| { | |
| "architectures": [ | |
| "HawkForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "modeling_hawk.HawkConfig", | |
| "AutoModel": "modeling_hawk.HawkModel", | |
| "AutoModelForCausalLM": "modeling_hawk.HawkForCausalLM", | |
| "AutoModelForSequenceClassification": "modeling_hawk.HawkForSequenceClassification" | |
| }, | |
| "bos_token_id": 1, | |
| "conv_kernel": 4, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "max_position_embeddings": 1024, | |
| "mlp_expansion": 2, | |
| "model_type": "hawk_rglru", | |
| "n_embd": 576, | |
| "n_layer": 18, | |
| "pad_token_id": 3, | |
| "rglru_c": 8.0, | |
| "rmsnorm_eps": 1e-06, | |
| "rnn_width": 864, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.3.0", | |
| "vocab_size": 39953 | |
| } | |