{ "activation_checkpointing": true, "depth": 8, "eval_batch_size": 4, "img_size": 512, "mfu_percent": null, "num_params": 77575312, "num_steps": 3, "patch_size": 14, "peak_vram_mb": 3815.390625, "smoke_test": true, "tokenizer_dataset": "multivision", "total_seconds": 87.76879477500916, "total_tokens": 1572864, "train_batch_size": 4, "train_dataset": "multivision", "training_seconds": 0.0, "use_moe": false, "use_vision": true, "val_bpb": 2.706730086299882, "vision_dim": 128, "vision_heads": 4, "vision_layers": 4 }