| { | |
| "architectures": [ | |
| "LoopedForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_babyloop.BabyloopConfig", | |
| "AutoModelForCausalLM": "modeling_babyloop.LoopedForCausalLM", | |
| "AutoModel": "modeling_babyloop.LoopedModel" | |
| }, | |
| "bias": false, | |
| "bos_token_id": 1, | |
| "d_model": 768, | |
| "dropout": 0.0, | |
| "eos_token_id": 2, | |
| "ffn_hidden": 2048, | |
| "fusion": null, | |
| "hidden_size": 768, | |
| "inject_input": true, | |
| "k": 4, | |
| "max_position_embeddings": 512, | |
| "max_seq_len": 512, | |
| "model_type": "babyloop", | |
| "n_coda": 0, | |
| "n_core": 12, | |
| "n_heads": 12, | |
| "n_layers": 12, | |
| "n_prelude": 0, | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "pad_token_id": 0, | |
| "rms_eps": 1e-05, | |
| "rope_base": 10000.0, | |
| "tie_embeddings": true, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.51.3", | |
| "vocab_size": 16000 | |
| } |