| { | |
| "architectures": [ | |
| "FastAnkhForMaskedLM" | |
| ], | |
| "attn_backend": "sdpa", | |
| "auto_map": { | |
| "AutoConfig": "modeling_ankh.FastAnkhConfig", | |
| "AutoModel": "modeling_ankh.FastAnkhModel", | |
| "AutoModelForMaskedLM": "modeling_ankh.FastAnkhForMaskedLM", | |
| "AutoModelForSequenceClassification": "modeling_ankh.FastAnkhForSequenceClassification", | |
| "AutoModelForTokenClassification": "modeling_ankh.FastAnkhForTokenClassification" | |
| }, | |
| "d_ff": 3072, | |
| "d_kv": 64, | |
| "d_model": 768, | |
| "dense_act_fn": "gelu_new", | |
| "dtype": "float32", | |
| "eos_token_id": 1, | |
| "initializer_factor": 1.0, | |
| "layer_norm_epsilon": 1e-06, | |
| "model_type": "fast_ankh", | |
| "num_heads": 12, | |
| "num_layers": 48, | |
| "pad_token_id": 0, | |
| "relative_attention_max_distance": 128, | |
| "relative_attention_num_buckets": 64, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.6", | |
| "vocab_size": 144 | |
| } | |