{ "vocab_size": 7829, "max_seq_len": 256, "n_layers": 6, "n_heads": 8, "d_model": 256, "d_ff": 1024, "dropout": 0.1, "pad_id": 0, "bos_id": 1, "eos_id": 2, "sep_id": 3, "epochs": 5, "batch_size": 64, "grad_accum": 4, "lr": 0.0003, "warmup_steps": 500, "weight_decay": 0.01, "grad_clip": 1.0, "max_new_tokens": 250, "val_split": 0.05, "seed": 42, "model_type": "farmbot-gpt", "architecture": "decoder-only transformer", "training_examples": 377500, "best_val_loss": 0.0010096110614960102, "status": "draft-365k" }