| { |
| "adapter_layer_norm": false, |
| "adapter_reduction_factor": 2, |
| "adapter_reuse_layer_norm": true, |
| "architectures": [ |
| "XmodForMaskedLM" |
| ], |
| "attention_probs_dropout_prob": 0.1, |
| "bos_token_id": 0, |
| "classifier_dropout": null, |
| "eos_token_id": 2, |
| "hidden_act": "gelu", |
| "hidden_dropout_prob": 0.1, |
| "hidden_size": 768, |
| "initializer_range": 0.02, |
| "intermediate_size": 3072, |
| "languages": [ |
| "ar_AR", |
| "en_XX", |
| "fi_FI", |
| "fr_XX", |
| "hi_IN", |
| "id_ID", |
| "ka_GE", |
| "ko_KR", |
| "ru_RU", |
| "sw_KE", |
| "ta_IN", |
| "th_TH", |
| "vi_VN" |
| ], |
| "layer_norm_eps": 1e-05, |
| "ln_before_adapter": true, |
| "max_position_embeddings": 514, |
| "model_type": "xmod", |
| "num_attention_heads": 12, |
| "num_hidden_layers": 12, |
| "pad_token_id": 1, |
| "position_embedding_type": "absolute", |
| "pre_norm": false, |
| "torch_dtype": "float32", |
| "transformers_version": "4.26.0.dev0", |
| "type_vocab_size": 1, |
| "use_cache": true, |
| "vocab_size": 250002 |
| } |
|
|
