| { | |
| "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": [ | |
| "af_ZA", | |
| "am_ET", | |
| "ar_AR", | |
| "be_BY", | |
| "bn_IN", | |
| "ca_ES", | |
| "cs_CZ", | |
| "cy_GB", | |
| "da_DK", | |
| "en_XX", | |
| "eo_EO", | |
| "et_EE", | |
| "eu_ES", | |
| "fa_IR", | |
| "fi_FI", | |
| "fr_XX", | |
| "ga_IE", | |
| "gl_ES", | |
| "gu_IN", | |
| "ha_NG", | |
| "hi_IN", | |
| "hr_HR", | |
| "hu_HU", | |
| "hy_AM", | |
| "id_ID", | |
| "is_IS", | |
| "it_IT", | |
| "ka_GE", | |
| "ko_KR", | |
| "ku_TR", | |
| "la_VA", | |
| "lt_LT", | |
| "lv_LV", | |
| "mk_MK", | |
| "ml_IN", | |
| "mn_MN", | |
| "ms_MY", | |
| "ne_NP", | |
| "nl_XX", | |
| "no_XX", | |
| "pl_PL", | |
| "ps_AF", | |
| "pt_XX", | |
| "ro_RO", | |
| "ru_RU", | |
| "sa_IN", | |
| "sd_PK", | |
| "si_LK", | |
| "sk_SK", | |
| "sl_SI", | |
| "so_SO", | |
| "sq_AL", | |
| "sr_RS", | |
| "sv_SE", | |
| "sw_KE", | |
| "ta_IN", | |
| "te_IN", | |
| "th_TH", | |
| "tl_XX", | |
| "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 | |
| } | |