| { | |
| "architectures": [ | |
| "ChessForCausalLM" | |
| ], | |
| "attention_type": "gqa", | |
| "bos_token_id": 1, | |
| "dropout": 0.1, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "ffn_type": "swiglu", | |
| "layer_norm_epsilon": 1e-05, | |
| "legal_loss_weight": 0.0, | |
| "model_type": "chess_transformer", | |
| "n_ctx": 256, | |
| "n_embd": 136, | |
| "n_head": 4, | |
| "n_inner": 192, | |
| "n_kv_heads": 2, | |
| "n_layer": 7, | |
| "pad_token_id": 0, | |
| "pos_encoding": "rope", | |
| "rope_theta": 10000.0, | |
| "tie_weights": true, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.52.4", | |
| "vocab_size": 1682 | |
| } | |