| { |
| "architectures": [ |
| "BidirLMModel" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "auto_map": { |
| "AutoConfig": "configuration_bidirlm.BidirLMConfig", |
| "AutoModel": "modeling_bidirlm.BidirLMModel", |
| "AutoModelForMaskedLM": "modeling_bidirlm.BidirLMForMaskedLM", |
| "AutoModelForPreTraining": "modeling_bidirlm.BidirLMPreTrainedModel", |
| "AutoModelForSequenceClassification": "modeling_bidirlm.BidirLMForSequenceClassification", |
| "AutoModelForTokenClassification": "modeling_bidirlm.BidirLMForTokenClassification" |
| }, |
| "bos_token_id": 151644, |
| "clf_pooling": "late", |
| "dtype": "bfloat16", |
| "eos_token_id": 151645, |
| "head_dim": 128, |
| "hidden_act": "silu", |
| "hidden_size": 2048, |
| "initializer_range": 0.02, |
| "intermediate_size": 6144, |
| "mask_token": "<|mask|>", |
| "mask_token_id": 151663, |
| "max_position_embeddings": 32768, |
| "max_window_layers": 28, |
| "model_type": "bidirlm", |
| "num_attention_heads": 16, |
| "num_hidden_layers": 28, |
| "num_key_value_heads": 8, |
| "rms_norm_eps": 1e-06, |
| "rope_scaling": null, |
| "rope_theta": 1000000, |
| "tie_word_embeddings": true, |
| "transformers_version": "4.57.6", |
| "vocab_size": 151936 |
| } |
|
|