Overriding config with config/train_wiki_gutenberg.py: # train a 0.5B (16L/1536) gpt2-bpe model on the combined wikipedia+gutenberg # corpus (~9B tok). The capacity-scaling run: 2x params vs our 16L/1024 (253M), # Chinchilla-near at 1 epoch (~17 tok/param). Overridden by launch flags. out_dir = 'out-wiki-gutenberg-530m-1ep' eval_interval = 2000 eval_iters = 200 log_interval = 100 always_save_checkpoint = True # local resume granularity; hf_backup.py mirrors to the Hub wandb_log = False wandb_project = 'wiki_gutenberg' wandb_run_name = '0.5b' dataset = 'wiki_gutenberg' gradient_accumulation_steps = 1 batch_size = 256 block_size = 256 # baby GPT model :) (overridden by launch flags -> 16L/1536/24head ~530M) n_layer = 6 n_head = 6 n_embd = 384 dropout = 0.1 learning_rate = 1e-3 max_iters = 5000 lr_decay_iters = 5000 min_lr = 1e-4 beta2 = 0.95 warmup_iters = 100 device = 'cpu' compile = False Overriding: device = cuda Overriding: compile = True Overriding: eval_interval = 2000 Overriding: eval_iters = 200 Overriding: log_interval = 100 Overriding: always_save_checkpoint = True Overriding: block_size = 256 Overriding: batch_size = 256 Overriding: gradient_accumulation_steps = 1 Overriding: n_layer = 16 Overriding: n_head = 24 Overriding: n_embd = 1536 Overriding: dropout = 0.1 Overriding: learning_rate = 0.0012 Overriding: min_lr = 0.00012 Overriding: warmup_iters = 5000 Overriding: beta2 = 0.95 Overriding: max_iters = 137108 Overriding: lr_decay_iters = 137108 Overriding: out_dir = out-wiki-gutenberg-530m-1ep tokens per iteration will be: 65,536 Initializing a new model from scratch defaulting to vocab_size of GPT-2 to 50304 (50257 rounded up for efficiency) number of parameters: 530.30M /root/nanoGPT/train.py:199: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. scaler = torch.cuda.amp.GradScaler(enabled=(dtype == 'float16')) num decayed parameter tensors: 66, with 530,644,992 parameters num non-decayed parameter tensors: 33, with 50,688 parameters using fused AdamW: True compiling the model... (takes a ~minute) step 0: train loss 11.1036, val loss 11.1013 iter 0: loss 11.1082, time 89170.69ms, mfu -100.00% iter 100: loss 7.1267, time 437.34ms, mfu 156.45% iter 200: loss 6.4479, time 442.17ms, mfu 156.28% iter 300: loss 5.8774, time 442.33ms, mfu 156.12% iter 400: loss 5.6738, time 440.82ms, mfu 156.03% iter 500: loss 5.3547, time 442.74ms, mfu 155.88% iter 600: loss 5.1492, time 441.76ms, mfu 155.78% iter 700: loss 5.0054, time 441.16ms, mfu 155.71% iter 800: loss 4.8405, time 442.20ms, mfu 155.61% iter 900: loss 4.7264, time 441.09ms, mfu 155.56% iter 1000: loss 4.5948, time 444.95ms, mfu 155.38% iter 1100: loss 4.3694, time 442.62ms, mfu 155.30% iter 1200: loss 4.3759, time 445.97ms, mfu 155.11% iter 1300: loss 4.2373, time 441.98ms, mfu 155.08% iter 1400: loss 4.1923, time 443.71ms, mfu 154.99% iter 1500: loss 4.1416, time 446.06ms, mfu 154.83% iter 1600: loss 4.0585, time 444.95ms, mfu 154.73% iter 1700: loss 4.1305, time 442.71ms, mfu 154.71% iter 1800: loss 4.0767, time 446.73ms, mfu 154.55% iter 1900: loss 4.0259, time 442.95ms, mfu 154.55% step 2000: train loss 3.9183, val loss 3.9155 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 2000: loss 3.9544, time 62017.87ms, mfu 139.20% iter 2100: loss 3.9057, time 446.44ms, mfu 140.61% iter 2200: loss 3.8796, time 446.73ms, mfu 141.86% iter 2300: loss 3.8484, time 447.67ms, mfu 142.96% iter 2400: loss 3.9056, time 444.35ms, mfu 144.06% iter 2500: loss 3.8618, time 446.88ms, mfu 144.97% iter 2600: loss 3.8273, time 446.55ms, mfu 145.79% iter 2700: loss 3.8333, time 440.68ms, mfu 146.74% iter 2800: loss 3.8259, time 441.91ms, mfu 147.55% iter 2900: loss 3.7795, time 448.62ms, mfu 148.04% iter 3000: loss 3.7829, time 449.51ms, mfu 148.46% iter 3100: loss 3.8019, time 444.32ms, mfu 149.01% iter 3200: loss 3.8028, time 448.94ms, mfu 149.35% iter 3300: loss 3.7317, time 445.18ms, mfu 149.79% iter 3400: loss 3.6775, time 444.78ms, mfu 150.19% iter 3500: loss 3.7106, time 443.69ms, mfu 150.59% iter 3600: loss 3.7242, time 444.40ms, mfu 150.93% iter 3700: loss 3.7189, time 447.56ms, mfu 151.12% iter 3800: loss 3.6343, time 447.22ms, mfu 151.31% iter 3900: loss 3.6177, time 446.55ms, mfu 151.50% step 4000: train loss 3.6156, val loss 3.6195 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 4000: loss 3.6261, time 66001.87ms, mfu 136.45% iter 4100: loss 3.6635, time 439.91ms, mfu 138.36% iter 4200: loss 3.6292, time 439.72ms, mfu 140.09% iter 4300: loss 3.6484, time 442.83ms, mfu 141.53% iter 4400: loss 3.6215, time 445.03ms, mfu 142.75% iter 4500: loss 3.6225, time 442.49ms, mfu 143.94% iter 4600: loss 3.6168, time 446.89ms, mfu 144.85% iter 4700: loss 3.6826, time 443.07ms, mfu 145.81% iter 4800: loss 3.5770, time 442.51ms, mfu 146.69% iter 4900: loss 3.6001, time 444.92ms, mfu 147.40% iter 5000: loss 3.6241, time 440.84ms, mfu 148.18% iter 5100: loss 3.5333, time 447.42ms, mfu 148.66% iter 5200: loss 3.6191, time 442.71ms, mfu 149.24% iter 5300: loss 3.6189, time 448.83ms, mfu 149.56% iter 5400: loss 3.6240, time 443.82ms, mfu 150.02% iter 5500: loss 3.4967, time 445.14ms, mfu 150.39% iter 5600: loss 3.4990, time 441.33ms, mfu 150.86% iter 5700: loss 3.5285, time 442.22ms, mfu 151.24% iter 5800: loss 3.5127, time 444.95ms, mfu 151.50% iter 5900: loss 3.4877, time 442.02ms, mfu 151.82% step 6000: train loss 3.4517, val loss 3.4570 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 6000: loss 3.3989, time 65862.07ms, mfu 136.75% iter 6100: loss 3.4812, time 439.08ms, mfu 138.65% iter 6200: loss 3.5478, time 442.09ms, mfu 140.27% iter 6300: loss 3.4423, time 442.49ms, mfu 141.70% iter 6400: loss 3.4052, time 443.41ms, mfu 142.96% iter 6500: loss 3.5069, time 442.02ms, mfu 144.14% iter 6600: loss 3.3985, time 441.76ms, mfu 145.22% iter 6700: loss 3.4148, time 446.59ms, mfu 146.02% iter 6800: loss 3.4067, time 441.78ms, mfu 146.90% iter 6900: loss 3.3920, time 442.39ms, mfu 147.68% iter 7000: loss 3.3811, time 441.98ms, mfu 148.39% iter 7100: loss 3.4921, time 445.73ms, mfu 148.90% iter 7200: loss 3.4252, time 445.87ms, mfu 149.36% iter 7300: loss 3.4486, time 442.58ms, mfu 149.88% iter 7400: loss 3.4207, time 444.25ms, mfu 150.29% iter 7500: loss 3.4161, time 444.91ms, mfu 150.64% iter 7600: loss 3.3954, time 442.74ms, mfu 151.03% iter 7700: loss 3.4477, time 441.99ms, mfu 151.41% iter 7800: loss 3.3723, time 442.45ms, mfu 151.73% iter 7900: loss 3.4532, time 443.68ms, mfu 151.98% step 8000: train loss 3.3458, val loss 3.3517 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 8000: loss 3.3585, time 65769.94ms, mfu 136.89% iter 8100: loss 3.4348, time 443.89ms, mfu 138.61% iter 8200: loss 3.3941, time 440.23ms, mfu 140.29% iter 8300: loss 3.4014, time 443.97ms, mfu 141.67% iter 8400: loss 3.4147, time 440.55ms, mfu 143.04% iter 8500: loss 3.4033, time 442.57ms, mfu 144.19% iter 8600: loss 3.3590, time 441.40ms, mfu 145.27% iter 8700: loss 3.3398, time 441.06ms, mfu 146.26% iter 8800: loss 3.3703, time 445.86ms, mfu 146.98% iter 8900: loss 3.3603, time 444.12ms, mfu 147.69% iter 9000: loss 3.3583, time 446.80ms, mfu 148.23% iter 9100: loss 3.3942, time 441.60ms, mfu 148.90% iter 9200: loss 3.4014, time 445.52ms, mfu 149.37% iter 9300: loss 3.3763, time 442.88ms, mfu 149.88% iter 9400: loss 3.3114, time 444.65ms, mfu 150.28% iter 9500: loss 3.4656, time 444.21ms, mfu 150.66% iter 9600: loss 3.2709, time 443.68ms, mfu 151.01% iter 9700: loss 3.3287, time 441.56ms, mfu 151.41% iter 9800: loss 3.3598, time 444.98ms, mfu 151.64% iter 9900: loss 3.2859, time 441.92ms, mfu 151.96% step 10000: train loss 3.2856, val loss 3.2934 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 10000: loss 3.3140, time 65707.54ms, mfu 136.87% iter 10100: loss 3.3347, time 440.20ms, mfu 138.72% iter 10200: loss 3.3091, time 441.62ms, mfu 140.34% iter 10300: loss 3.3442, time 444.88ms, mfu 141.69% iter 10400: loss 3.4077, time 442.05ms, mfu 143.00% iter 10500: loss 3.3773, time 444.11ms, mfu 144.10% iter 10600: loss 3.3197, time 445.05ms, mfu 145.07% iter 10700: loss 3.3867, time 445.74ms, mfu 145.91% iter 10800: loss 3.3132, time 443.69ms, mfu 146.74% iter 10900: loss 3.2219, time 442.60ms, mfu 147.53% iter 11000: loss 3.3162, time 442.65ms, mfu 148.23% iter 11100: loss 3.2492, time 443.92ms, mfu 148.82% iter 11200: loss 3.2580, time 442.62ms, mfu 149.40% iter 11300: loss 3.3678, time 442.80ms, mfu 149.91% iter 11400: loss 3.3408, time 443.72ms, mfu 150.34% iter 11500: loss 3.3912, time 441.13ms, mfu 150.81% iter 11600: loss 3.2979, time 445.79ms, mfu 151.08% iter 11700: loss 3.2721, time 443.37ms, mfu 151.40% iter 11800: loss 3.2679, time 443.61ms, mfu 151.69% iter 11900: loss 3.3370, time 440.43ms, mfu 152.05% step 12000: train loss 3.2428, val loss 3.2486 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 12000: loss 3.2503, time 65678.03ms, mfu 136.95% iter 12100: loss 3.3662, time 438.84ms, mfu 138.85% iter 12200: loss 3.2599, time 442.94ms, mfu 140.41% iter 12300: loss 3.2895, time 442.09ms, mfu 141.85% iter 12400: loss 3.2313, time 441.04ms, mfu 143.17% iter 12500: loss 3.3367, time 441.28ms, mfu 144.36% iter 12600: loss 3.2648, time 446.60ms, mfu 145.25% iter 12700: loss 3.2849, time 447.12ms, mfu 146.02% iter 12800: loss 3.2787, time 443.23ms, mfu 146.86% iter 12900: loss 3.3005, time 445.94ms, mfu 147.52% iter 13000: loss 3.2938, time 442.06ms, mfu 148.24% iter 13100: loss 3.2868, time 447.35ms, mfu 148.71% iter 13200: loss 3.2834, time 441.03ms, mfu 149.35% iter 13300: loss 3.2777, time 440.04ms, mfu 149.97% iter 13400: loss 3.2641, time 442.06ms, mfu 150.45% iter 13500: loss 3.2625, time 443.11ms, mfu 150.84% iter 13600: loss 3.2251, time 444.52ms, mfu 151.15% iter 13700: loss 3.2294, time 444.16ms, mfu 151.44% iter 13800: loss 3.2571, time 443.35ms, mfu 151.73% iter 13900: loss 3.1879, time 443.60ms, mfu 151.98% step 14000: train loss 3.2173, val loss 3.2256 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 14000: loss 3.2374, time 65827.28ms, mfu 136.89% iter 14100: loss 3.2433, time 441.60ms, mfu 138.69% iter 14200: loss 3.2675, time 443.25ms, mfu 140.26% iter 14300: loss 3.2100, time 441.87ms, mfu 141.72% iter 14400: loss 3.2757, time 444.32ms, mfu 142.94% iter 14500: loss 3.3211, time 443.98ms, mfu 144.06% iter 14600: loss 3.2441, time 441.54ms, mfu 145.15% iter 14700: loss 3.2467, time 440.77ms, mfu 146.16% iter 14800: loss 3.3270, time 444.99ms, mfu 146.92% iter 14900: loss 3.2327, time 442.88ms, mfu 147.67% iter 15000: loss 3.2169, time 442.76ms, mfu 148.36% iter 15100: loss 3.2549, time 440.97ms, mfu 149.04% iter 15200: loss 3.2018, time 446.53ms, mfu 149.46% iter 15300: loss 3.2698, time 442.74ms, mfu 149.97% iter 15400: loss 3.2472, time 441.88ms, mfu 150.45% iter 15500: loss 3.2258, time 445.46ms, mfu 150.77% iter 15600: loss 3.1997, time 440.81ms, mfu 151.21% iter 15700: loss 3.1950, time 443.81ms, mfu 151.51% iter 15800: loss 3.3171, time 442.84ms, mfu 151.81% iter 15900: loss 3.2615, time 445.16ms, mfu 152.00% step 16000: train loss 3.1977, val loss 3.2048 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 16000: loss 3.2691, time 65701.88ms, mfu 136.90% iter 16100: loss 3.2874, time 441.43ms, mfu 138.71% iter 16200: loss 3.1908, time 442.66ms, mfu 140.30% iter 16300: loss 3.1912, time 443.20ms, mfu 141.70% iter 16400: loss 3.2039, time 442.97ms, mfu 142.98% iter 16500: loss 3.1949, time 440.96ms, mfu 144.20% iter 16600: loss 3.2290, time 443.31ms, mfu 145.21% iter 16700: loss 3.2693, time 445.50ms, mfu 146.05% iter 16800: loss 3.2097, time 445.26ms, mfu 146.81% iter 16900: loss 3.2467, time 442.98ms, mfu 147.57% iter 17000: loss 3.2187, time 442.79ms, mfu 148.27% iter 17100: loss 3.2884, time 443.43ms, mfu 148.87% iter 17200: loss 3.2612, time 445.50ms, mfu 149.34% iter 17300: loss 3.2020, time 442.55ms, mfu 149.87% iter 17400: loss 3.2556, time 444.66ms, mfu 150.27% iter 17500: loss 3.2447, time 442.91ms, mfu 150.69% iter 17600: loss 3.2983, time 440.10ms, mfu 151.17% iter 17700: loss 3.1734, time 444.52ms, mfu 151.44% iter 17800: loss 3.2511, time 442.56ms, mfu 151.76% iter 17900: loss 3.2137, time 445.42ms, mfu 151.94% step 18000: train loss 3.1773, val loss 3.1889 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 18000: loss 3.2322, time 65615.98ms, mfu 136.85% iter 18100: loss 3.2220, time 440.36ms, mfu 138.71% iter 18200: loss 3.3049, time 440.15ms, mfu 140.38% iter 18300: loss 3.2544, time 446.33ms, mfu 141.67% iter 18400: loss 3.2650, time 444.13ms, mfu 142.91% iter 18500: loss 3.2118, time 441.45ms, mfu 144.12% iter 18600: loss 3.1895, time 442.50ms, mfu 145.17% iter 18700: loss 3.1941, time 443.30ms, mfu 146.09% iter 18800: loss 3.2778, time 444.64ms, mfu 146.87% iter 18900: loss 3.2339, time 441.31ms, mfu 147.68% iter 19000: loss 3.1870, time 441.61ms, mfu 148.41% iter 19100: loss 3.2831, time 443.72ms, mfu 148.99% iter 19200: loss 3.2928, time 441.89ms, mfu 149.57% iter 19300: loss 3.2986, time 445.25ms, mfu 149.98% iter 19400: loss 3.2456, time 445.22ms, mfu 150.35% iter 19500: loss 3.1557, time 445.29ms, mfu 150.68% iter 19600: loss 3.1754, time 444.03ms, mfu 151.02% iter 19700: loss 3.1769, time 443.85ms, mfu 151.33% iter 19800: loss 3.2197, time 443.63ms, mfu 151.62% iter 19900: loss 3.1796, time 441.84ms, mfu 151.95% step 20000: train loss 3.1663, val loss 3.1786 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 20000: loss 3.2533, time 65627.46ms, mfu 136.86% iter 20100: loss 3.2156, time 440.87ms, mfu 138.69% iter 20200: loss 3.2796, time 440.86ms, mfu 140.34% iter 20300: loss 3.2432, time 439.73ms, mfu 141.87% iter 20400: loss 3.2318, time 440.62ms, mfu 143.21% iter 20500: loss 3.1981, time 440.47ms, mfu 144.42% iter 20600: loss 3.2269, time 442.96ms, mfu 145.42% iter 20700: loss 3.2127, time 441.25ms, mfu 146.39% iter 20800: loss 3.2265, time 440.43ms, mfu 147.28% iter 20900: loss 3.2139, time 444.35ms, mfu 147.95% iter 21000: loss 3.2253, time 440.23ms, mfu 148.70% iter 21100: loss 3.1392, time 436.57ms, mfu 149.50% iter 21200: loss 3.2089, time 440.38ms, mfu 150.09% iter 21300: loss 3.1923, time 439.31ms, mfu 150.65% iter 21400: loss 3.2165, time 437.87ms, mfu 151.22% iter 21500: loss 3.1946, time 440.78ms, mfu 151.62% iter 21600: loss 3.2736, time 439.86ms, mfu 152.01% iter 21700: loss 3.1535, time 437.99ms, mfu 152.43% iter 21800: loss 3.2071, time 440.49ms, mfu 152.72% iter 21900: loss 3.1913, time 441.04ms, mfu 152.96% step 22000: train loss 3.1568, val loss 3.1641 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 22000: loss 3.1692, time 65310.84ms, mfu 137.77% iter 22100: loss 3.2188, time 440.52ms, mfu 139.52% iter 22200: loss 3.1753, time 440.64ms, mfu 141.10% iter 22300: loss 3.3106, time 437.58ms, mfu 142.63% iter 22400: loss 3.1838, time 440.55ms, mfu 143.89% iter 22500: loss 3.2047, time 441.35ms, mfu 145.01% iter 22600: loss 3.2066, time 441.71ms, mfu 146.00% iter 22700: loss 3.2010, time 440.30ms, mfu 146.94% iter 22800: loss 3.1363, time 441.55ms, mfu 147.74% iter 22900: loss 3.1682, time 441.18ms, mfu 148.47% iter 23000: loss 3.2113, time 443.77ms, mfu 149.04% iter 23100: loss 3.2214, time 442.58ms, mfu 149.60% iter 23200: loss 3.1835, time 440.88ms, mfu 150.16% iter 23300: loss 3.2083, time 441.10ms, mfu 150.65% iter 23400: loss 3.2242, time 441.47ms, mfu 151.09% iter 23500: loss 3.1649, time 441.47ms, mfu 151.48% iter 23600: loss 3.1838, time 440.25ms, mfu 151.87% iter 23700: loss 3.1758, time 440.56ms, mfu 152.21% iter 23800: loss 3.2205, time 439.44ms, mfu 152.56% iter 23900: loss 3.2537, time 438.36ms, mfu 152.91% step 24000: train loss 3.1421, val loss 3.1553 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 24000: loss 3.1881, time 65282.99ms, mfu 137.73% iter 24100: loss 3.2003, time 439.85ms, mfu 139.51% iter 24200: loss 3.2393, time 441.07ms, mfu 141.07% iter 24300: loss 3.2287, time 438.57ms, mfu 142.56% iter 24400: loss 3.2247, time 440.37ms, mfu 143.85% iter 24500: loss 3.1927, time 440.19ms, mfu 145.00% iter 24600: loss 3.2176, time 441.60ms, mfu 146.00% iter 24700: loss 3.1701, time 440.06ms, mfu 146.95% iter 24800: loss 3.2210, time 440.44ms, mfu 147.79% iter 24900: loss 3.1681, time 438.30ms, mfu 148.62% iter 25000: loss 3.2443, time 441.50ms, mfu 149.25% iter 25100: loss 3.2527, time 442.08ms, mfu 149.80% iter 25200: loss 3.1772, time 440.98ms, mfu 150.34% iter 25300: loss 3.1544, time 440.34ms, mfu 150.84% iter 25400: loss 3.1639, time 439.02ms, mfu 151.34% iter 25500: loss 3.1479, time 440.53ms, mfu 151.74% iter 25600: loss 3.2178, time 440.46ms, mfu 152.10% iter 25700: loss 3.1651, time 442.39ms, mfu 152.36% iter 25800: loss 3.1976, time 440.25ms, mfu 152.66% iter 25900: loss 3.1531, time 441.09ms, mfu 152.91% step 26000: train loss 3.1323, val loss 3.1432 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 26000: loss 3.2047, time 65653.33ms, mfu 137.72% iter 26100: loss 3.1503, time 441.37ms, mfu 139.45% iter 26200: loss 3.2074, time 440.35ms, mfu 141.04% iter 26300: loss 3.2216, time 440.95ms, mfu 142.46% iter 26400: loss 3.1873, time 441.01ms, mfu 143.72% iter 26500: loss 3.2559, time 441.12ms, mfu 144.86% iter 26600: loss 3.1597, time 442.74ms, mfu 145.83% iter 26700: loss 3.1844, time 441.18ms, mfu 146.76% iter 26800: loss 3.1551, time 440.31ms, mfu 147.62% iter 26900: loss 3.2272, time 440.56ms, mfu 148.39% iter 27000: loss 3.1884, time 439.58ms, mfu 149.11% iter 27100: loss 3.2449, time 439.77ms, mfu 149.76% iter 27200: loss 3.2506, time 439.02ms, mfu 150.37% iter 27300: loss 3.1501, time 438.69ms, mfu 150.93% iter 27400: loss 3.1378, time 440.86ms, mfu 151.36% iter 27500: loss 3.1627, time 440.59ms, mfu 151.75% iter 27600: loss 3.1861, time 437.84ms, mfu 152.20% iter 27700: loss 3.1644, time 438.50ms, mfu 152.58% iter 27800: loss 3.1470, time 442.58ms, mfu 152.78% iter 27900: loss 3.1677, time 440.85ms, mfu 153.03% step 28000: train loss 3.1270, val loss 3.1334 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 28000: loss 3.1931, time 65432.01ms, mfu 137.83% iter 28100: loss 3.2114, time 437.68ms, mfu 139.68% iter 28200: loss 3.2457, time 437.80ms, mfu 141.34% iter 28300: loss 3.1986, time 439.94ms, mfu 142.76% iter 28400: loss 3.1461, time 439.26ms, mfu 144.06% iter 28500: loss 3.1284, time 442.32ms, mfu 145.12% iter 28600: loss 3.2687, time 439.77ms, mfu 146.17% iter 28700: loss 3.2131, time 440.94ms, mfu 147.07% iter 28800: loss 3.1142, time 439.01ms, mfu 147.95% iter 28900: loss 3.1451, time 441.84ms, mfu 148.64% iter 29000: loss 3.1589, time 440.80ms, mfu 149.29% iter 29100: loss 3.1450, time 443.06ms, mfu 149.81% iter 29200: loss 3.1225, time 441.04ms, mfu 150.34% iter 29300: loss 3.1363, time 440.94ms, mfu 150.82% iter 29400: loss 3.1213, time 441.59ms, mfu 151.24% iter 29500: loss 3.1392, time 441.64ms, mfu 151.60% iter 29600: loss 3.1850, time 438.42ms, mfu 152.05% iter 29700: loss 3.1586, time 440.93ms, mfu 152.36% iter 29800: loss 3.2396, time 443.05ms, mfu 152.57% iter 29900: loss 3.2580, time 434.33ms, mfu 153.07% step 30000: train loss 3.1217, val loss 3.1208 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 30000: loss 3.2306, time 65346.39ms, mfu 137.86% iter 30100: loss 3.1761, time 440.21ms, mfu 139.62% iter 30200: loss 3.1694, time 436.11ms, mfu 141.35% iter 30300: loss 3.1369, time 438.80ms, mfu 142.80% iter 30400: loss 3.2328, time 442.19ms, mfu 144.00% iter 30500: loss 3.1111, time 439.89ms, mfu 145.15% iter 30600: loss 3.1187, time 435.06ms, mfu 146.36% iter 30700: loss 3.1222, time 438.36ms, mfu 147.33% iter 30800: loss 3.1422, time 441.10ms, mfu 148.11% iter 30900: loss 3.1692, time 442.65ms, mfu 148.76% iter 31000: loss 3.1235, time 440.40ms, mfu 149.42% iter 31100: loss 3.1903, time 437.54ms, mfu 150.11% iter 31200: loss 3.1061, time 437.73ms, mfu 150.73% iter 31300: loss 3.2265, time 440.03ms, mfu 151.21% iter 31400: loss 3.1683, time 440.02ms, mfu 151.64% iter 31500: loss 3.1792, time 440.73ms, mfu 152.00% iter 31600: loss 3.1770, time 439.67ms, mfu 152.36% iter 31700: loss 3.1152, time 440.78ms, mfu 152.65% iter 31800: loss 3.1440, time 440.87ms, mfu 152.90% iter 31900: loss 3.1648, time 443.14ms, mfu 153.05% step 32000: train loss 3.1174, val loss 3.1209 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 32000: loss 3.2021, time 65882.30ms, mfu 137.85% iter 32100: loss 3.1142, time 440.49ms, mfu 139.60% iter 32200: loss 3.1869, time 440.16ms, mfu 141.18% iter 32300: loss 3.1862, time 441.76ms, mfu 142.55% iter 32400: loss 3.0835, time 439.22ms, mfu 143.87% iter 32500: loss 3.1186, time 441.53ms, mfu 144.98% iter 32600: loss 3.1332, time 440.10ms, mfu 146.03% iter 32700: loss 3.1653, time 441.09ms, mfu 146.94% iter 32800: loss 3.0876, time 439.37ms, mfu 147.82% iter 32900: loss 3.1725, time 440.65ms, mfu 148.56% iter 33000: loss 3.1411, time 438.29ms, mfu 149.32% iter 33100: loss 3.1240, time 440.70ms, mfu 149.91% iter 33200: loss 3.1866, time 442.01ms, mfu 150.40% iter 33300: loss 3.1937, time 441.39ms, mfu 150.86% iter 33400: loss 3.1438, time 441.20ms, mfu 151.28% iter 33500: loss 3.1431, time 438.93ms, mfu 151.74% iter 33600: loss 3.1765, time 441.70ms, mfu 152.06% iter 33700: loss 3.1790, time 439.30ms, mfu 152.43% iter 33800: loss 3.1901, time 440.92ms, mfu 152.70% iter 33900: loss 3.1656, time 441.90ms, mfu 152.92% step 34000: train loss 3.1030, val loss 3.1095 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 34000: loss 3.1539, time 65373.02ms, mfu 137.73% iter 34100: loss 3.1435, time 441.38ms, mfu 139.46% iter 34200: loss 3.1812, time 442.57ms, mfu 140.97% iter 34300: loss 3.1066, time 442.16ms, mfu 142.35% iter 34400: loss 3.1261, time 443.71ms, mfu 143.53% iter 34500: loss 3.1057, time 440.65ms, mfu 144.71% iter 34600: loss 3.2013, time 441.57ms, mfu 145.73% iter 34700: loss 3.1375, time 440.28ms, mfu 146.70% iter 34800: loss 3.1631, time 442.39ms, mfu 147.49% iter 34900: loss 3.1176, time 440.82ms, mfu 148.27% iter 35000: loss 3.0846, time 440.21ms, mfu 148.98% iter 35100: loss 3.0308, time 437.86ms, mfu 149.71% iter 35200: loss 3.0116, time 437.97ms, mfu 150.36% iter 35300: loss 3.1386, time 439.07ms, mfu 150.91% iter 35400: loss 3.2202, time 437.69ms, mfu 151.45% iter 35500: loss 3.1263, time 438.35ms, mfu 151.91% iter 35600: loss 3.1341, time 438.88ms, mfu 152.31% iter 35700: loss 3.1480, time 439.99ms, mfu 152.63% iter 35800: loss 3.1823, time 437.49ms, mfu 153.01% iter 35900: loss 3.1703, time 437.42ms, mfu 153.35% step 36000: train loss 3.0919, val loss 3.1089 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 36000: loss 3.0796, time 65524.32ms, mfu 138.12% iter 36100: loss 3.1635, time 440.45ms, mfu 139.84% iter 36200: loss 3.1232, time 442.05ms, mfu 141.33% iter 36300: loss 3.2023, time 442.16ms, mfu 142.67% iter 36400: loss 3.0824, time 439.02ms, mfu 143.99% iter 36500: loss 3.2438, time 441.87ms, mfu 145.08% iter 36600: loss 3.0976, time 440.20ms, mfu 146.11% iter 36700: loss 3.1506, time 438.43ms, mfu 147.11% iter 36800: loss 3.1652, time 438.18ms, mfu 148.01% iter 36900: loss 3.1624, time 440.07ms, mfu 148.76% iter 37000: loss 3.1852, time 438.43ms, mfu 149.49% iter 37100: loss 3.1130, time 440.50ms, mfu 150.07% iter 37200: loss 3.1409, time 440.45ms, mfu 150.60% iter 37300: loss 3.1261, time 439.34ms, mfu 151.11% iter 37400: loss 3.1568, time 439.02ms, mfu 151.59% iter 37500: loss 3.2170, time 439.44ms, mfu 152.00% iter 37600: loss 3.1721, time 438.55ms, mfu 152.40% iter 37700: loss 3.1254, time 439.89ms, mfu 152.71% iter 37800: loss 3.2083, time 438.50ms, mfu 153.04% iter 37900: loss 3.1972, time 439.99ms, mfu 153.29% step 38000: train loss 3.0881, val loss 3.1060 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 38000: loss 3.0752, time 65400.61ms, mfu 138.07% iter 38100: loss 3.1241, time 439.69ms, mfu 139.82% iter 38200: loss 3.1406, time 440.15ms, mfu 141.38% iter 38300: loss 3.1196, time 441.75ms, mfu 142.73% iter 38400: loss 3.0871, time 439.76ms, mfu 144.02% iter 38500: loss 3.1002, time 439.26ms, mfu 145.19% iter 38600: loss 3.1627, time 442.09ms, mfu 146.15% iter 38700: loss 3.1359, time 440.17ms, mfu 147.08% iter 38800: loss 3.1181, time 439.34ms, mfu 147.94% iter 38900: loss 3.0825, time 439.96ms, mfu 148.70% iter 39000: loss 3.1528, time 441.37ms, mfu 149.33% iter 39100: loss 3.1497, time 438.43ms, mfu 150.01% iter 39200: loss 3.1530, time 441.22ms, mfu 150.51% iter 39300: loss 3.1517, time 441.30ms, mfu 150.97% iter 39400: loss 3.1346, time 438.62ms, mfu 151.47% iter 39500: loss 3.2056, time 440.11ms, mfu 151.87% iter 39600: loss 3.1069, time 439.89ms, mfu 152.23% iter 39700: loss 3.1575, time 441.61ms, mfu 152.50% iter 39800: loss 3.1291, time 439.76ms, mfu 152.81% iter 39900: loss 3.1384, time 440.10ms, mfu 153.08% step 40000: train loss 3.0795, val loss 3.0926 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 40000: loss 3.1666, time 65356.60ms, mfu 137.87% iter 40100: loss 3.1513, time 440.58ms, mfu 139.62% iter 40200: loss 3.1636, time 438.94ms, mfu 141.24% iter 40300: loss 3.1182, time 441.61ms, mfu 142.61% iter 40400: loss 3.2074, time 439.74ms, mfu 143.91% iter 40500: loss 3.1014, time 440.35ms, mfu 145.06% iter 40600: loss 3.0895, time 441.24ms, mfu 146.06% iter 40700: loss 3.1138, time 440.28ms, mfu 146.99% iter 40800: loss 3.0745, time 439.96ms, mfu 147.84% iter 40900: loss 3.1401, time 442.18ms, mfu 148.53% iter 41000: loss 3.0290, time 441.06ms, mfu 149.19% iter 41100: loss 3.1356, time 441.09ms, mfu 149.79% iter 41200: loss 3.1947, time 439.15ms, mfu 150.39% iter 41300: loss 3.1699, time 440.93ms, mfu 150.87% iter 41400: loss 3.1631, time 441.37ms, mfu 151.28% iter 41500: loss 3.0933, time 438.71ms, mfu 151.75% iter 41600: loss 3.1015, time 439.08ms, mfu 152.16% iter 41700: loss 3.1541, time 439.18ms, mfu 152.52% iter 41800: loss 3.1475, time 440.14ms, mfu 152.81% iter 41900: loss 3.1103, time 438.70ms, mfu 153.13% step 42000: train loss 3.0792, val loss 3.0862 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 42000: loss 3.2236, time 65229.12ms, mfu 137.92% iter 42100: loss 3.1352, time 441.86ms, mfu 139.61% iter 42200: loss 3.1045, time 441.40ms, mfu 141.15% iter 42300: loss 3.1690, time 442.42ms, mfu 142.50% iter 42400: loss 3.1851, time 441.56ms, mfu 143.75% iter 42500: loss 3.0816, time 441.01ms, mfu 144.89% iter 42600: loss 3.1736, time 441.96ms, mfu 145.88% iter 42700: loss 3.1924, time 441.26ms, mfu 146.80% iter 42800: loss 3.1813, time 440.16ms, mfu 147.66% iter 42900: loss 3.1326, time 441.00ms, mfu 148.41% iter 43000: loss 3.1395, time 441.12ms, mfu 149.08% iter 43100: loss 3.1630, time 440.65ms, mfu 149.70% iter 43200: loss 3.1415, time 441.35ms, mfu 150.23% iter 43300: loss 3.2113, time 440.30ms, mfu 150.75% iter 43400: loss 3.1254, time 441.50ms, mfu 151.17% iter 43500: loss 3.1059, time 441.31ms, mfu 151.56% iter 43600: loss 3.1080, time 441.81ms, mfu 151.89% iter 43700: loss 3.1519, time 441.20ms, mfu 152.21% iter 43800: loss 3.1158, time 441.66ms, mfu 152.48% iter 43900: loss 3.1154, time 439.62ms, mfu 152.79% step 44000: train loss 3.0703, val loss 3.0790 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 44000: loss 3.1420, time 65549.70ms, mfu 137.62% iter 44100: loss 3.1244, time 442.45ms, mfu 139.32% iter 44200: loss 3.0865, time 439.73ms, mfu 140.95% iter 44300: loss 3.1255, time 440.76ms, mfu 142.38% iter 44400: loss 3.1418, time 441.55ms, mfu 143.63% iter 44500: loss 3.1720, time 442.80ms, mfu 144.72% iter 44600: loss 3.1223, time 440.02ms, mfu 145.80% iter 44700: loss 3.1375, time 443.06ms, mfu 146.66% iter 44800: loss 3.1846, time 442.80ms, mfu 147.45% iter 44900: loss 3.1498, time 442.65ms, mfu 148.16% iter 45000: loss 3.0990, time 440.46ms, mfu 148.88% iter 45100: loss 3.1634, time 440.51ms, mfu 149.52% iter 45200: loss 3.0561, time 442.28ms, mfu 150.04% iter 45300: loss 3.0795, time 440.67ms, mfu 150.56% iter 45400: loss 3.1325, time 440.73ms, mfu 151.03% iter 45500: loss 3.0683, time 440.33ms, mfu 151.47% iter 45600: loss 3.0709, time 440.97ms, mfu 151.84% iter 45700: loss 3.1183, time 440.89ms, mfu 152.17% iter 45800: loss 3.0583, time 439.54ms, mfu 152.52% iter 45900: loss 3.1016, time 442.65ms, mfu 152.72% step 46000: train loss 3.0685, val loss 3.0794 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 46000: loss 3.1274, time 65180.68ms, mfu 137.56% iter 46100: loss 3.1506, time 437.55ms, mfu 139.44% iter 46200: loss 3.1104, time 437.89ms, mfu 141.12% iter 46300: loss 3.0399, time 437.23ms, mfu 142.66% iter 46400: loss 3.1694, time 437.96ms, mfu 144.01% iter 46500: loss 3.1508, time 440.17ms, mfu 145.16% iter 46600: loss 3.0851, time 443.05ms, mfu 146.08% iter 46700: loss 3.2149, time 438.40ms, mfu 147.08% iter 46800: loss 3.1135, time 440.46ms, mfu 147.91% iter 46900: loss 3.1074, time 440.79ms, mfu 148.64% iter 47000: loss 3.1327, time 440.78ms, mfu 149.30% iter 47100: loss 3.0934, time 441.38ms, mfu 149.87% iter 47200: loss 3.1263, time 439.12ms, mfu 150.46% iter 47300: loss 3.1413, time 439.19ms, mfu 151.00% iter 47400: loss 3.1168, time 439.84ms, mfu 151.45% iter 47500: loss 3.1493, time 438.65ms, mfu 151.90% iter 47600: loss 3.1645, time 441.57ms, mfu 152.21% iter 47700: loss 3.1114, time 442.51ms, mfu 152.45% iter 47800: loss 3.1660, time 441.51ms, mfu 152.70% iter 47900: loss 3.0851, time 441.90ms, mfu 152.92% step 48000: train loss 3.0560, val loss 3.0647 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 48000: loss 3.0687, time 65406.66ms, mfu 137.73% iter 48100: loss 3.1452, time 441.18ms, mfu 139.46% iter 48200: loss 3.1037, time 438.71ms, mfu 141.11% iter 48300: loss 3.1458, time 439.62ms, mfu 142.57% iter 48400: loss 3.0012, time 440.53ms, mfu 143.84% iter 48500: loss 3.0982, time 441.51ms, mfu 144.95% iter 48600: loss 3.0754, time 439.63ms, mfu 146.02% iter 48700: loss 3.2024, time 440.63ms, mfu 146.95% iter 48800: loss 3.0808, time 440.69ms, mfu 147.78% iter 48900: loss 3.1253, time 438.94ms, mfu 148.59% iter 49000: loss 3.0566, time 440.89ms, mfu 149.25% iter 49100: loss 3.1114, time 439.45ms, mfu 149.89% iter 49200: loss 3.0741, time 439.79ms, mfu 150.46% iter 49300: loss 3.1150, time 441.63ms, mfu 150.91% iter 49400: loss 3.1475, time 440.70ms, mfu 151.34% iter 49500: loss 3.0570, time 440.60ms, mfu 151.74% iter 49600: loss 3.1096, time 439.08ms, mfu 152.15% iter 49700: loss 3.0753, time 440.17ms, mfu 152.47% iter 49800: loss 3.1272, time 439.30ms, mfu 152.80% iter 49900: loss 3.0495, time 441.76ms, mfu 153.01% step 50000: train loss 3.0468, val loss 3.0641 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 50000: loss 3.1086, time 65335.35ms, mfu 137.81% iter 50100: loss 3.0666, time 440.85ms, mfu 139.55% iter 50200: loss 3.1276, time 439.71ms, mfu 141.16% iter 50300: loss 3.0820, time 438.84ms, mfu 142.63% iter 50400: loss 3.0822, time 438.22ms, mfu 143.98% iter 50500: loss 3.0515, time 439.91ms, mfu 145.14% iter 50600: loss 3.0729, time 440.66ms, mfu 146.15% iter 50700: loss 3.0959, time 440.80ms, mfu 147.06% iter 50800: loss 3.1247, time 442.07ms, mfu 147.83% iter 50900: loss 3.0912, time 441.28ms, mfu 148.55% iter 51000: loss 3.0463, time 443.04ms, mfu 149.14% iter 51100: loss 3.1229, time 439.78ms, mfu 149.78% iter 51200: loss 3.0375, time 441.10ms, mfu 150.32% iter 51300: loss 3.0534, time 440.89ms, mfu 150.80% iter 51400: loss 3.1410, time 441.94ms, mfu 151.20% iter 51500: loss 3.1354, time 441.35ms, mfu 151.59% iter 51600: loss 3.1039, time 441.20ms, mfu 151.94% iter 51700: loss 3.0724, time 439.71ms, mfu 152.30% iter 51800: loss 3.0448, time 440.62ms, mfu 152.60% iter 51900: loss 3.0713, time 440.05ms, mfu 152.89% step 52000: train loss 3.0438, val loss 3.0566 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 52000: loss 3.0551, time 65444.41ms, mfu 137.70% iter 52100: loss 3.1701, time 438.14ms, mfu 139.55% iter 52200: loss 3.1098, time 437.78ms, mfu 141.22% iter 52300: loss 3.1294, time 441.31ms, mfu 142.61% iter 52400: loss 3.0976, time 441.41ms, mfu 143.85% iter 52500: loss 3.0292, time 440.10ms, mfu 145.01% iter 52600: loss 3.0322, time 439.29ms, mfu 146.08% iter 52700: loss 3.0833, time 440.24ms, mfu 147.02% iter 52800: loss 3.0982, time 440.34ms, mfu 147.85% iter 52900: loss 3.0738, time 440.22ms, mfu 148.61% iter 53000: loss 3.1700, time 440.62ms, mfu 149.28% iter 53100: loss 3.0691, time 439.03ms, mfu 149.93% iter 53200: loss 3.0483, time 437.08ms, mfu 150.59% iter 53300: loss 3.0580, time 436.85ms, mfu 151.20% iter 53400: loss 3.1423, time 437.35ms, mfu 151.72% iter 53500: loss 3.0353, time 437.62ms, mfu 152.18% iter 53600: loss 3.0790, time 438.44ms, mfu 152.57% iter 53700: loss 3.0630, time 439.01ms, mfu 152.90% iter 53800: loss 3.0492, time 436.80ms, mfu 153.27% iter 53900: loss 3.1117, time 439.91ms, mfu 153.50% step 54000: train loss 3.0398, val loss 3.0534 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 54000: loss 3.1341, time 65440.45ms, mfu 138.25% iter 54100: loss 3.0274, time 439.48ms, mfu 140.00% iter 54200: loss 3.0515, time 440.98ms, mfu 141.51% iter 54300: loss 3.0363, time 440.06ms, mfu 142.91% iter 54400: loss 3.1620, time 441.17ms, mfu 144.13% iter 54500: loss 3.1331, time 440.40ms, mfu 145.25% iter 54600: loss 3.1015, time 439.82ms, mfu 146.28% iter 54700: loss 3.1023, time 439.37ms, mfu 147.23% iter 54800: loss 3.0969, time 441.32ms, mfu 148.01% iter 54900: loss 3.0559, time 440.49ms, mfu 148.74% iter 55000: loss 3.0933, time 439.48ms, mfu 149.43% iter 55100: loss 3.0770, time 440.10ms, mfu 150.04% iter 55200: loss 3.0865, time 439.85ms, mfu 150.59% iter 55300: loss 3.0621, time 438.69ms, mfu 151.13% iter 55400: loss 3.0879, time 438.94ms, mfu 151.60% iter 55500: loss 3.0152, time 440.65ms, mfu 151.97% iter 55600: loss 3.0510, time 440.87ms, mfu 152.29% iter 55700: loss 3.0697, time 442.77ms, mfu 152.51% iter 55800: loss 3.0890, time 436.43ms, mfu 152.94% iter 55900: loss 3.0304, time 440.71ms, mfu 153.17% step 56000: train loss 3.0321, val loss 3.0409 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 56000: loss 3.0638, time 65377.31ms, mfu 137.96% iter 56100: loss 3.0231, time 439.57ms, mfu 139.73% iter 56200: loss 3.0997, time 439.14ms, mfu 141.34% iter 56300: loss 3.1754, time 437.52ms, mfu 142.84% iter 56400: loss 3.0942, time 440.29ms, mfu 144.10% iter 56500: loss 3.0885, time 438.94ms, mfu 145.27% iter 56600: loss 3.0345, time 439.80ms, mfu 146.30% iter 56700: loss 3.0617, time 440.04ms, mfu 147.22% iter 56800: loss 3.1482, time 439.09ms, mfu 148.08% iter 56900: loss 3.1352, time 439.00ms, mfu 148.86% iter 57000: loss 3.0218, time 435.67ms, mfu 149.68% iter 57100: loss 3.0228, time 439.07ms, mfu 150.29% iter 57200: loss 3.0959, time 437.88ms, mfu 150.89% iter 57300: loss 3.1141, time 439.84ms, mfu 151.36% iter 57400: loss 3.1529, time 438.63ms, mfu 151.82% iter 57500: loss 3.0456, time 439.39ms, mfu 152.21% iter 57600: loss 3.0641, time 440.19ms, mfu 152.53% iter 57700: loss 3.0254, time 440.62ms, mfu 152.81% iter 57800: loss 3.1110, time 437.25ms, mfu 153.17% iter 57900: loss 3.0839, time 440.61ms, mfu 153.39% step 58000: train loss 3.0274, val loss 3.0375 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 58000: loss 3.1437, time 65514.69ms, mfu 138.15% iter 58100: loss 3.0870, time 438.90ms, mfu 139.93% iter 58200: loss 3.0118, time 441.30ms, mfu 141.44% iter 58300: loss 3.0542, time 440.42ms, mfu 142.83% iter 58400: loss 3.0486, time 444.09ms, mfu 143.95% iter 58500: loss 3.0095, time 440.40ms, mfu 145.09% iter 58600: loss 3.0491, time 442.61ms, mfu 146.04% iter 58700: loss 3.0548, time 440.76ms, mfu 146.96% iter 58800: loss 3.0602, time 440.24ms, mfu 147.81% iter 58900: loss 3.0851, time 442.68ms, mfu 148.48% iter 59000: loss 3.1125, time 441.40ms, mfu 149.13% iter 59100: loss 3.1091, time 440.05ms, mfu 149.77% iter 59200: loss 3.0696, time 439.04ms, mfu 150.38% iter 59300: loss 3.0043, time 441.42ms, mfu 150.84% iter 59400: loss 3.0714, time 440.00ms, mfu 151.31% iter 59500: loss 3.0090, time 441.06ms, mfu 151.69% iter 59600: loss 3.0509, time 441.22ms, mfu 152.03% iter 59700: loss 2.9839, time 439.31ms, mfu 152.40% iter 59800: loss 3.0458, time 442.26ms, mfu 152.63% iter 59900: loss 3.0372, time 439.29ms, mfu 152.94% step 60000: train loss 3.0196, val loss 3.0302 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 60000: loss 3.0854, time 65424.17ms, mfu 137.75% iter 60100: loss 3.0641, time 436.63ms, mfu 139.65% iter 60200: loss 3.0030, time 438.99ms, mfu 141.27% iter 60300: loss 3.0349, time 439.89ms, mfu 142.69% iter 60400: loss 3.0690, time 439.14ms, mfu 144.01% iter 60500: loss 3.0861, time 439.68ms, mfu 145.17% iter 60600: loss 3.0207, time 440.36ms, mfu 146.19% iter 60700: loss 3.1289, time 440.17ms, mfu 147.11% iter 60800: loss 3.0466, time 438.89ms, mfu 147.99% iter 60900: loss 3.0868, time 440.52ms, mfu 148.72% iter 61000: loss 3.0582, time 439.24ms, mfu 149.43% iter 61100: loss 3.1366, time 438.59ms, mfu 150.08% iter 61200: loss 3.0483, time 440.14ms, mfu 150.62% iter 61300: loss 3.0071, time 438.43ms, mfu 151.16% iter 61400: loss 3.0812, time 439.24ms, mfu 151.63% iter 61500: loss 3.0668, time 439.06ms, mfu 152.05% iter 61600: loss 3.1011, time 439.12ms, mfu 152.42% iter 61700: loss 3.0613, time 439.92ms, mfu 152.73% iter 61800: loss 3.0067, time 441.30ms, mfu 152.96% iter 61900: loss 3.0275, time 439.99ms, mfu 153.22% step 62000: train loss 3.0096, val loss 3.0239 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 62000: loss 3.1109, time 65298.10ms, mfu 138.00% iter 62100: loss 3.1113, time 439.43ms, mfu 139.77% iter 62200: loss 3.0218, time 441.20ms, mfu 141.30% iter 62300: loss 3.0534, time 439.83ms, mfu 142.73% iter 62400: loss 3.0612, time 440.07ms, mfu 144.00% iter 62500: loss 3.0278, time 439.53ms, mfu 145.17% iter 62600: loss 3.1147, time 438.57ms, mfu 146.25% iter 62700: loss 3.0488, time 439.77ms, mfu 147.19% iter 62800: loss 3.0801, time 438.67ms, mfu 148.06% iter 62900: loss 3.0712, time 440.71ms, mfu 148.78% iter 63000: loss 3.1122, time 438.73ms, mfu 149.50% iter 63100: loss 3.1121, time 438.09ms, mfu 150.17% iter 63200: loss 3.0891, time 438.62ms, mfu 150.75% iter 63300: loss 3.0539, time 439.92ms, mfu 151.23% iter 63400: loss 3.0250, time 440.59ms, mfu 151.63% iter 63500: loss 3.0018, time 441.27ms, mfu 151.98% iter 63600: loss 3.0495, time 440.77ms, mfu 152.30% iter 63700: loss 3.0530, time 440.74ms, mfu 152.60% iter 63800: loss 3.0913, time 441.20ms, mfu 152.84% iter 63900: loss 3.1316, time 440.95ms, mfu 153.08% step 64000: train loss 3.0051, val loss 3.0152 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 64000: loss 3.0311, time 65547.11ms, mfu 137.87% iter 64100: loss 3.0031, time 438.93ms, mfu 139.67% iter 64200: loss 3.0113, time 437.60ms, mfu 141.34% iter 64300: loss 3.0274, time 438.37ms, mfu 142.82% iter 64400: loss 2.9791, time 437.53ms, mfu 144.17% iter 64500: loss 3.1084, time 440.40ms, mfu 145.29% iter 64600: loss 3.0577, time 437.58ms, mfu 146.40% iter 64700: loss 3.0359, time 437.50ms, mfu 147.40% iter 64800: loss 3.0027, time 442.06ms, mfu 148.13% iter 64900: loss 3.0366, time 440.09ms, mfu 148.87% iter 65000: loss 3.0392, time 440.47ms, mfu 149.51% iter 65100: loss 3.0057, time 441.29ms, mfu 150.07% iter 65200: loss 3.0551, time 438.87ms, mfu 150.65% iter 65300: loss 3.0577, time 442.30ms, mfu 151.05% iter 65400: loss 2.9782, time 440.84ms, mfu 151.47% iter 65500: loss 3.0549, time 440.91ms, mfu 151.84% iter 65600: loss 3.0731, time 440.99ms, mfu 152.17% iter 65700: loss 3.0597, time 440.05ms, mfu 152.50% iter 65800: loss 3.0591, time 439.47ms, mfu 152.82% iter 65900: loss 3.0704, time 440.06ms, mfu 153.09% step 66000: train loss 2.9948, val loss 3.0059 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 66000: loss 2.9691, time 65714.02ms, mfu 137.88% iter 66100: loss 2.9972, time 438.43ms, mfu 139.70% iter 66200: loss 3.1161, time 437.73ms, mfu 141.36% iter 66300: loss 3.0164, time 438.58ms, mfu 142.83% iter 66400: loss 3.0650, time 439.98ms, mfu 144.09% iter 66500: loss 3.1190, time 438.58ms, mfu 145.28% iter 66600: loss 3.0110, time 438.56ms, mfu 146.36% iter 66700: loss 3.0292, time 439.49ms, mfu 147.29% iter 66800: loss 3.0599, time 438.23ms, mfu 148.17% iter 66900: loss 3.0774, time 438.24ms, mfu 148.97% iter 67000: loss 3.0405, time 438.14ms, mfu 149.69% iter 67100: loss 3.0034, time 438.40ms, mfu 150.33% iter 67200: loss 3.0306, time 438.40ms, mfu 150.90% iter 67300: loss 3.0122, time 437.84ms, mfu 151.44% iter 67400: loss 3.0555, time 438.40ms, mfu 151.90% iter 67500: loss 3.0834, time 438.13ms, mfu 152.33% iter 67600: loss 3.0676, time 438.21ms, mfu 152.71% iter 67700: loss 3.0356, time 438.19ms, mfu 153.05% iter 67800: loss 3.0711, time 437.53ms, mfu 153.38% iter 67900: loss 3.0364, time 440.20ms, mfu 153.59% step 68000: train loss 2.9898, val loss 3.0041 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 68000: loss 3.0174, time 65342.47ms, mfu 138.33% iter 68100: loss 3.0396, time 442.89ms, mfu 139.95% iter 68200: loss 3.0384, time 438.06ms, mfu 141.57% iter 68300: loss 3.0651, time 437.81ms, mfu 143.04% iter 68400: loss 3.0246, time 439.26ms, mfu 144.32% iter 68500: loss 3.0212, time 439.19ms, mfu 145.46% iter 68600: loss 3.0281, time 438.28ms, mfu 146.53% iter 68700: loss 2.9349, time 440.20ms, mfu 147.42% iter 68800: loss 3.0149, time 438.29ms, mfu 148.29% iter 68900: loss 2.9418, time 439.87ms, mfu 149.01% iter 69000: loss 2.9298, time 440.16ms, mfu 149.66% iter 69100: loss 3.0403, time 439.74ms, mfu 150.25% iter 69200: loss 3.0286, time 440.03ms, mfu 150.77% iter 69300: loss 3.0987, time 437.97ms, mfu 151.32% iter 69400: loss 3.0637, time 439.35ms, mfu 151.76% iter 69500: loss 2.9498, time 442.28ms, mfu 152.05% iter 69600: loss 3.1574, time 440.45ms, mfu 152.38% iter 69700: loss 3.0499, time 439.77ms, mfu 152.70% iter 69800: loss 3.0068, time 442.84ms, mfu 152.88% iter 69900: loss 3.0940, time 441.11ms, mfu 153.11% step 70000: train loss 2.9858, val loss 2.9930 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 70000: loss 3.0232, time 65356.00ms, mfu 137.90% iter 70100: loss 3.1126, time 434.98ms, mfu 139.84% iter 70200: loss 3.0199, time 438.90ms, mfu 141.44% iter 70300: loss 3.0699, time 438.10ms, mfu 142.92% iter 70400: loss 3.0581, time 443.47ms, mfu 144.05% iter 70500: loss 2.9961, time 441.07ms, mfu 145.16% iter 70600: loss 3.0263, time 438.86ms, mfu 146.24% iter 70700: loss 2.9721, time 440.47ms, mfu 147.15% iter 70800: loss 3.0791, time 440.17ms, mfu 147.97% iter 70900: loss 3.0591, time 440.57ms, mfu 148.71% iter 71000: loss 3.0493, time 441.22ms, mfu 149.34% iter 71100: loss 3.0477, time 440.28ms, mfu 149.95% iter 71200: loss 2.9989, time 441.75ms, mfu 150.44% iter 71300: loss 2.9560, time 440.06ms, mfu 150.95% iter 71400: loss 3.0193, time 440.04ms, mfu 151.40% iter 71500: loss 3.0037, time 440.68ms, mfu 151.79% iter 71600: loss 2.9993, time 437.52ms, mfu 152.25% iter 71700: loss 3.0268, time 440.95ms, mfu 152.54% iter 71800: loss 2.9997, time 439.07ms, mfu 152.87% iter 71900: loss 3.0690, time 440.83ms, mfu 153.10% step 72000: train loss 2.9735, val loss 2.9862 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 72000: loss 3.0948, time 65598.84ms, mfu 137.90% iter 72100: loss 3.0025, time 439.80ms, mfu 139.66% iter 72200: loss 3.0854, time 436.41ms, mfu 141.37% iter 72300: loss 3.0258, time 439.93ms, mfu 142.79% iter 72400: loss 3.0217, time 437.84ms, mfu 144.14% iter 72500: loss 3.0330, time 439.45ms, mfu 145.29% iter 72600: loss 3.0372, time 439.98ms, mfu 146.31% iter 72700: loss 3.0036, time 439.52ms, mfu 147.25% iter 72800: loss 3.0626, time 440.25ms, mfu 148.07% iter 72900: loss 3.0605, time 437.06ms, mfu 148.91% iter 73000: loss 2.9687, time 441.76ms, mfu 149.51% iter 73100: loss 3.0497, time 438.09ms, mfu 150.18% iter 73200: loss 3.0122, time 440.22ms, mfu 150.70% iter 73300: loss 3.0715, time 439.92ms, mfu 151.19% iter 73400: loss 3.0639, time 439.37ms, mfu 151.64% iter 73500: loss 3.0497, time 438.90ms, mfu 152.06% iter 73600: loss 3.0005, time 439.80ms, mfu 152.41% iter 73700: loss 3.0588, time 439.21ms, mfu 152.75% iter 73800: loss 3.0406, time 439.66ms, mfu 153.04% iter 73900: loss 2.9904, time 438.64ms, mfu 153.33% step 74000: train loss 2.9638, val loss 2.9808 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 74000: loss 3.0316, time 65488.44ms, mfu 138.10% iter 74100: loss 3.1043, time 437.11ms, mfu 139.95% iter 74200: loss 2.9795, time 437.80ms, mfu 141.58% iter 74300: loss 3.0361, time 437.77ms, mfu 143.05% iter 74400: loss 2.9249, time 437.77ms, mfu 144.37% iter 74500: loss 3.0242, time 437.82ms, mfu 145.56% iter 74600: loss 2.9896, time 436.90ms, mfu 146.67% iter 74700: loss 3.0461, time 437.00ms, mfu 147.66% iter 74800: loss 2.9698, time 438.96ms, mfu 148.48% iter 74900: loss 3.0505, time 437.60ms, mfu 149.27% iter 75000: loss 3.0266, time 436.93ms, mfu 150.00% iter 75100: loss 2.9543, time 438.34ms, mfu 150.61% iter 75200: loss 3.0689, time 439.56ms, mfu 151.11% iter 75300: loss 3.0272, time 437.55ms, mfu 151.64% iter 75400: loss 3.0343, time 438.84ms, mfu 152.07% iter 75500: loss 2.9909, time 438.34ms, mfu 152.47% iter 75600: loss 2.9259, time 433.45ms, mfu 153.01% iter 75700: loss 2.9975, time 437.86ms, mfu 153.33% iter 75800: loss 2.9459, time 437.86ms, mfu 153.63% iter 75900: loss 3.0009, time 437.00ms, mfu 153.92% step 76000: train loss 2.9573, val loss 2.9723 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 76000: loss 2.9639, time 65298.98ms, mfu 138.63% iter 76100: loss 2.9857, time 437.93ms, mfu 140.39% iter 76200: loss 2.9169, time 438.19ms, mfu 141.97% iter 76300: loss 3.0461, time 438.74ms, mfu 143.37% iter 76400: loss 3.0179, time 437.18ms, mfu 144.68% iter 76500: loss 3.0035, time 437.48ms, mfu 145.85% iter 76600: loss 2.9923, time 438.13ms, mfu 146.88% iter 76700: loss 3.0297, time 438.27ms, mfu 147.81% iter 76800: loss 2.9872, time 439.00ms, mfu 148.61% iter 76900: loss 3.0313, time 438.51ms, mfu 149.35% iter 77000: loss 2.9735, time 438.09ms, mfu 150.03% iter 77100: loss 3.0751, time 436.92ms, mfu 150.69% iter 77200: loss 2.9456, time 437.85ms, mfu 151.25% iter 77300: loss 2.9293, time 438.58ms, mfu 151.72% iter 77400: loss 2.9897, time 437.26ms, mfu 152.20% iter 77500: loss 3.0412, time 437.71ms, mfu 152.61% iter 77600: loss 3.0513, time 437.83ms, mfu 152.98% iter 77700: loss 3.0317, time 438.25ms, mfu 153.29% iter 77800: loss 3.0361, time 437.38ms, mfu 153.61% iter 77900: loss 2.9983, time 438.15ms, mfu 153.86% step 78000: train loss 2.9514, val loss 2.9628 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 78000: loss 3.0739, time 65598.55ms, mfu 138.58% iter 78100: loss 3.0574, time 437.45ms, mfu 140.36% iter 78200: loss 2.9734, time 437.92ms, mfu 141.95% iter 78300: loss 3.0352, time 439.09ms, mfu 143.34% iter 78400: loss 2.9283, time 438.14ms, mfu 144.62% iter 78500: loss 3.0011, time 437.72ms, mfu 145.79% iter 78600: loss 2.9932, time 438.82ms, mfu 146.80% iter 78700: loss 3.0094, time 436.45ms, mfu 147.80% iter 78800: loss 2.9800, time 438.74ms, mfu 148.61% iter 78900: loss 2.9542, time 440.57ms, mfu 149.28% iter 79000: loss 2.9856, time 438.44ms, mfu 149.96% iter 79100: loss 2.9302, time 439.43ms, mfu 150.53% iter 79200: loss 2.9779, time 441.39ms, mfu 150.98% iter 79300: loss 3.0164, time 440.53ms, mfu 151.41% iter 79400: loss 2.9777, time 442.27ms, mfu 151.74% iter 79500: loss 3.0307, time 440.36ms, mfu 152.11% iter 79600: loss 3.0342, time 435.90ms, mfu 152.59% iter 79700: loss 3.0273, time 440.87ms, mfu 152.85% iter 79800: loss 3.0280, time 437.85ms, mfu 153.19% iter 79900: loss 2.9910, time 439.48ms, mfu 153.44% step 80000: train loss 2.9408, val loss 2.9569 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 80000: loss 2.9765, time 65411.26ms, mfu 138.20% iter 80100: loss 3.0190, time 439.59ms, mfu 139.95% iter 80200: loss 2.9609, time 440.15ms, mfu 141.50% iter 80300: loss 3.0224, time 439.79ms, mfu 142.90% iter 80400: loss 3.0200, time 438.31ms, mfu 144.22% iter 80500: loss 2.9863, time 440.80ms, mfu 145.32% iter 80600: loss 2.9867, time 438.36ms, mfu 146.40% iter 80700: loss 2.9357, time 438.45ms, mfu 147.36% iter 80800: loss 2.9972, time 438.61ms, mfu 148.23% iter 80900: loss 3.0588, time 439.56ms, mfu 148.97% iter 81000: loss 3.0695, time 437.13ms, mfu 149.72% iter 81100: loss 2.9412, time 439.88ms, mfu 150.31% iter 81200: loss 3.0060, time 438.25ms, mfu 150.89% iter 81300: loss 3.0475, time 440.86ms, mfu 151.32% iter 81400: loss 3.0078, time 442.07ms, mfu 151.66% iter 81500: loss 2.9458, time 440.70ms, mfu 152.02% iter 81600: loss 3.0411, time 440.26ms, mfu 152.36% iter 81700: loss 3.0199, time 442.20ms, mfu 152.60% iter 81800: loss 2.9923, time 441.32ms, mfu 152.84% iter 81900: loss 2.9766, time 442.32ms, mfu 153.03% step 82000: train loss 2.9362, val loss 2.9501 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 82000: loss 2.9465, time 65582.34ms, mfu 137.83% iter 82100: loss 2.9394, time 439.73ms, mfu 139.60% iter 82200: loss 3.0818, time 440.35ms, mfu 141.18% iter 82300: loss 3.0278, time 436.19ms, mfu 142.75% iter 82400: loss 2.9714, time 439.33ms, mfu 144.05% iter 82500: loss 2.8894, time 438.94ms, mfu 145.23% iter 82600: loss 2.9588, time 438.57ms, mfu 146.31% iter 82700: loss 3.0096, time 439.47ms, mfu 147.25% iter 82800: loss 2.9659, time 438.53ms, mfu 148.12% iter 82900: loss 2.9508, time 441.04ms, mfu 148.83% iter 83000: loss 3.0372, time 440.44ms, mfu 149.48% iter 83100: loss 3.0274, time 441.16ms, mfu 150.04% iter 83200: loss 2.9807, time 438.32ms, mfu 150.64% iter 83300: loss 2.9766, time 438.18ms, mfu 151.19% iter 83400: loss 2.9950, time 441.22ms, mfu 151.58% iter 83500: loss 2.9684, time 440.90ms, mfu 151.94% iter 83600: loss 3.0050, time 440.87ms, mfu 152.27% iter 83700: loss 2.9657, time 441.82ms, mfu 152.53% iter 83800: loss 2.9481, time 439.63ms, mfu 152.84% iter 83900: loss 2.9321, time 439.32ms, mfu 153.13% step 84000: train loss 2.9262, val loss 2.9444 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 84000: loss 3.0083, time 65493.75ms, mfu 137.92% iter 84100: loss 2.9745, time 441.36ms, mfu 139.63% iter 84200: loss 2.9842, time 441.42ms, mfu 141.17% iter 84300: loss 2.9390, time 441.32ms, mfu 142.55% iter 84400: loss 2.9650, time 442.82ms, mfu 143.75% iter 84500: loss 2.9894, time 441.60ms, mfu 144.87% iter 84600: loss 3.0643, time 437.55ms, mfu 146.02% iter 84700: loss 2.9741, time 441.67ms, mfu 146.91% iter 84800: loss 2.9471, time 440.87ms, mfu 147.74% iter 84900: loss 3.0156, time 439.49ms, mfu 148.53% iter 85000: loss 2.9154, time 440.09ms, mfu 149.23% iter 85100: loss 2.9607, time 441.57ms, mfu 149.80% iter 85200: loss 3.0374, time 442.13ms, mfu 150.29% iter 85300: loss 2.9413, time 438.47ms, mfu 150.87% iter 85400: loss 2.9965, time 438.95ms, mfu 151.37% iter 85500: loss 2.9431, time 439.23ms, mfu 151.81% iter 85600: loss 2.9472, time 441.03ms, mfu 152.14% iter 85700: loss 2.9410, time 441.67ms, mfu 152.42% iter 85800: loss 2.9860, time 441.14ms, mfu 152.69% iter 85900: loss 2.9026, time 437.76ms, mfu 153.05% step 86000: train loss 2.9146, val loss 2.9330 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 86000: loss 2.9082, time 65547.86ms, mfu 137.85% iter 86100: loss 2.9350, time 438.56ms, mfu 139.66% iter 86200: loss 2.9690, time 437.37ms, mfu 141.34% iter 86300: loss 2.9492, time 437.77ms, mfu 142.84% iter 86400: loss 2.9826, time 436.61ms, mfu 144.22% iter 86500: loss 2.9921, time 439.23ms, mfu 145.38% iter 86600: loss 2.8777, time 434.29ms, mfu 146.59% iter 86700: loss 2.9166, time 438.70ms, mfu 147.53% iter 86800: loss 2.9873, time 438.81ms, mfu 148.37% iter 86900: loss 2.9912, time 438.29ms, mfu 149.14% iter 87000: loss 2.9583, time 440.15ms, mfu 149.77% iter 87100: loss 2.9258, time 441.78ms, mfu 150.28% iter 87200: loss 2.9614, time 439.83ms, mfu 150.81% iter 87300: loss 3.0091, time 440.73ms, mfu 151.26% iter 87400: loss 2.9660, time 439.67ms, mfu 151.69% iter 87500: loss 2.9393, time 438.90ms, mfu 152.11% iter 87600: loss 2.9854, time 438.23ms, mfu 152.51% iter 87700: loss 3.0175, time 439.13ms, mfu 152.84% iter 87800: loss 2.9602, time 440.80ms, mfu 153.08% iter 87900: loss 2.9994, time 441.11ms, mfu 153.28% step 88000: train loss 2.9052, val loss 2.9267 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 88000: loss 2.9541, time 65709.25ms, mfu 138.06% iter 88100: loss 2.9144, time 440.60ms, mfu 139.78% iter 88200: loss 2.9477, time 436.93ms, mfu 141.46% iter 88300: loss 2.9622, time 440.49ms, mfu 142.85% iter 88400: loss 2.9708, time 440.44ms, mfu 144.10% iter 88500: loss 3.0075, time 441.13ms, mfu 145.20% iter 88600: loss 2.9409, time 440.86ms, mfu 146.20% iter 88700: loss 2.8936, time 440.64ms, mfu 147.11% iter 88800: loss 2.9464, time 438.66ms, mfu 147.99% iter 88900: loss 2.9645, time 438.04ms, mfu 148.81% iter 89000: loss 2.9264, time 437.94ms, mfu 149.56% iter 89100: loss 3.0051, time 438.51ms, mfu 150.20% iter 89200: loss 2.9246, time 441.27ms, mfu 150.69% iter 89300: loss 2.8754, time 441.59ms, mfu 151.11% iter 89400: loss 2.9401, time 439.79ms, mfu 151.56% iter 89500: loss 2.9174, time 439.98ms, mfu 151.95% iter 89600: loss 2.9097, time 442.98ms, mfu 152.20% iter 89700: loss 2.9532, time 441.70ms, mfu 152.47% iter 89800: loss 2.9357, time 440.28ms, mfu 152.77% iter 89900: loss 2.9332, time 440.80ms, mfu 153.01% step 90000: train loss 2.9026, val loss 2.9190 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 90000: loss 2.9688, time 65412.83ms, mfu 137.82% iter 90100: loss 2.9655, time 441.77ms, mfu 139.52% iter 90200: loss 2.9756, time 442.40ms, mfu 141.04% iter 90300: loss 2.9108, time 442.07ms, mfu 142.41% iter 90400: loss 2.9303, time 440.97ms, mfu 143.68% iter 90500: loss 2.9480, time 442.29ms, mfu 144.79% iter 90600: loss 2.9380, time 441.09ms, mfu 145.82% iter 90700: loss 2.8538, time 440.57ms, mfu 146.77% iter 90800: loss 2.9550, time 439.72ms, mfu 147.65% iter 90900: loss 2.8678, time 442.74ms, mfu 148.34% iter 91000: loss 2.9443, time 441.51ms, mfu 149.00% iter 91100: loss 2.9807, time 441.07ms, mfu 149.61% iter 91200: loss 2.9519, time 442.53ms, mfu 150.11% iter 91300: loss 2.9118, time 443.00ms, mfu 150.55% iter 91400: loss 2.8929, time 441.83ms, mfu 150.98% iter 91500: loss 3.0036, time 441.27ms, mfu 151.39% iter 91600: loss 2.8623, time 440.88ms, mfu 151.77% iter 91700: loss 3.0222, time 440.24ms, mfu 152.13% iter 91800: loss 2.9247, time 439.33ms, mfu 152.49% iter 91900: loss 2.9979, time 441.10ms, mfu 152.75% step 92000: train loss 2.8951, val loss 2.9064 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 92000: loss 2.9599, time 65688.85ms, mfu 137.58% iter 92100: loss 2.9686, time 439.00ms, mfu 139.41% iter 92200: loss 2.9342, time 440.28ms, mfu 141.01% iter 92300: loss 2.9535, time 436.45ms, mfu 142.58% iter 92400: loss 2.9564, time 439.53ms, mfu 143.89% iter 92500: loss 2.9179, time 438.95ms, mfu 145.09% iter 92600: loss 2.9405, time 440.72ms, mfu 146.11% iter 92700: loss 3.0298, time 440.05ms, mfu 147.04% iter 92800: loss 2.9647, time 440.84ms, mfu 147.86% iter 92900: loss 2.9731, time 440.29ms, mfu 148.61% iter 93000: loss 2.8835, time 442.02ms, mfu 149.23% iter 93100: loss 2.9397, time 440.14ms, mfu 149.85% iter 93200: loss 2.9325, time 439.87ms, mfu 150.42% iter 93300: loss 2.9509, time 440.69ms, mfu 150.91% iter 93400: loss 2.9944, time 440.90ms, mfu 151.33% iter 93500: loss 2.9185, time 440.65ms, mfu 151.73% iter 93600: loss 2.9569, time 440.10ms, mfu 152.10% iter 93700: loss 2.9348, time 440.70ms, mfu 152.42% iter 93800: loss 2.9680, time 439.96ms, mfu 152.73% iter 93900: loss 2.9438, time 441.06ms, mfu 152.97% step 94000: train loss 2.8807, val loss 2.9015 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 94000: loss 2.9329, time 65846.10ms, mfu 137.77% iter 94100: loss 2.9846, time 441.87ms, mfu 139.48% iter 94200: loss 2.9531, time 441.94ms, mfu 141.01% iter 94300: loss 2.9500, time 440.87ms, mfu 142.43% iter 94400: loss 2.9647, time 443.10ms, mfu 143.63% iter 94500: loss 2.9319, time 441.18ms, mfu 144.78% iter 94600: loss 2.9485, time 440.18ms, mfu 145.84% iter 94700: loss 3.0074, time 441.72ms, mfu 146.75% iter 94800: loss 2.9065, time 441.31ms, mfu 147.58% iter 94900: loss 2.9364, time 440.77ms, mfu 148.34% iter 95000: loss 2.9349, time 441.64ms, mfu 149.00% iter 95100: loss 2.9150, time 438.78ms, mfu 149.69% iter 95200: loss 2.9206, time 437.76ms, mfu 150.35% iter 95300: loss 2.9083, time 438.51ms, mfu 150.92% iter 95400: loss 2.9255, time 437.82ms, mfu 151.46% iter 95500: loss 2.9246, time 438.02ms, mfu 151.93% iter 95600: loss 3.0054, time 438.74ms, mfu 152.33% iter 95700: loss 2.9140, time 442.86ms, mfu 152.55% iter 95800: loss 2.9589, time 440.32ms, mfu 152.83% iter 95900: loss 2.8783, time 439.88ms, mfu 153.10% step 96000: train loss 2.8689, val loss 2.8914 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 96000: loss 2.8810, time 65397.45ms, mfu 137.90% iter 96100: loss 2.9268, time 441.61ms, mfu 139.60% iter 96200: loss 2.9547, time 442.89ms, mfu 141.09% iter 96300: loss 2.9126, time 441.65ms, mfu 142.47% iter 96400: loss 2.9130, time 441.00ms, mfu 143.74% iter 96500: loss 2.9455, time 439.69ms, mfu 144.93% iter 96600: loss 2.9946, time 441.06ms, mfu 145.95% iter 96700: loss 2.9099, time 441.26ms, mfu 146.86% iter 96800: loss 2.9280, time 441.13ms, mfu 147.68% iter 96900: loss 2.8698, time 441.52ms, mfu 148.41% iter 97000: loss 2.9649, time 438.83ms, mfu 149.16% iter 97100: loss 2.8705, time 434.39ms, mfu 150.00% iter 97200: loss 2.8748, time 440.30ms, mfu 150.54% iter 97300: loss 2.9543, time 439.79ms, mfu 151.04% iter 97400: loss 2.9275, time 438.32ms, mfu 151.55% iter 97500: loss 2.9206, time 442.56ms, mfu 151.85% iter 97600: loss 2.9960, time 438.72ms, mfu 152.26% iter 97700: loss 2.8706, time 441.10ms, mfu 152.55% iter 97800: loss 2.9292, time 441.47ms, mfu 152.79% iter 97900: loss 2.9589, time 439.12ms, mfu 153.09% step 98000: train loss 2.8713, val loss 2.8859 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 98000: loss 2.9433, time 65360.28ms, mfu 137.89% iter 98100: loss 2.9236, time 437.82ms, mfu 139.73% iter 98200: loss 2.9142, time 437.63ms, mfu 141.39% iter 98300: loss 2.9327, time 439.98ms, mfu 142.80% iter 98400: loss 2.8660, time 441.16ms, mfu 144.03% iter 98500: loss 2.9172, time 438.12ms, mfu 145.24% iter 98600: loss 2.8958, time 441.72ms, mfu 146.21% iter 98700: loss 2.9251, time 438.77ms, mfu 147.18% iter 98800: loss 3.0009, time 441.02ms, mfu 147.98% iter 98900: loss 2.9627, time 440.88ms, mfu 148.70% iter 99000: loss 2.9117, time 438.90ms, mfu 149.42% iter 99100: loss 2.9761, time 442.32ms, mfu 149.94% iter 99200: loss 2.9181, time 439.01ms, mfu 150.54% iter 99300: loss 2.9122, time 441.91ms, mfu 150.96% iter 99400: loss 2.9218, time 438.08ms, mfu 151.49% iter 99500: loss 2.8803, time 439.53ms, mfu 151.90% iter 99600: loss 2.9193, time 439.50ms, mfu 152.28% iter 99700: loss 2.8830, time 441.20ms, mfu 152.56% iter 99800: loss 2.8586, time 440.78ms, mfu 152.83% iter 99900: loss 2.8837, time 442.08ms, mfu 153.02% step 100000: train loss 2.8606, val loss 2.8797 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 100000: loss 2.9445, time 65533.23ms, mfu 137.82% iter 100100: loss 2.8718, time 439.55ms, mfu 139.61% iter 100200: loss 2.8828, time 440.85ms, mfu 141.17% iter 100300: loss 2.9010, time 440.61ms, mfu 142.58% iter 100400: loss 2.9367, time 441.15ms, mfu 143.83% iter 100500: loss 2.9438, time 440.24ms, mfu 144.99% iter 100600: loss 2.8464, time 441.28ms, mfu 146.00% iter 100700: loss 2.9042, time 440.06ms, mfu 146.94% iter 100800: loss 2.9074, time 439.75ms, mfu 147.81% iter 100900: loss 2.8091, time 441.73ms, mfu 148.52% iter 101000: loss 2.8996, time 441.06ms, mfu 149.18% iter 101100: loss 2.8789, time 439.35ms, mfu 149.83% iter 101200: loss 2.8777, time 440.51ms, mfu 150.38% iter 101300: loss 2.8334, time 440.83ms, mfu 150.86% iter 101400: loss 2.8216, time 440.47ms, mfu 151.31% iter 101500: loss 2.8976, time 442.36ms, mfu 151.65% iter 101600: loss 2.8487, time 440.83ms, mfu 152.00% iter 101700: loss 2.9260, time 439.90ms, mfu 152.36% iter 101800: loss 2.8934, time 440.74ms, mfu 152.64% iter 101900: loss 2.9018, time 438.32ms, mfu 152.99% step 102000: train loss 2.8502, val loss 2.8641 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 102000: loss 2.8634, time 65283.53ms, mfu 137.80% iter 102100: loss 2.8315, time 440.46ms, mfu 139.55% iter 102200: loss 2.8964, time 438.88ms, mfu 141.18% iter 102300: loss 2.8236, time 438.63ms, mfu 142.66% iter 102400: loss 2.8673, time 439.93ms, mfu 143.95% iter 102500: loss 2.8717, time 440.87ms, mfu 145.08% iter 102600: loss 2.9232, time 441.14ms, mfu 146.08% iter 102700: loss 2.9135, time 438.75ms, mfu 147.06% iter 102800: loss 2.8939, time 439.90ms, mfu 147.91% iter 102900: loss 2.8452, time 440.20ms, mfu 148.66% iter 103000: loss 2.9538, time 440.85ms, mfu 149.32% iter 103100: loss 2.8981, time 440.72ms, mfu 149.91% iter 103200: loss 2.8443, time 440.97ms, mfu 150.43% iter 103300: loss 2.8879, time 440.18ms, mfu 150.93% iter 103400: loss 2.9201, time 440.99ms, mfu 151.36% iter 103500: loss 2.9173, time 441.14ms, mfu 151.73% iter 103600: loss 2.9144, time 442.33ms, mfu 152.03% iter 103700: loss 2.9383, time 435.56ms, mfu 152.53% iter 103800: loss 2.8762, time 444.99ms, mfu 152.65% iter 103900: loss 2.8043, time 437.58ms, mfu 153.02% step 104000: train loss 2.8393, val loss 2.8539 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 104000: loss 2.8683, time 65516.72ms, mfu 137.83% iter 104100: loss 2.8933, time 436.07ms, mfu 139.73% iter 104200: loss 2.9221, time 440.92ms, mfu 141.28% iter 104300: loss 2.8598, time 441.44ms, mfu 142.65% iter 104400: loss 2.8540, time 438.76ms, mfu 143.98% iter 104500: loss 2.8742, time 441.23ms, mfu 145.09% iter 104600: loss 2.8288, time 442.52ms, mfu 146.04% iter 104700: loss 2.8813, time 440.89ms, mfu 146.95% iter 104800: loss 2.8738, time 440.62ms, mfu 147.79% iter 104900: loss 2.8902, time 439.91ms, mfu 148.56% iter 105000: loss 2.8792, time 441.89ms, mfu 149.19% iter 105100: loss 2.8769, time 441.90ms, mfu 149.75% iter 105200: loss 2.8926, time 441.14ms, mfu 150.29% iter 105300: loss 2.8640, time 438.58ms, mfu 150.86% iter 105400: loss 2.8965, time 440.09ms, mfu 151.32% iter 105500: loss 2.8628, time 443.68ms, mfu 151.61% iter 105600: loss 2.9195, time 442.92ms, mfu 151.90% iter 105700: loss 2.9042, time 440.99ms, mfu 152.22% iter 105800: loss 2.8231, time 439.44ms, mfu 152.57% iter 105900: loss 2.8982, time 442.37ms, mfu 152.78% step 106000: train loss 2.8309, val loss 2.8478 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 106000: loss 2.8691, time 65609.33ms, mfu 137.61% iter 106100: loss 2.8959, time 440.65ms, mfu 139.37% iter 106200: loss 2.9283, time 439.90ms, mfu 140.99% iter 106300: loss 2.9250, time 440.80ms, mfu 142.41% iter 106400: loss 2.8535, time 441.25ms, mfu 143.68% iter 106500: loss 2.8930, time 440.56ms, mfu 144.84% iter 106600: loss 2.9439, time 441.26ms, mfu 145.86% iter 106700: loss 2.8818, time 440.05ms, mfu 146.82% iter 106800: loss 2.8768, time 440.56ms, mfu 147.67% iter 106900: loss 2.8946, time 441.53ms, mfu 148.40% iter 107000: loss 2.9425, time 441.62ms, mfu 149.05% iter 107100: loss 2.9048, time 439.78ms, mfu 149.71% iter 107200: loss 2.9147, time 441.06ms, mfu 150.25% iter 107300: loss 2.9014, time 439.32ms, mfu 150.80% iter 107400: loss 2.8473, time 439.92ms, mfu 151.27% iter 107500: loss 2.8262, time 442.30ms, mfu 151.61% iter 107600: loss 2.8969, time 438.53ms, mfu 152.05% iter 107700: loss 2.9108, time 441.37ms, mfu 152.35% iter 107800: loss 2.8964, time 440.98ms, mfu 152.63% iter 107900: loss 2.8772, time 441.58ms, mfu 152.86% step 108000: train loss 2.8179, val loss 2.8391 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 108000: loss 2.9332, time 65294.80ms, mfu 137.68% iter 108100: loss 2.9012, time 440.60ms, mfu 139.44% iter 108200: loss 2.9023, time 440.68ms, mfu 141.02% iter 108300: loss 2.8371, time 440.97ms, mfu 142.44% iter 108400: loss 2.8989, time 443.29ms, mfu 143.63% iter 108500: loss 2.8963, time 439.92ms, mfu 144.82% iter 108600: loss 2.8682, time 438.51ms, mfu 145.94% iter 108700: loss 2.9415, time 440.50ms, mfu 146.88% iter 108800: loss 2.9263, time 440.74ms, mfu 147.71% iter 108900: loss 2.8211, time 440.43ms, mfu 148.48% iter 109000: loss 2.8775, time 442.16ms, mfu 149.10% iter 109100: loss 2.8797, time 441.23ms, mfu 149.70% iter 109200: loss 2.8588, time 439.91ms, mfu 150.28% iter 109300: loss 2.8618, time 439.47ms, mfu 150.82% iter 109400: loss 2.8647, time 441.18ms, mfu 151.25% iter 109500: loss 2.7807, time 440.38ms, mfu 151.66% iter 109600: loss 2.8184, time 441.41ms, mfu 152.00% iter 109700: loss 2.8834, time 441.13ms, mfu 152.31% iter 109800: loss 2.9062, time 439.78ms, mfu 152.63% iter 109900: loss 2.8404, time 439.23ms, mfu 152.95% step 110000: train loss 2.8121, val loss 2.8344 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 110000: loss 2.9468, time 65452.59ms, mfu 137.76% iter 110100: loss 2.8952, time 440.41ms, mfu 139.52% iter 110200: loss 2.8594, time 442.24ms, mfu 141.04% iter 110300: loss 2.8343, time 441.45ms, mfu 142.43% iter 110400: loss 2.8454, time 440.00ms, mfu 143.74% iter 110500: loss 2.8636, time 441.09ms, mfu 144.88% iter 110600: loss 2.8063, time 440.61ms, mfu 145.92% iter 110700: loss 2.9544, time 443.86ms, mfu 146.74% iter 110800: loss 2.8225, time 441.31ms, mfu 147.57% iter 110900: loss 2.9179, time 442.33ms, mfu 148.28% iter 111000: loss 2.8416, time 440.25ms, mfu 148.99% iter 111100: loss 2.8758, time 442.93ms, mfu 149.54% iter 111200: loss 2.8718, time 440.38ms, mfu 150.13% iter 111300: loss 2.9011, time 439.00ms, mfu 150.70% iter 111400: loss 2.8385, time 442.41ms, mfu 151.09% iter 111500: loss 2.8232, time 444.76ms, mfu 151.37% iter 111600: loss 2.9330, time 441.64ms, mfu 151.72% iter 111700: loss 2.8223, time 440.51ms, mfu 152.08% iter 111800: loss 2.8453, time 440.18ms, mfu 152.42% iter 111900: loss 2.7992, time 440.96ms, mfu 152.69% step 112000: train loss 2.8069, val loss 2.8247 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 112000: loss 2.8589, time 65653.75ms, mfu 137.53% iter 112100: loss 2.8063, time 433.34ms, mfu 139.56% iter 112200: loss 2.8442, time 441.00ms, mfu 141.12% iter 112300: loss 2.7892, time 435.95ms, mfu 142.70% iter 112400: loss 2.8514, time 443.14ms, mfu 143.87% iter 112500: loss 2.8236, time 442.51ms, mfu 144.95% iter 112600: loss 2.8906, time 442.76ms, mfu 145.91% iter 112700: loss 2.8318, time 439.14ms, mfu 146.90% iter 112800: loss 2.8489, time 440.93ms, mfu 147.72% iter 112900: loss 2.8398, time 440.04ms, mfu 148.50% iter 113000: loss 2.8333, time 439.81ms, mfu 149.21% iter 113100: loss 2.8870, time 439.47ms, mfu 149.86% iter 113200: loss 2.8974, time 440.93ms, mfu 150.39% iter 113300: loss 2.8274, time 440.48ms, mfu 150.88% iter 113400: loss 2.9033, time 439.30ms, mfu 151.37% iter 113500: loss 2.8073, time 441.36ms, mfu 151.73% iter 113600: loss 2.8538, time 442.87ms, mfu 152.01% iter 113700: loss 2.8755, time 439.71ms, mfu 152.37% iter 113800: loss 2.8030, time 442.40ms, mfu 152.60% iter 113900: loss 2.8522, time 437.92ms, mfu 152.96% step 114000: train loss 2.7994, val loss 2.8177 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 114000: loss 2.7186, time 65533.58ms, mfu 137.77% iter 114100: loss 2.8597, time 443.73ms, mfu 139.41% iter 114200: loss 2.7810, time 440.44ms, mfu 141.01% iter 114300: loss 2.9057, time 439.41ms, mfu 142.48% iter 114400: loss 2.8561, time 441.48ms, mfu 143.73% iter 114500: loss 2.8427, time 440.72ms, mfu 144.88% iter 114600: loss 2.8784, time 441.59ms, mfu 145.88% iter 114700: loss 2.8019, time 442.30ms, mfu 146.77% iter 114800: loss 2.8505, time 440.93ms, mfu 147.61% iter 114900: loss 2.8134, time 440.00ms, mfu 148.40% iter 115000: loss 2.8616, time 439.42ms, mfu 149.13% iter 115100: loss 2.8605, time 441.81ms, mfu 149.70% iter 115200: loss 2.8474, time 443.10ms, mfu 150.17% iter 115300: loss 2.8767, time 439.68ms, mfu 150.72% iter 115400: loss 2.8730, time 434.14ms, mfu 151.40% iter 115500: loss 2.8667, time 441.00ms, mfu 151.78% iter 115600: loss 2.7871, time 439.12ms, mfu 152.18% iter 115700: loss 2.8599, time 441.86ms, mfu 152.45% iter 115800: loss 2.8359, time 434.86ms, mfu 152.94% iter 115900: loss 2.8446, time 444.20ms, mfu 153.05% step 116000: train loss 2.7871, val loss 2.8135 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 116000: loss 2.8704, time 65761.15ms, mfu 137.85% iter 116100: loss 2.8968, time 438.61ms, mfu 139.66% iter 116200: loss 2.8724, time 441.28ms, mfu 141.20% iter 116300: loss 2.8824, time 441.08ms, mfu 142.59% iter 116400: loss 2.8774, time 435.40ms, mfu 144.05% iter 116500: loss 2.8419, time 441.45ms, mfu 145.14% iter 116600: loss 2.7817, time 442.87ms, mfu 146.08% iter 116700: loss 2.8297, time 442.16ms, mfu 146.94% iter 116800: loss 2.7375, time 439.17ms, mfu 147.83% iter 116900: loss 2.8947, time 440.89ms, mfu 148.56% iter 117000: loss 2.8754, time 440.55ms, mfu 149.24% iter 117100: loss 2.7665, time 439.41ms, mfu 149.89% iter 117200: loss 2.7990, time 444.02ms, mfu 150.31% iter 117300: loss 2.8387, time 441.11ms, mfu 150.79% iter 117400: loss 2.8929, time 442.40ms, mfu 151.17% iter 117500: loss 2.7976, time 442.49ms, mfu 151.52% iter 117600: loss 2.8134, time 439.71ms, mfu 151.93% iter 117700: loss 2.7726, time 440.54ms, mfu 152.27% iter 117800: loss 2.8762, time 443.03ms, mfu 152.48% iter 117900: loss 2.8190, time 440.43ms, mfu 152.77% step 118000: train loss 2.7863, val loss 2.8032 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 118000: loss 2.8672, time 65690.25ms, mfu 137.60% iter 118100: loss 2.8376, time 443.01ms, mfu 139.28% iter 118200: loss 2.8407, time 440.66ms, mfu 140.88% iter 118300: loss 2.8197, time 442.26ms, mfu 142.26% iter 118400: loss 2.7998, time 438.98ms, mfu 143.62% iter 118500: loss 2.8570, time 440.67ms, mfu 144.79% iter 118600: loss 2.8095, time 439.62ms, mfu 145.87% iter 118700: loss 2.8174, time 443.19ms, mfu 146.72% iter 118800: loss 2.8729, time 441.74ms, mfu 147.54% iter 118900: loss 2.8126, time 438.70ms, mfu 148.38% iter 119000: loss 2.7807, time 441.84ms, mfu 149.03% iter 119100: loss 2.8243, time 441.92ms, mfu 149.61% iter 119200: loss 2.8336, time 440.16ms, mfu 150.19% iter 119300: loss 2.8085, time 439.77ms, mfu 150.73% iter 119400: loss 2.8617, time 441.80ms, mfu 151.14% iter 119500: loss 2.8478, time 439.28ms, mfu 151.61% iter 119600: loss 2.8275, time 443.04ms, mfu 151.89% iter 119700: loss 2.8272, time 441.46ms, mfu 152.20% iter 119800: loss 2.8077, time 439.17ms, mfu 152.56% iter 119900: loss 2.8324, time 440.91ms, mfu 152.82% step 120000: train loss 2.7764, val loss 2.7948 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 120000: loss 2.8556, time 65414.84ms, mfu 137.64% iter 120100: loss 2.8183, time 439.73ms, mfu 139.44% iter 120200: loss 2.7740, time 441.18ms, mfu 141.00% iter 120300: loss 2.7821, time 440.20ms, mfu 142.45% iter 120400: loss 2.7554, time 433.85ms, mfu 143.97% iter 120500: loss 2.8011, time 441.45ms, mfu 145.07% iter 120600: loss 2.8431, time 444.78ms, mfu 145.95% iter 120700: loss 2.8372, time 441.21ms, mfu 146.86% iter 120800: loss 2.8175, time 432.67ms, mfu 147.99% iter 120900: loss 2.7497, time 442.54ms, mfu 148.65% iter 121000: loss 2.7956, time 440.99ms, mfu 149.30% iter 121100: loss 2.8291, time 440.01ms, mfu 149.92% iter 121200: loss 2.7597, time 439.76ms, mfu 150.49% iter 121300: loss 2.7891, time 442.61ms, mfu 150.90% iter 121400: loss 2.8242, time 439.47ms, mfu 151.38% iter 121500: loss 2.8993, time 438.32ms, mfu 151.85% iter 121600: loss 2.8140, time 438.95ms, mfu 152.25% iter 121700: loss 2.7881, time 439.68ms, mfu 152.59% iter 121800: loss 2.7911, time 441.33ms, mfu 152.83% iter 121900: loss 2.8824, time 441.26ms, mfu 153.05% step 122000: train loss 2.7670, val loss 2.7888 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 122000: loss 2.8614, time 65505.33ms, mfu 137.85% iter 122100: loss 2.8434, time 438.94ms, mfu 139.66% iter 122200: loss 2.7899, time 439.73ms, mfu 141.25% iter 122300: loss 2.8125, time 440.51ms, mfu 142.66% iter 122400: loss 2.7955, time 441.64ms, mfu 143.88% iter 122500: loss 2.7594, time 444.32ms, mfu 144.89% iter 122600: loss 2.8907, time 443.27ms, mfu 145.84% iter 122700: loss 2.8648, time 440.72ms, mfu 146.78% iter 122800: loss 2.8149, time 441.06ms, mfu 147.61% iter 122900: loss 2.8159, time 439.95ms, mfu 148.41% iter 123000: loss 2.8114, time 445.07ms, mfu 148.94% iter 123100: loss 2.8256, time 441.07ms, mfu 149.56% iter 123200: loss 2.8304, time 443.19ms, mfu 150.04% iter 123300: loss 2.7887, time 440.80ms, mfu 150.56% iter 123400: loss 2.7741, time 440.07ms, mfu 151.05% iter 123500: loss 2.8599, time 441.63ms, mfu 151.44% iter 123600: loss 2.7530, time 442.85ms, mfu 151.74% iter 123700: loss 2.7967, time 440.83ms, mfu 152.09% iter 123800: loss 2.8116, time 441.59ms, mfu 152.37% iter 123900: loss 2.8388, time 442.57ms, mfu 152.60% step 124000: train loss 2.7536, val loss 2.7844 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 124000: loss 2.7749, time 65548.67ms, mfu 137.44% iter 124100: loss 2.8764, time 437.81ms, mfu 139.32% iter 124200: loss 2.8399, time 439.51ms, mfu 140.96% iter 124300: loss 2.8156, time 441.41ms, mfu 142.36% iter 124400: loss 2.8278, time 436.19ms, mfu 143.81% iter 124500: loss 2.7536, time 440.73ms, mfu 144.96% iter 124600: loss 2.7980, time 438.64ms, mfu 146.06% iter 124700: loss 2.8289, time 443.08ms, mfu 146.90% iter 124800: loss 2.7836, time 440.77ms, mfu 147.73% iter 124900: loss 2.7926, time 435.95ms, mfu 148.65% iter 125000: loss 2.8268, time 441.68ms, mfu 149.28% iter 125100: loss 2.7899, time 439.24ms, mfu 149.93% iter 125200: loss 2.7296, time 439.46ms, mfu 150.50% iter 125300: loss 2.8671, time 440.25ms, mfu 150.99% iter 125400: loss 2.8313, time 442.21ms, mfu 151.37% iter 125500: loss 2.8739, time 440.14ms, mfu 151.77% iter 125600: loss 2.8147, time 441.52ms, mfu 152.09% iter 125700: loss 2.8039, time 440.30ms, mfu 152.42% iter 125800: loss 2.8488, time 438.91ms, mfu 152.77% iter 125900: loss 2.7709, time 439.64ms, mfu 153.06% step 126000: train loss 2.7582, val loss 2.7787 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 126000: loss 2.8225, time 65705.47ms, mfu 137.85% iter 126100: loss 2.7964, time 439.59ms, mfu 139.63% iter 126200: loss 2.7949, time 434.45ms, mfu 141.42% iter 126300: loss 2.7715, time 441.63ms, mfu 142.77% iter 126400: loss 2.8160, time 441.32ms, mfu 144.00% iter 126500: loss 2.8389, time 441.04ms, mfu 145.11% iter 126600: loss 2.7874, time 442.28ms, mfu 146.07% iter 126700: loss 2.7688, time 440.45ms, mfu 147.00% iter 126800: loss 2.7711, time 442.12ms, mfu 147.77% iter 126900: loss 2.7036, time 441.05ms, mfu 148.51% iter 127000: loss 2.8151, time 440.12ms, mfu 149.20% iter 127100: loss 2.8384, time 441.08ms, mfu 149.79% iter 127200: loss 2.7610, time 440.01ms, mfu 150.36% iter 127300: loss 2.7682, time 441.71ms, mfu 150.82% iter 127400: loss 2.8408, time 440.31ms, mfu 151.28% iter 127500: loss 2.8193, time 442.28ms, mfu 151.62% iter 127600: loss 2.8509, time 442.31ms, mfu 151.92% iter 127700: loss 2.8121, time 440.67ms, mfu 152.26% iter 127800: loss 2.7416, time 442.17ms, mfu 152.51% iter 127900: loss 2.8069, time 444.85ms, mfu 152.64% step 128000: train loss 2.7548, val loss 2.7764 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 128000: loss 2.7691, time 65564.38ms, mfu 137.48% iter 128100: loss 2.7954, time 440.80ms, mfu 139.25% iter 128200: loss 2.8105, time 440.43ms, mfu 140.86% iter 128300: loss 2.7383, time 441.48ms, mfu 142.27% iter 128400: loss 2.8281, time 441.35ms, mfu 143.55% iter 128500: loss 2.7763, time 443.87ms, mfu 144.61% iter 128600: loss 2.8198, time 441.87ms, mfu 145.63% iter 128700: loss 2.8018, time 439.97ms, mfu 146.62% iter 128800: loss 2.8550, time 443.92ms, mfu 147.37% iter 128900: loss 2.7483, time 441.07ms, mfu 148.15% iter 129000: loss 2.7486, time 442.83ms, mfu 148.78% iter 129100: loss 2.8854, time 442.26ms, mfu 149.37% iter 129200: loss 2.7922, time 443.30ms, mfu 149.87% iter 129300: loss 2.7137, time 446.38ms, mfu 150.21% iter 129400: loss 2.8317, time 439.50ms, mfu 150.76% iter 129500: loss 2.7223, time 441.18ms, mfu 151.19% iter 129600: loss 2.8584, time 439.49ms, mfu 151.64% iter 129700: loss 2.8009, time 442.30ms, mfu 151.95% iter 129800: loss 2.9103, time 441.96ms, mfu 152.23% iter 129900: loss 2.7666, time 442.79ms, mfu 152.46% step 130000: train loss 2.7483, val loss 2.7697 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 130000: loss 2.8124, time 65799.62ms, mfu 137.32% iter 130100: loss 2.8043, time 441.01ms, mfu 139.10% iter 130200: loss 2.7715, time 440.01ms, mfu 140.74% iter 130300: loss 2.8107, time 439.34ms, mfu 142.24% iter 130400: loss 2.7556, time 442.41ms, mfu 143.48% iter 130500: loss 2.7954, time 441.54ms, mfu 144.63% iter 130600: loss 2.8161, time 440.13ms, mfu 145.71% iter 130700: loss 2.7706, time 441.41ms, mfu 146.64% iter 130800: loss 2.8254, time 440.91ms, mfu 147.49% iter 130900: loss 2.7722, time 444.19ms, mfu 148.15% iter 131000: loss 2.7176, time 441.95ms, mfu 148.82% iter 131100: loss 2.7961, time 439.31ms, mfu 149.51% iter 131200: loss 2.7558, time 440.95ms, mfu 150.07% iter 131300: loss 2.8392, time 441.34ms, mfu 150.57% iter 131400: loss 2.8425, time 442.77ms, mfu 150.97% iter 131500: loss 2.7823, time 439.03ms, mfu 151.45% iter 131600: loss 2.6663, time 438.23ms, mfu 151.92% iter 131700: loss 2.7957, time 443.53ms, mfu 152.15% iter 131800: loss 2.7258, time 443.30ms, mfu 152.37% iter 131900: loss 2.8340, time 441.09ms, mfu 152.65% step 132000: train loss 2.7441, val loss 2.7680 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 132000: loss 2.7797, time 65457.70ms, mfu 137.49% iter 132100: loss 2.8091, time 441.48ms, mfu 139.24% iter 132200: loss 2.7777, time 441.62ms, mfu 140.81% iter 132300: loss 2.7519, time 442.63ms, mfu 142.18% iter 132400: loss 2.8004, time 440.96ms, mfu 143.48% iter 132500: loss 2.7831, time 439.82ms, mfu 144.69% iter 132600: loss 2.7491, time 442.04ms, mfu 145.70% iter 132700: loss 2.7711, time 441.93ms, mfu 146.61% iter 132800: loss 2.7886, time 441.62ms, mfu 147.44% iter 132900: loss 2.7077, time 438.57ms, mfu 148.30% iter 133000: loss 2.7744, time 442.46ms, mfu 148.93% iter 133100: loss 2.8471, time 443.40ms, mfu 149.47% iter 133200: loss 2.7616, time 439.65ms, mfu 150.09% iter 133300: loss 2.8758, time 440.86ms, mfu 150.60% iter 133400: loss 2.7872, time 440.40ms, mfu 151.07% iter 133500: loss 2.8764, time 439.62ms, mfu 151.53% iter 133600: loss 2.7800, time 441.22ms, mfu 151.88% iter 133700: loss 2.8041, time 444.59ms, mfu 152.08% iter 133800: loss 2.8194, time 441.35ms, mfu 152.38% iter 133900: loss 2.7547, time 439.98ms, mfu 152.69% step 134000: train loss 2.7390, val loss 2.7650 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 134000: loss 2.7572, time 65663.64ms, mfu 137.53% iter 134100: loss 2.7551, time 440.54ms, mfu 139.31% iter 134200: loss 2.7672, time 439.60ms, mfu 140.94% iter 134300: loss 2.8065, time 442.56ms, mfu 142.31% iter 134400: loss 2.7587, time 441.61ms, mfu 143.57% iter 134500: loss 2.7616, time 441.45ms, mfu 144.71% iter 134600: loss 2.7910, time 435.63ms, mfu 145.94% iter 134700: loss 2.8514, time 440.35ms, mfu 146.89% iter 134800: loss 2.7414, time 441.74ms, mfu 147.69% iter 134900: loss 2.8629, time 440.16ms, mfu 148.46% iter 135000: loss 2.8098, time 439.35ms, mfu 149.19% iter 135100: loss 2.8385, time 443.71ms, mfu 149.69% iter 135200: loss 2.7048, time 440.65ms, mfu 150.25% iter 135300: loss 2.7254, time 444.26ms, mfu 150.63% iter 135400: loss 2.6914, time 441.42ms, mfu 151.06% iter 135500: loss 2.7797, time 441.67ms, mfu 151.45% iter 135600: loss 2.8389, time 442.64ms, mfu 151.76% iter 135700: loss 2.8208, time 441.67ms, mfu 152.08% iter 135800: loss 2.7976, time 441.08ms, mfu 152.38% iter 135900: loss 2.7482, time 440.82ms, mfu 152.66% step 136000: train loss 2.7352, val loss 2.7654 saving checkpoint to out-wiki-gutenberg-530m-1ep iter 136000: loss 2.7647, time 65743.26ms, mfu 137.50% iter 136100: loss 2.8031, time 440.62ms, mfu 139.28% iter 136200: loss 2.7676, time 440.96ms, mfu 140.87% iter 136300: loss 2.8472, time 441.35ms, mfu 142.28% iter 136400: loss 2.7656, time 437.28ms, mfu 143.70% iter 136500: loss 2.8012, time 441.28ms, mfu 144.84%