| { | |
| "architectures": [ | |
| "EuroBertForMaskedLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token": "<|begin_of_text|>", | |
| "bos_token_id": 128000, | |
| "eos_token": "<|end_of_text|>", | |
| "eos_token_id": 128001, | |
| "pad_token": "<|end_of_text|>", | |
| "pad_token_id": 128001, | |
| "mask_token": "<|mask|>", | |
| "mask_token_id": 128002, | |
| "hidden_act": "silu", | |
| "hidden_dropout": 0.0, | |
| "hidden_size": 2304, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 6144, | |
| "max_position_embeddings": 8192, | |
| "model_type": "eurobert", | |
| "num_attention_heads": 18, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 6, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 250000, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.40.0.dev0", | |
| "vocab_size": 128256, | |
| "is_decoder": false | |
| } |