onnx-internal-testing/tiny-random-ModernBertDecoderForCausalLM

1{2  "architectures": [3    "ModernBertDecoderForCausalLM"4  ],5  "attention_bias": false,6  "attention_dropout": 0.0,7  "bos_token": "[CLS]",8  "bos_token_id": 50281,9  "causal_mask": true,10  "classifier_activation": "gelu",11  "classifier_bias": false,12  "classifier_dropout": 0.0,13  "classifier_pooling": "mean",14  "cls_token_id": 50281,15  "decoder_bias": true,16  "deterministic_flash_attn": false,17  "embedding_dropout": 0.0,18  "eos_token": "[SEP]",19  "eos_token_id": 50282,20  "global_attn_every_n_layers": 3,21  "global_rope_theta": 160000.0,22  "gradient_checkpointing": false,23  "hidden_activation": "gelu",24  "hidden_size": 32,25  "initializer_cutoff_factor": 2.0,26  "initializer_range": 0.02,27  "intermediate_size": 32,28  "is_causal": true,29  "layer_norm_eps": 1e-05,30  "layer_types": [31    "full_attention",32    "sliding_attention"33  ],34  "local_rope_theta": 160000.0,35  "masked_prediction": false,36  "max_position_embeddings": 7999,37  "mlp_bias": false,38  "mlp_dropout": 0.0,39  "model_type": "modernbert-decoder",40  "norm_bias": false,41  "norm_eps": 1e-05,42  "num_attention_heads": 4,43  "num_hidden_layers": 2,44  "pad_token_id": 50283,45  "position_embedding_type": "sans_pos",46  "reference_compile": false,47  "sep_token_id": 50282,48  "sliding_window": 64,49  "tokenizer_class": "PreTrainedTokenizerFast",50  "torch_dtype": "float32",51  "transformers_version": "4.54.0.dev0",52  "unk_token": "[UNK]",53  "use_cache": true,54  "vocab_size": 5036855}56