Buckets:
| device=cuda params=3.43M bf16=True compile=False | |
| train games=368242 val games=19273 | |
| step=50 loss=6.6869 lr=3.00e-05 tokens/s=605344 | |
| step=100 loss=6.0171 lr=6.00e-05 tokens/s=658209 | |
| step=150 loss=5.3918 lr=9.00e-05 tokens/s=655713 | |
| step=200 loss=5.1521 lr=1.20e-04 tokens/s=656484 | |
| step=250 loss=5.1038 lr=1.50e-04 tokens/s=655852 | |
| step=300 loss=4.9399 lr=1.80e-04 tokens/s=657504 | |
| step=350 loss=4.7401 lr=2.10e-04 tokens/s=658316 | |
| step=400 loss=4.6398 lr=2.40e-04 tokens/s=652765 | |
| step=450 loss=4.5840 lr=2.70e-04 tokens/s=654749 | |
| step=500 loss=4.4888 lr=3.00e-04 tokens/s=655142 | |
| step=550 loss=4.4290 lr=3.00e-04 tokens/s=649054 | |
| step=600 loss=4.3759 lr=3.00e-04 tokens/s=653402 | |
| step=650 loss=4.3510 lr=3.00e-04 tokens/s=655609 | |
| step=700 loss=4.2784 lr=3.00e-04 tokens/s=652756 | |
| step=750 loss=4.2578 lr=3.00e-04 tokens/s=648627 | |
| step=800 loss=4.2221 lr=2.99e-04 tokens/s=652690 | |
| step=850 loss=4.1846 lr=2.99e-04 tokens/s=645393 | |
| step=900 loss=4.1585 lr=2.99e-04 tokens/s=652467 | |
| step=950 loss=4.1400 lr=2.99e-04 tokens/s=646742 | |
| step=1000 loss=4.1307 lr=2.99e-04 tokens/s=657810 | |
| step=1000 val_loss=4.1157 val_acc=0.1248 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=4.1157) | |
| step=1050 loss=4.1286 lr=2.98e-04 tokens/s=507683 | |
| step=1100 loss=4.1158 lr=2.98e-04 tokens/s=646758 | |
| step=1150 loss=4.0930 lr=2.98e-04 tokens/s=649028 | |
| step=1200 loss=4.0844 lr=2.97e-04 tokens/s=650025 | |
| step=1250 loss=4.0605 lr=2.97e-04 tokens/s=647861 | |
| step=1300 loss=4.0477 lr=2.96e-04 tokens/s=659500 | |
| step=1350 loss=4.0522 lr=2.96e-04 tokens/s=651295 | |
| step=1400 loss=4.0189 lr=2.95e-04 tokens/s=647677 | |
| step=1450 loss=4.0096 lr=2.95e-04 tokens/s=647136 | |
| step=1500 loss=4.0123 lr=2.94e-04 tokens/s=652339 | |
| step=1550 loss=3.9918 lr=2.94e-04 tokens/s=650727 | |
| step=1600 loss=3.9925 lr=2.93e-04 tokens/s=649967 | |
| step=1650 loss=3.9574 lr=2.93e-04 tokens/s=649149 | |
| step=1700 loss=3.9644 lr=2.92e-04 tokens/s=654601 | |
| step=1750 loss=3.9357 lr=2.91e-04 tokens/s=645359 | |
| step=1800 loss=3.9396 lr=2.91e-04 tokens/s=648631 | |
| step=1850 loss=3.9387 lr=2.90e-04 tokens/s=649498 | |
| step=1900 loss=3.9088 lr=2.89e-04 tokens/s=652218 | |
| step=1950 loss=3.9099 lr=2.88e-04 tokens/s=646045 | |
| step=2000 loss=3.9072 lr=2.88e-04 tokens/s=650717 | |
| step=2000 val_loss=3.8907 val_acc=0.1290 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.8907) | |
| saved full checkpoint to models/run_full/last.pt | |
| step=2050 loss=3.8960 lr=2.87e-04 tokens/s=568269 | |
| step=2100 loss=3.8937 lr=2.86e-04 tokens/s=646252 | |
| step=2150 loss=3.8785 lr=2.85e-04 tokens/s=644773 | |
| step=2200 loss=3.8870 lr=2.84e-04 tokens/s=647763 | |
| step=2250 loss=3.8693 lr=2.83e-04 tokens/s=652554 | |
| step=2300 loss=3.8681 lr=2.82e-04 tokens/s=645917 | |
| step=2350 loss=3.8521 lr=2.81e-04 tokens/s=650041 | |
| step=2400 loss=3.8553 lr=2.80e-04 tokens/s=645862 | |
| step=2450 loss=3.8349 lr=2.79e-04 tokens/s=646051 | |
| step=2500 loss=3.8432 lr=2.78e-04 tokens/s=648691 | |
| step=2550 loss=3.8205 lr=2.77e-04 tokens/s=645036 | |
| step=2600 loss=3.8337 lr=2.76e-04 tokens/s=649681 | |
| step=2650 loss=3.8118 lr=2.75e-04 tokens/s=653748 | |
| step=2700 loss=3.8101 lr=2.74e-04 tokens/s=649435 | |
| step=2750 loss=3.8054 lr=2.73e-04 tokens/s=653789 | |
| step=2800 loss=3.8101 lr=2.71e-04 tokens/s=646519 | |
| step=2850 loss=3.7902 lr=2.70e-04 tokens/s=647344 | |
| step=2900 loss=3.7822 lr=2.69e-04 tokens/s=648579 | |
| step=2950 loss=3.7789 lr=2.68e-04 tokens/s=644780 | |
| step=3000 loss=3.7880 lr=2.66e-04 tokens/s=644366 | |
| step=3000 val_loss=3.7698 val_acc=0.1300 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.7698) | |
| step=3050 loss=3.7673 lr=2.65e-04 tokens/s=564595 | |
| step=3100 loss=3.7654 lr=2.64e-04 tokens/s=647492 | |
| step=3150 loss=3.7553 lr=2.62e-04 tokens/s=646671 | |
| step=3200 loss=3.7439 lr=2.61e-04 tokens/s=652591 | |
| step=3250 loss=3.7528 lr=2.60e-04 tokens/s=648956 | |
| step=3300 loss=3.7429 lr=2.58e-04 tokens/s=649383 | |
| step=3350 loss=3.7520 lr=2.57e-04 tokens/s=644017 | |
| step=3400 loss=3.7357 lr=2.55e-04 tokens/s=652538 | |
| step=3450 loss=3.7503 lr=2.54e-04 tokens/s=649292 | |
| step=3500 loss=3.7253 lr=2.52e-04 tokens/s=644803 | |
| step=3550 loss=3.7216 lr=2.51e-04 tokens/s=652838 | |
| step=3600 loss=3.7127 lr=2.49e-04 tokens/s=650294 | |
| step=3650 loss=3.7256 lr=2.48e-04 tokens/s=655413 | |
| step=3700 loss=3.7148 lr=2.46e-04 tokens/s=653472 | |
| step=3750 loss=3.6982 lr=2.45e-04 tokens/s=657740 | |
| step=3800 loss=3.7105 lr=2.43e-04 tokens/s=653779 | |
| step=3850 loss=3.7004 lr=2.41e-04 tokens/s=649294 | |
| step=3900 loss=3.7193 lr=2.40e-04 tokens/s=650983 | |
| step=3950 loss=3.6842 lr=2.38e-04 tokens/s=648987 | |
| step=4000 loss=3.6908 lr=2.37e-04 tokens/s=645780 | |
| step=4000 val_loss=3.6678 val_acc=0.1331 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.6678) | |
| saved full checkpoint to models/run_full/last.pt | |
| step=4050 loss=3.6903 lr=2.35e-04 tokens/s=562327 | |
| step=4100 loss=3.6793 lr=2.33e-04 tokens/s=650480 | |
| step=4150 loss=3.6674 lr=2.31e-04 tokens/s=648029 | |
| step=4200 loss=3.6734 lr=2.30e-04 tokens/s=645908 | |
| step=4250 loss=3.6681 lr=2.28e-04 tokens/s=649744 | |
| step=4300 loss=3.6486 lr=2.26e-04 tokens/s=649635 | |
| step=4350 loss=3.6558 lr=2.24e-04 tokens/s=648360 | |
| step=4400 loss=3.6411 lr=2.23e-04 tokens/s=652156 | |
| step=4450 loss=3.6508 lr=2.21e-04 tokens/s=647129 | |
| step=4500 loss=3.6321 lr=2.19e-04 tokens/s=655967 | |
| step=4550 loss=3.6250 lr=2.17e-04 tokens/s=650098 | |
| step=4600 loss=3.6183 lr=2.15e-04 tokens/s=644691 | |
| step=4650 loss=3.6175 lr=2.13e-04 tokens/s=645412 | |
| step=4700 loss=3.6197 lr=2.12e-04 tokens/s=648750 | |
| step=4750 loss=3.6197 lr=2.10e-04 tokens/s=648957 | |
| step=4800 loss=3.6075 lr=2.08e-04 tokens/s=649524 | |
| step=4850 loss=3.6154 lr=2.06e-04 tokens/s=644630 | |
| step=4900 loss=3.6127 lr=2.04e-04 tokens/s=647808 | |
| step=4950 loss=3.5966 lr=2.02e-04 tokens/s=654803 | |
| step=5000 loss=3.5783 lr=2.00e-04 tokens/s=646539 | |
| step=5000 val_loss=3.5794 val_acc=0.1362 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.5794) | |
| step=5050 loss=3.6003 lr=1.98e-04 tokens/s=568030 | |
| step=5100 loss=3.5946 lr=1.96e-04 tokens/s=652864 | |
| step=5150 loss=3.5892 lr=1.94e-04 tokens/s=651122 | |
| step=5200 loss=3.5768 lr=1.92e-04 tokens/s=650155 | |
| step=5250 loss=3.5746 lr=1.90e-04 tokens/s=654272 | |
| step=5300 loss=3.5845 lr=1.88e-04 tokens/s=649189 | |
| step=5350 loss=3.5705 lr=1.87e-04 tokens/s=648502 | |
| step=5400 loss=3.5667 lr=1.85e-04 tokens/s=645674 | |
| step=5450 loss=3.5605 lr=1.83e-04 tokens/s=646637 | |
| step=5500 loss=3.5521 lr=1.81e-04 tokens/s=648717 | |
| step=5550 loss=3.5432 lr=1.79e-04 tokens/s=645135 | |
| step=5600 loss=3.5486 lr=1.76e-04 tokens/s=646696 | |
| step=5650 loss=3.5422 lr=1.74e-04 tokens/s=645679 | |
| step=5700 loss=3.5221 lr=1.72e-04 tokens/s=647129 | |
| step=5750 loss=3.5257 lr=1.70e-04 tokens/s=646527 | |
| step=5800 loss=3.5123 lr=1.68e-04 tokens/s=645406 | |
| step=5850 loss=3.5202 lr=1.66e-04 tokens/s=647427 | |
| step=5900 loss=3.5165 lr=1.64e-04 tokens/s=646285 | |
| step=5950 loss=3.5024 lr=1.62e-04 tokens/s=649661 | |
| step=6000 loss=3.4833 lr=1.60e-04 tokens/s=646107 | |
| step=6000 val_loss=3.4592 val_acc=0.1599 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.4592) | |
| saved full checkpoint to models/run_full/last.pt | |
| step=6050 loss=3.4843 lr=1.58e-04 tokens/s=565632 | |
| step=6100 loss=3.4749 lr=1.56e-04 tokens/s=649806 | |
| step=6150 loss=3.4727 lr=1.54e-04 tokens/s=644283 | |
| step=6200 loss=3.4574 lr=1.52e-04 tokens/s=646619 | |
| step=6250 loss=3.4787 lr=1.50e-04 tokens/s=650323 | |
| step=6300 loss=3.4478 lr=1.48e-04 tokens/s=648368 | |
| step=6350 loss=3.4310 lr=1.46e-04 tokens/s=649585 | |
| step=6400 loss=3.4063 lr=1.44e-04 tokens/s=648821 | |
| step=6450 loss=3.4271 lr=1.42e-04 tokens/s=648296 | |
| step=6500 loss=3.4219 lr=1.40e-04 tokens/s=653115 | |
| step=6550 loss=3.4261 lr=1.38e-04 tokens/s=648551 | |
| step=6600 loss=3.3942 lr=1.36e-04 tokens/s=648403 | |
| step=6650 loss=3.3945 lr=1.34e-04 tokens/s=647262 | |
| step=6700 loss=3.3768 lr=1.32e-04 tokens/s=645650 | |
| step=6750 loss=3.3594 lr=1.30e-04 tokens/s=645159 | |
| step=6800 loss=3.3543 lr=1.28e-04 tokens/s=644470 | |
| step=6850 loss=3.3543 lr=1.26e-04 tokens/s=657750 | |
| step=6900 loss=3.3328 lr=1.24e-04 tokens/s=644780 | |
| step=6950 loss=3.3128 lr=1.21e-04 tokens/s=648328 | |
| step=7000 loss=3.3153 lr=1.19e-04 tokens/s=643013 | |
| step=7000 val_loss=3.2710 val_acc=0.2038 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.2710) | |
| step=7050 loss=3.3335 lr=1.17e-04 tokens/s=570511 | |
| step=7100 loss=3.3108 lr=1.15e-04 tokens/s=646086 | |
| step=7150 loss=3.2824 lr=1.13e-04 tokens/s=649372 | |
| step=7200 loss=3.3036 lr=1.12e-04 tokens/s=645160 | |
| step=7250 loss=3.2910 lr=1.10e-04 tokens/s=648005 | |
| step=7300 loss=3.2758 lr=1.08e-04 tokens/s=647553 | |
| step=7350 loss=3.2637 lr=1.06e-04 tokens/s=644124 | |
| step=7400 loss=3.2364 lr=1.04e-04 tokens/s=647694 | |
| step=7450 loss=3.2612 lr=1.02e-04 tokens/s=649421 | |
| step=7500 loss=3.2440 lr=9.98e-05 tokens/s=644405 | |
| step=7550 loss=3.2507 lr=9.78e-05 tokens/s=643748 | |
| step=7600 loss=3.2201 lr=9.59e-05 tokens/s=648603 | |
| step=7650 loss=3.2008 lr=9.40e-05 tokens/s=650269 | |
| step=7700 loss=3.1980 lr=9.21e-05 tokens/s=650190 | |
| step=7750 loss=3.1916 lr=9.02e-05 tokens/s=648036 | |
| step=7800 loss=3.1887 lr=8.84e-05 tokens/s=644013 | |
| step=7850 loss=3.1614 lr=8.65e-05 tokens/s=645610 | |
| step=7900 loss=3.1984 lr=8.47e-05 tokens/s=653082 | |
| step=7950 loss=3.1655 lr=8.28e-05 tokens/s=651783 | |
| step=8000 loss=3.1442 lr=8.10e-05 tokens/s=641902 | |
| step=8000 val_loss=3.0922 val_acc=0.2378 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=3.0922) | |
| saved full checkpoint to models/run_full/last.pt | |
| step=8050 loss=3.1524 lr=7.92e-05 tokens/s=563466 | |
| step=8100 loss=3.1286 lr=7.74e-05 tokens/s=647859 | |
| step=8150 loss=3.1475 lr=7.56e-05 tokens/s=644617 | |
| step=8200 loss=3.1346 lr=7.38e-05 tokens/s=650563 | |
| step=8250 loss=3.1153 lr=7.21e-05 tokens/s=643957 | |
| step=8300 loss=3.1291 lr=7.03e-05 tokens/s=651232 | |
| step=8350 loss=3.1007 lr=6.86e-05 tokens/s=646257 | |
| step=8400 loss=3.1427 lr=6.69e-05 tokens/s=645649 | |
| step=8450 loss=3.0885 lr=6.52e-05 tokens/s=644504 | |
| step=8500 loss=3.1135 lr=6.35e-05 tokens/s=647955 | |
| step=8550 loss=3.0873 lr=6.18e-05 tokens/s=646751 | |
| step=8600 loss=3.1149 lr=6.02e-05 tokens/s=648015 | |
| step=8650 loss=3.0754 lr=5.86e-05 tokens/s=653452 | |
| step=8700 loss=3.0770 lr=5.69e-05 tokens/s=646571 | |
| step=8750 loss=3.1066 lr=5.53e-05 tokens/s=648826 | |
| step=8800 loss=3.0757 lr=5.38e-05 tokens/s=645261 | |
| step=8850 loss=3.0756 lr=5.22e-05 tokens/s=649998 | |
| step=8900 loss=3.0475 lr=5.06e-05 tokens/s=645425 | |
| step=8950 loss=3.0455 lr=4.91e-05 tokens/s=651097 | |
| step=9000 loss=3.0308 lr=4.76e-05 tokens/s=647662 | |
| step=9000 val_loss=2.9850 val_acc=0.2530 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=2.9850) | |
| step=9050 loss=3.0317 lr=4.61e-05 tokens/s=568104 | |
| step=9100 loss=3.0222 lr=4.47e-05 tokens/s=646393 | |
| step=9150 loss=3.0333 lr=4.32e-05 tokens/s=646682 | |
| step=9200 loss=3.0221 lr=4.18e-05 tokens/s=648731 | |
| step=9250 loss=3.0192 lr=4.04e-05 tokens/s=644730 | |
| step=9300 loss=3.0526 lr=3.90e-05 tokens/s=651169 | |
| step=9350 loss=3.0147 lr=3.76e-05 tokens/s=644826 | |
| step=9400 loss=3.0214 lr=3.63e-05 tokens/s=646448 | |
| step=9450 loss=3.0113 lr=3.49e-05 tokens/s=646192 | |
| step=9500 loss=3.0175 lr=3.36e-05 tokens/s=645201 | |
| step=9550 loss=3.0156 lr=3.24e-05 tokens/s=650393 | |
| step=9600 loss=3.0058 lr=3.11e-05 tokens/s=646733 | |
| step=9650 loss=3.0161 lr=2.99e-05 tokens/s=642951 | |
| step=9700 loss=2.9871 lr=2.86e-05 tokens/s=647503 | |
| step=9750 loss=3.0119 lr=2.75e-05 tokens/s=646996 | |
| step=9800 loss=3.0166 lr=2.63e-05 tokens/s=651381 | |
| step=9850 loss=3.0158 lr=2.51e-05 tokens/s=648943 | |
| step=9900 loss=3.0112 lr=2.40e-05 tokens/s=646281 | |
| step=9950 loss=3.0129 lr=2.29e-05 tokens/s=648573 | |
| step=10000 loss=2.9643 lr=2.18e-05 tokens/s=653568 | |
| step=10000 val_loss=2.9297 val_acc=0.2572 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=2.9297) | |
| saved full checkpoint to models/run_full/last.pt | |
| step=10050 loss=2.9819 lr=2.08e-05 tokens/s=563254 | |
| step=10100 loss=2.9844 lr=1.98e-05 tokens/s=646296 | |
| step=10150 loss=3.0113 lr=1.88e-05 tokens/s=646930 | |
| step=10200 loss=2.9931 lr=1.78e-05 tokens/s=647201 | |
| step=10250 loss=2.9738 lr=1.68e-05 tokens/s=653204 | |
| step=10300 loss=2.9858 lr=1.59e-05 tokens/s=641268 | |
| step=10350 loss=2.9726 lr=1.50e-05 tokens/s=655348 | |
| step=10400 loss=2.9804 lr=1.41e-05 tokens/s=648610 | |
| step=10450 loss=2.9737 lr=1.32e-05 tokens/s=644742 | |
| step=10500 loss=2.9654 lr=1.24e-05 tokens/s=648839 | |
| step=10550 loss=2.9993 lr=1.16e-05 tokens/s=647406 | |
| step=10600 loss=2.9652 lr=1.08e-05 tokens/s=652403 | |
| step=10650 loss=2.9624 lr=1.01e-05 tokens/s=648790 | |
| step=10700 loss=2.9628 lr=9.36e-06 tokens/s=649639 | |
| step=10750 loss=2.9901 lr=8.66e-06 tokens/s=650415 | |
| step=10800 loss=2.9791 lr=7.99e-06 tokens/s=648769 | |
| step=10850 loss=2.9754 lr=7.34e-06 tokens/s=640275 | |
| step=10900 loss=2.9496 lr=6.72e-06 tokens/s=645462 | |
| step=10950 loss=2.9940 lr=6.13e-06 tokens/s=650760 | |
| step=11000 loss=2.9641 lr=5.56e-06 tokens/s=642584 | |
| step=11000 val_loss=2.9066 val_acc=0.2596 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=2.9066) | |
| step=11050 loss=2.9748 lr=5.02e-06 tokens/s=568324 | |
| step=11100 loss=2.9788 lr=4.51e-06 tokens/s=644133 | |
| step=11150 loss=2.9653 lr=4.03e-06 tokens/s=646159 | |
| step=11200 loss=2.9757 lr=3.57e-06 tokens/s=643553 | |
| step=11250 loss=2.9430 lr=3.14e-06 tokens/s=648271 | |
| step=11300 loss=2.9524 lr=2.73e-06 tokens/s=642272 | |
| step=11350 loss=2.9572 lr=2.36e-06 tokens/s=653117 | |
| step=11400 loss=2.9677 lr=2.01e-06 tokens/s=644456 | |
| step=11450 loss=2.9680 lr=1.69e-06 tokens/s=650522 | |
| step=11500 loss=2.9917 lr=1.40e-06 tokens/s=646930 | |
| step=11550 loss=2.9589 lr=1.13e-06 tokens/s=632073 | |
| step=11600 loss=2.9693 lr=8.95e-07 tokens/s=644645 | |
| step=11650 loss=2.9931 lr=6.85e-07 tokens/s=651521 | |
| step=11700 loss=2.9634 lr=5.03e-07 tokens/s=646709 | |
| step=11750 loss=2.9758 lr=3.50e-07 tokens/s=644369 | |
| step=11800 loss=2.9409 lr=2.24e-07 tokens/s=651068 | |
| step=11850 loss=2.9885 lr=1.26e-07 tokens/s=644769 | |
| step=11900 loss=2.9510 lr=5.60e-08 tokens/s=643677 | |
| step=11950 loss=2.9770 lr=1.40e-08 tokens/s=648222 | |
| step=12000 loss=2.9604 lr=0.00e+00 tokens/s=645851 | |
| step=12000 val_loss=2.9020 val_acc=0.2602 | |
| saved best checkpoint to models/run_full/best.pt (val_loss=2.9020) | |
| saved full checkpoint to models/run_full/last.pt | |
Xet Storage Details
- Size:
- 13.7 kB
- Xet hash:
- 3824c76ad8e518bbfbab8f9bf03b96c94a64e300e5b19f2a9fdd305d3cb480b2
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.