{ "activation_checkpointing": true, "dataset": "tinystories", "depth": 8, "eval_batch_size": 4, "mfu_percent": 10.217708574629848, "num_params": 75498000, "num_steps": 480, "peak_vram_mb": 3033.13671875, "smoke_test": false, "total_seconds": 7569.189307928085, "total_tokens": 251658240, "train_batch_size": 4, "training_seconds": 7208.875112533569, "val_bpb": 0.3975104956810562 }