| { | |
| "_name_or_path": "facebook/opt-350m", | |
| "_remove_final_layer_norm": false, | |
| "activation_dropout": 0.0, | |
| "activation_function": "relu", | |
| "architectures": [ | |
| "OPTForSequenceClassification" | |
| ], | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 2, | |
| "do_layer_norm_before": false, | |
| "dropout": 0.1, | |
| "enable_bias": true, | |
| "eos_token_id": 2, | |
| "ffn_dim": 4096, | |
| "hidden_size": 1024, | |
| "id2label": { | |
| "0": "expository", | |
| "1": "procedural", | |
| "2": "non_math", | |
| "3": "probing_exploring", | |
| "4": "other_math" | |
| }, | |
| "init_std": 0.02, | |
| "label2id": { | |
| "expository": 0, | |
| "non_math": 2, | |
| "other_math": 4, | |
| "probing_exploring": 3, | |
| "procedural": 1 | |
| }, | |
| "layer_norm_elementwise_affine": true, | |
| "layerdrop": 0.0, | |
| "max_position_embeddings": 2048, | |
| "model_type": "opt", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 24, | |
| "pad_token_id": 1, | |
| "prefix": "</s>", | |
| "problem_type": "single_label_classification", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.25.1", | |
| "use_cache": true, | |
| "vocab_size": 50272, | |
| "word_embed_proj_dim": 512 | |
| } | |