss_gb=6.52 cpu_load=69.99 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11700 loss=1.9701 lr=5.03e-07 tokens/s=2781556 | |
| step=11700 system rss_gb=6.52 cpu_load=66.46 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11750 loss=1.9783 lr=3.50e-07 tokens/s=2769448 | |
| step=11750 system rss_gb=6.52 cpu_load=59.55 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11800 loss=1.9749 lr=2.24e-07 tokens/s=2776357 | |
| step=11800 system rss_gb=6.51 cpu_load=53.87 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11850 loss=1.9699 lr=1.26e-07 tokens/s=2800255 | |
| step=11850 system rss_gb=6.52 cpu_load=48.36 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11900 loss=1.9705 lr=5.60e-08 tokens/s=2769766 | |
| step=11900 system rss_gb=6.51 cpu_load=43.78 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=11950 loss=1.9689 lr=1.40e-08 tokens/s=2743501 | |
| step=11950 system rss_gb=6.51 cpu_load=40.21 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=12000 loss=1.9698 lr=0.00e+00 tokens/s=2772602 | |
| step=12000 system rss_gb=6.51 cpu_load=38.89 n_cpus=128.00 vram_alloc_gb=0.56 vram_used_gb=35.05 vram_total_gb=47.38 | |
| step=12000 val_loss=2.2853 val_acc=0.1969 | |
| step=12000 neighbor query material=+3 r2q3k/ppp3pp/1Pnp4/4pr1n/1PB3b1/2PPBN2/3N1PPP/R2Q1RK1 b - - 0 1 | |
| top sim=1.000 material=+3 r2qnrk1/1p1Nb2p/p1n1ppp1/3p4/3P1P2/2PB1N1R/PP4PP/R1BQ2K1 b - - 0 1 | |
| step=12000 neighbor query material=-6 6r1/2pk4/1p1p4/5p1p/PP4q1/2P2pPp/5P1K/4R1R1 b - - 0 1 | |
| top sim=0.980 material=+0 8/5pkp/1ppR2p1/2n5/7P/Pp3NP1/1P2qP2/3R2K1 w - - 0 1 | |
| step=12000 neighbor query material=+0 r1bqk2r/ppppbppp/2n2n2/4p3/2BPP3/5N2/PPP2PPP/RNBQ1RK1 b kq - 0 1 | |
| top sim=1.000 material=+0 r1b1kn1r/ppq1bppp/2pp1n2/4p3/2PPP3/2N1BN1P/PP2BPP1/R2Q1RK1 b kq - 0 1 | |
| step=12000 neighbor board report saved to data/models/d256l3/neighbor_examples.html | |
| step=12000 eval plot saved to data/models/d256l3/pca_phase.png | |
| step=12000 board_turn_acc=0.8500 board_castle_acc=0.4375 board_piece_acc=0.1391 board_prior_baseline=0.5267 random_piece_acc=0.2363 span_piece_acc=0.1401 cos_to_move_child=0.9991 cos_to_corrupted=0.9240 test_r2_material=0.9773 pc1_piece_corr=-0.9860 | |
| saved full checkpoint to data/models/d256l3/last.pt | |
Xet Storage Details
- Size:
- 66.7 kB
- Xet hash:
- fba56004bfc8f2f30532cb68cb0b2936c5c0ed993019e6e794edd64e11e61cd8
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.