{ "step": 300000, "recipe": { "vocab_size": 8192, "num_hidden_layers": 12, "hidden_size": 128, "hidden_size_per_layer_input": 16, "intermediate_size": 384, "num_attention_heads": 4, "num_key_value_heads": 1, "head_dim": 32, "global_head_dim": 32, "sliding_window": 128, "max_position_embeddings": 2048, "full_attention_every": 4 }, "args": { "output_dir": "runs\\tiny-gemma4-v3", "data_dir": "data", "tokenizer_dir": "runs\\tiny-gemma4-v3\\tokenizer", "recipe": "v3", "data_mode": null, "max_steps": 300000, "batch_size": 32, "block_size": 256, "gradient_accumulation_steps": 1, "learning_rate": 0.0003, "min_learning_rate": 3e-05, "warmup_steps": 1000, "weight_decay": 0.1, "grad_clip": 1.0, "save_every": 10000, "eval_every": 1000, "eval_batches": 50, "train_char_limit": null, "valid_char_limit": null, "num_workers": 0, "seed": 1337, "resume": false, "compile": false }, "time": "2026-07-10 16:09:40" }