| { | |
| "_name_or_path": "ashercn97/ashbert-v003", | |
| "architectures": [ | |
| "NeoBERTLMHead" | |
| ], | |
| "decoder_init_range": 0.02, | |
| "dim_head": 64, | |
| "embedding_init_range": 0.02, | |
| "hidden_size": 512, | |
| "intermediate_size": 1536.0, | |
| "kwargs": { | |
| "_commit_hash": "c244da032abb502393447b29a144c12569f6c42f", | |
| "architectures": [ | |
| "NeoBERTLMHead" | |
| ], | |
| "attn_implementation": null, | |
| "dim_head": 64, | |
| "kwargs": {}, | |
| "model_type": "neobert", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.45.2" | |
| }, | |
| "max_length": 256, | |
| "model_type": "neobert", | |
| "norm_eps": 1e-06, | |
| "num_attention_heads": 8, | |
| "num_hidden_layers": 8, | |
| "pad_token_id": 0, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.45.2", | |
| "vocab_size": 30522 | |
| } | |