| {"event": "run_start", "step": 4400, "max_steps": 5764, "model_params_M": 9.853, "eff_batch": 1024, "amp_dtype": "torch.float16", "attn_backend": "chunked", "started_at": "2026-06-21T05:02:45.037316"} |
| {"step": 4420, "epoch": 0.767, "loss": 2.4509, "ppl": 11.599, "lr": 6.8262e-05, "lr_sched": "cosine", "grad_norm": 0.7006, "tokens": 4630179840, "tok_s": 69873.9, "elapsed_s": 371.1, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4440, "epoch": 0.77, "loss": 2.40301, "ppl": 11.056, "lr": 6.7187e-05, "lr_sched": "cosine", "grad_norm": 0.7597, "tokens": 4651130880, "tok_s": 70266.3, "elapsed_s": 669.8, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4460, "epoch": 0.774, "loss": 2.42283, "ppl": 11.278, "lr": 6.6126e-05, "lr_sched": "cosine", "grad_norm": 0.6205, "tokens": 4672081920, "tok_s": 70267.5, "elapsed_s": 968.0, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4480, "epoch": 0.777, "loss": 2.37378, "ppl": 10.738, "lr": 6.5077e-05, "lr_sched": "cosine", "grad_norm": 0.6263, "tokens": 4693032960, "tok_s": 70255.0, "elapsed_s": 1266.3, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4500, "epoch": 0.781, "loss": 2.33904, "ppl": 10.371, "lr": 6.4042e-05, "lr_sched": "cosine", "grad_norm": 0.5782, "tokens": 4713984000, "tok_s": 70337.8, "elapsed_s": 1564.5, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4520, "epoch": 0.784, "loss": 2.37358, "ppl": 10.736, "lr": 6.302e-05, "lr_sched": "cosine", "grad_norm": 0.6556, "tokens": 4734935040, "tok_s": 70180.5, "elapsed_s": 1863.3, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4540, "epoch": 0.788, "loss": 2.31746, "ppl": 10.15, "lr": 6.2011e-05, "lr_sched": "cosine", "grad_norm": 0.631, "tokens": 4755886080, "tok_s": 70129.5, "elapsed_s": 2161.7, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4560, "epoch": 0.791, "loss": 2.38647, "ppl": 10.875, "lr": 6.1016e-05, "lr_sched": "cosine", "grad_norm": 0.7045, "tokens": 4776837120, "tok_s": 70062.0, "elapsed_s": 2460.5, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4580, "epoch": 0.795, "loss": 2.32614, "ppl": 10.238, "lr": 6.0035e-05, "lr_sched": "cosine", "grad_norm": 0.6115, "tokens": 4797788160, "tok_s": 69954.5, "elapsed_s": 2759.5, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4600, "epoch": 0.798, "loss": 2.38817, "ppl": 10.894, "lr": 5.9067e-05, "lr_sched": "cosine", "grad_norm": 0.6463, "tokens": 4818739200, "tok_s": 69957.6, "elapsed_s": 3058.6, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4620, "epoch": 0.802, "loss": 2.33083, "ppl": 10.286, "lr": 5.8114e-05, "lr_sched": "cosine", "grad_norm": 0.6187, "tokens": 4839690240, "tok_s": 70081.9, "elapsed_s": 3357.7, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4640, "epoch": 0.805, "loss": 2.39559, "ppl": 10.975, "lr": 5.7174e-05, "lr_sched": "cosine", "grad_norm": 0.5444, "tokens": 4860641280, "tok_s": 69969.9, "elapsed_s": 3656.8, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4660, "epoch": 0.808, "loss": 2.31579, "ppl": 10.133, "lr": 5.6249e-05, "lr_sched": "cosine", "grad_norm": 0.5401, "tokens": 4881592320, "tok_s": 69893.5, "elapsed_s": 3956.2, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4680, "epoch": 0.812, "loss": 2.40544, "ppl": 11.083, "lr": 5.5338e-05, "lr_sched": "cosine", "grad_norm": 0.7227, "tokens": 4902543360, "tok_s": 70336.4, "elapsed_s": 4255.1, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4700, "epoch": 0.815, "loss": 2.31736, "ppl": 10.149, "lr": 5.4442e-05, "lr_sched": "cosine", "grad_norm": 0.5292, "tokens": 4923494400, "tok_s": 70013.3, "elapsed_s": 4553.4, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4720, "epoch": 0.819, "loss": 2.39576, "ppl": 10.977, "lr": 5.356e-05, "lr_sched": "cosine", "grad_norm": 0.8587, "tokens": 4944445440, "tok_s": 69999.4, "elapsed_s": 4853.0, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4740, "epoch": 0.822, "loss": 2.29127, "ppl": 9.888, "lr": 5.2692e-05, "lr_sched": "cosine", "grad_norm": 0.6399, "tokens": 4965396480, "tok_s": 69954.2, "elapsed_s": 5152.5, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4760, "epoch": 0.826, "loss": 2.36583, "ppl": 10.653, "lr": 5.184e-05, "lr_sched": "cosine", "grad_norm": 0.6819, "tokens": 4986347520, "tok_s": 69983.4, "elapsed_s": 5452.1, "vram_gb": 0.67, "ram_pct": 18.1, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4780, "epoch": 0.829, "loss": 2.35854, "ppl": 10.576, "lr": 5.1002e-05, "lr_sched": "cosine", "grad_norm": 0.5546, "tokens": 5007298560, "tok_s": 69955.3, "elapsed_s": 5751.2, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4800, "epoch": 0.833, "loss": 2.30045, "ppl": 9.979, "lr": 5.018e-05, "lr_sched": "cosine", "grad_norm": 1.2005, "tokens": 5028249600, "tok_s": 70165.0, "elapsed_s": 6050.2, "vram_gb": 0.67, "ram_pct": 18.0, "disk_free_gb": 20.94, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4800, "epoch": 0.833, "val_loss": 2.40489, "val_ppl": 11.077, "best_val_ppl": 11.077, "is_best": true} |
| {"step": 4820, "epoch": 0.836, "loss": 2.24408, "ppl": 9.432, "lr": 4.9373e-05, "lr_sched": "cosine", "grad_norm": 0.7609, "tokens": 5049200640, "tok_s": 70208.3, "elapsed_s": 7406.8, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4840, "epoch": 0.84, "loss": 2.13003, "ppl": 8.415, "lr": 4.858e-05, "lr_sched": "cosine", "grad_norm": 0.7302, "tokens": 5070151680, "tok_s": 70044.1, "elapsed_s": 7705.9, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4860, "epoch": 0.843, "loss": 2.18098, "ppl": 8.855, "lr": 4.7804e-05, "lr_sched": "cosine", "grad_norm": 0.8219, "tokens": 5091102720, "tok_s": 70063.3, "elapsed_s": 8005.2, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4880, "epoch": 0.847, "loss": 2.15663, "ppl": 8.642, "lr": 4.7042e-05, "lr_sched": "cosine", "grad_norm": 0.7249, "tokens": 5112053760, "tok_s": 70298.9, "elapsed_s": 8303.9, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4900, "epoch": 0.85, "loss": 2.08825, "ppl": 8.071, "lr": 4.6296e-05, "lr_sched": "cosine", "grad_norm": 0.7222, "tokens": 5133004800, "tok_s": 70354.6, "elapsed_s": 8601.8, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4920, "epoch": 0.854, "loss": 2.01568, "ppl": 7.506, "lr": 4.5566e-05, "lr_sched": "cosine", "grad_norm": 0.8498, "tokens": 5153955840, "tok_s": 70152.1, "elapsed_s": 8900.3, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4940, "epoch": 0.857, "loss": 2.05463, "ppl": 7.804, "lr": 4.4852e-05, "lr_sched": "cosine", "grad_norm": 0.8497, "tokens": 5174906880, "tok_s": 70073.4, "elapsed_s": 9199.1, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4960, "epoch": 0.861, "loss": 2.10032, "ppl": 8.169, "lr": 4.4153e-05, "lr_sched": "cosine", "grad_norm": 0.8044, "tokens": 5195857920, "tok_s": 70284.1, "elapsed_s": 9497.0, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 4980, "epoch": 0.864, "loss": 2.10869, "ppl": 8.237, "lr": 4.347e-05, "lr_sched": "cosine", "grad_norm": 0.9414, "tokens": 5216808960, "tok_s": 70355.1, "elapsed_s": 9795.5, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5000, "epoch": 0.867, "loss": 2.08726, "ppl": 8.063, "lr": 4.2804e-05, "lr_sched": "cosine", "grad_norm": 0.7008, "tokens": 5237760000, "tok_s": 70316.3, "elapsed_s": 10093.4, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5020, "epoch": 0.871, "loss": 2.20945, "ppl": 9.111, "lr": 4.2153e-05, "lr_sched": "cosine", "grad_norm": 0.779, "tokens": 5258711040, "tok_s": 70159.3, "elapsed_s": 10392.0, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5040, "epoch": 0.874, "loss": 2.15376, "ppl": 8.617, "lr": 4.1518e-05, "lr_sched": "cosine", "grad_norm": 0.6971, "tokens": 5279662080, "tok_s": 70059.2, "elapsed_s": 10691.0, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5060, "epoch": 0.878, "loss": 2.10978, "ppl": 8.246, "lr": 4.09e-05, "lr_sched": "cosine", "grad_norm": 0.8154, "tokens": 5300613120, "tok_s": 70298.4, "elapsed_s": 10989.5, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5080, "epoch": 0.881, "loss": 2.15794, "ppl": 8.653, "lr": 4.0299e-05, "lr_sched": "cosine", "grad_norm": 0.674, "tokens": 5321564160, "tok_s": 70371.1, "elapsed_s": 11287.6, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5100, "epoch": 0.885, "loss": 2.08483, "ppl": 8.043, "lr": 3.9713e-05, "lr_sched": "cosine", "grad_norm": 0.6787, "tokens": 5342515200, "tok_s": 70255.2, "elapsed_s": 11585.7, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5120, "epoch": 0.888, "loss": 2.05963, "ppl": 7.843, "lr": 3.9144e-05, "lr_sched": "cosine", "grad_norm": 0.8739, "tokens": 5363466240, "tok_s": 70258.3, "elapsed_s": 11884.1, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5140, "epoch": 0.892, "loss": 2.10277, "ppl": 8.189, "lr": 3.8592e-05, "lr_sched": "cosine", "grad_norm": 0.8421, "tokens": 5384417280, "tok_s": 70155.5, "elapsed_s": 12182.5, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5160, "epoch": 0.895, "loss": 2.06721, "ppl": 7.903, "lr": 3.8056e-05, "lr_sched": "cosine", "grad_norm": 0.8163, "tokens": 5405368320, "tok_s": 70230.0, "elapsed_s": 12480.7, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5180, "epoch": 0.899, "loss": 2.11915, "ppl": 8.324, "lr": 3.7537e-05, "lr_sched": "cosine", "grad_norm": 1.0566, "tokens": 5426319360, "tok_s": 70364.1, "elapsed_s": 12778.8, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5200, "epoch": 0.902, "loss": 2.43887, "ppl": 11.46, "lr": 3.7035e-05, "lr_sched": "cosine", "grad_norm": 0.6095, "tokens": 5447270400, "tok_s": 70078.7, "elapsed_s": 13077.6, "vram_gb": 0.67, "ram_pct": 20.2, "disk_free_gb": 20.62, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5200, "epoch": 0.902, "val_loss": 2.40036, "val_ppl": 11.027, "best_val_ppl": 11.027, "is_best": true} |
| {"step": 5220, "epoch": 0.906, "loss": 2.40566, "ppl": 11.086, "lr": 3.655e-05, "lr_sched": "cosine", "grad_norm": 0.6325, "tokens": 5468221440, "tok_s": 70295.8, "elapsed_s": 14392.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5240, "epoch": 0.909, "loss": 2.45154, "ppl": 11.606, "lr": 3.6082e-05, "lr_sched": "cosine", "grad_norm": 0.6922, "tokens": 5489172480, "tok_s": 70161.4, "elapsed_s": 14691.5, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5260, "epoch": 0.913, "loss": 2.36491, "ppl": 10.643, "lr": 3.563e-05, "lr_sched": "cosine", "grad_norm": 0.6546, "tokens": 5510123520, "tok_s": 70177.7, "elapsed_s": 14990.8, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5280, "epoch": 0.916, "loss": 2.37539, "ppl": 10.755, "lr": 3.5196e-05, "lr_sched": "cosine", "grad_norm": 0.613, "tokens": 5531074560, "tok_s": 70229.2, "elapsed_s": 15290.0, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5300, "epoch": 0.92, "loss": 2.371, "ppl": 10.708, "lr": 3.4779e-05, "lr_sched": "cosine", "grad_norm": 0.6257, "tokens": 5552025600, "tok_s": 70344.5, "elapsed_s": 15588.8, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5320, "epoch": 0.923, "loss": 2.38127, "ppl": 10.819, "lr": 3.4379e-05, "lr_sched": "cosine", "grad_norm": 0.6278, "tokens": 5572976640, "tok_s": 69986.1, "elapsed_s": 15887.6, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5340, "epoch": 0.926, "loss": 2.37596, "ppl": 10.761, "lr": 3.3996e-05, "lr_sched": "cosine", "grad_norm": 0.665, "tokens": 5593927680, "tok_s": 69974.3, "elapsed_s": 16186.7, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5360, "epoch": 0.93, "loss": 2.3663, "ppl": 10.658, "lr": 3.363e-05, "lr_sched": "cosine", "grad_norm": 0.7365, "tokens": 5614878720, "tok_s": 69947.0, "elapsed_s": 16485.8, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5380, "epoch": 0.933, "loss": 2.39503, "ppl": 10.969, "lr": 3.3282e-05, "lr_sched": "cosine", "grad_norm": 0.5692, "tokens": 5635829760, "tok_s": 70147.1, "elapsed_s": 16785.2, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5400, "epoch": 0.937, "loss": 2.59303, "ppl": 13.37, "lr": 3.2951e-05, "lr_sched": "cosine", "grad_norm": 1.2027, "tokens": 5656780800, "tok_s": 70371.4, "elapsed_s": 17083.8, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5420, "epoch": 0.94, "loss": 2.54659, "ppl": 12.763, "lr": 3.2638e-05, "lr_sched": "cosine", "grad_norm": 0.4686, "tokens": 5677731840, "tok_s": 69953.3, "elapsed_s": 17382.1, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5440, "epoch": 0.944, "loss": 2.5106, "ppl": 12.312, "lr": 3.2342e-05, "lr_sched": "cosine", "grad_norm": 0.506, "tokens": 5698682880, "tok_s": 69931.3, "elapsed_s": 17681.0, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5460, "epoch": 0.947, "loss": 2.51406, "ppl": 12.355, "lr": 3.2063e-05, "lr_sched": "cosine", "grad_norm": 0.4599, "tokens": 5719633920, "tok_s": 70216.1, "elapsed_s": 17979.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5480, "epoch": 0.951, "loss": 2.50314, "ppl": 12.221, "lr": 3.1802e-05, "lr_sched": "cosine", "grad_norm": 0.4518, "tokens": 5740584960, "tok_s": 70086.1, "elapsed_s": 18278.5, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5500, "epoch": 0.954, "loss": 2.49455, "ppl": 12.116, "lr": 3.1558e-05, "lr_sched": "cosine", "grad_norm": 0.5578, "tokens": 5761536000, "tok_s": 70268.7, "elapsed_s": 18576.8, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5520, "epoch": 0.958, "loss": 2.37941, "ppl": 10.799, "lr": 3.1332e-05, "lr_sched": "cosine", "grad_norm": 0.6903, "tokens": 5782487040, "tok_s": 70246.8, "elapsed_s": 18875.8, "vram_gb": 0.67, "ram_pct": 20.3, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5540, "epoch": 0.961, "loss": 2.36308, "ppl": 10.624, "lr": 3.1124e-05, "lr_sched": "cosine", "grad_norm": 0.6258, "tokens": 5803438080, "tok_s": 70102.0, "elapsed_s": 19173.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5560, "epoch": 0.965, "loss": 2.37222, "ppl": 10.721, "lr": 3.0933e-05, "lr_sched": "cosine", "grad_norm": 0.5562, "tokens": 5824389120, "tok_s": 70309.8, "elapsed_s": 19472.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5580, "epoch": 0.968, "loss": 2.10531, "ppl": 8.21, "lr": 3.076e-05, "lr_sched": "cosine", "grad_norm": 1.4699, "tokens": 5845340160, "tok_s": 70154.2, "elapsed_s": 19771.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5600, "epoch": 0.972, "loss": 2.01128, "ppl": 7.473, "lr": 3.0605e-05, "lr_sched": "cosine", "grad_norm": 0.5457, "tokens": 5866291200, "tok_s": 70071.4, "elapsed_s": 20071.0, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5600, "epoch": 0.972, "val_loss": 2.46951, "val_ppl": 11.817, "best_val_ppl": 11.027, "is_best": false} |
| {"step": 5620, "epoch": 0.975, "loss": 1.96707, "ppl": 7.15, "lr": 3.0467e-05, "lr_sched": "cosine", "grad_norm": 0.4761, "tokens": 5887242240, "tok_s": 70324.4, "elapsed_s": 21385.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5640, "epoch": 0.978, "loss": 1.9579, "ppl": 7.084, "lr": 3.0347e-05, "lr_sched": "cosine", "grad_norm": 0.5355, "tokens": 5908193280, "tok_s": 70032.3, "elapsed_s": 21684.9, "vram_gb": 0.67, "ram_pct": 20.5, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5660, "epoch": 0.982, "loss": 1.95738, "ppl": 7.081, "lr": 3.0245e-05, "lr_sched": "cosine", "grad_norm": 0.5058, "tokens": 5929144320, "tok_s": 70148.6, "elapsed_s": 21983.9, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5680, "epoch": 0.985, "loss": 1.94553, "ppl": 6.997, "lr": 3.0161e-05, "lr_sched": "cosine", "grad_norm": 0.4914, "tokens": 5950095360, "tok_s": 70081.6, "elapsed_s": 22282.5, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5700, "epoch": 0.989, "loss": 1.91771, "ppl": 6.805, "lr": 3.0094e-05, "lr_sched": "cosine", "grad_norm": 0.5106, "tokens": 5971046400, "tok_s": 70202.8, "elapsed_s": 22581.5, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5720, "epoch": 0.992, "loss": 1.92384, "ppl": 6.847, "lr": 3.0045e-05, "lr_sched": "cosine", "grad_norm": 0.522, "tokens": 5991997440, "tok_s": 70349.5, "elapsed_s": 22880.0, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5740, "epoch": 0.996, "loss": 1.92147, "ppl": 6.831, "lr": 3.0014e-05, "lr_sched": "cosine", "grad_norm": 0.5118, "tokens": 6012948480, "tok_s": 69909.8, "elapsed_s": 23178.6, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
| {"step": 5760, "epoch": 0.999, "loss": 1.92361, "ppl": 6.846, "lr": 3.0001e-05, "lr_sched": "cosine", "grad_norm": 0.5959, "tokens": 6033899520, "tok_s": 70326.1, "elapsed_s": 23477.5, "vram_gb": 0.67, "ram_pct": 20.4, "disk_free_gb": 20.5, "attn_backend": "chunked", "amp_dtype": "torch.float16"} |
|
|