| { |
| "architectures": [ |
| "HEDForConditionalGeneration" |
| ], |
| "attention_probs_dropout_prob": 0.1, |
| "bos_token_id": 0, |
| "classifier_dropout": 0.1, |
| "d_ff": 3072, |
| "d_kv": 64, |
| "d_model": 768, |
| "decode_anchor_tokens": false, |
| "decoder_start_token_id": 0, |
| "dense_act_fn": "relu", |
| "dropout_rate": 0.1, |
| "eos_token_id": 2, |
| "eval_splits": null, |
| "feed_forward_proj": "relu", |
| "hidden_dropout_prob": 0.1, |
| "initializer_factor": 1.0, |
| "is_encoder_decoder": true, |
| "is_gated_act": false, |
| "layer_norm_epsilon": 1e-06, |
| "max_decoder_position_embeddings": 512, |
| "max_encoder_position_embeddings": 8192, |
| "model_type": "hed", |
| "num_decoder_layers": 6, |
| "num_heads": 12, |
| "num_layers": 6, |
| "pad_token_id": 1, |
| "pool_scheme": "first-token", |
| "pos_emb": "scaled-sinusoidal", |
| "position_embedding_type": "quaternion", |
| "prediction_head": false, |
| "relative_attention_max_distance": 128, |
| "relative_attention_num_buckets": 32, |
| "sep_token_id": 2, |
| "torch_dtype": "float32", |
| "transformers_version": "4.29.0.dev0", |
| "use_bias": false, |
| "use_cache": true, |
| "vocab_size": 50772 |
| } |
|
|
