| { | |
| "ESM": false, | |
| "MI_loss": false, | |
| "MOE": false, | |
| "_name_or_path": "allenai/scibert_scivocab_uncased", | |
| "a_col": "a", | |
| "add_during_eval": false, | |
| "architectures": [ | |
| "BertForSentenceSimilarity" | |
| ], | |
| "attention_probs_dropout_prob": 0.0, | |
| "b_col": "b", | |
| "classifier_dropout": null, | |
| "contact_head": false, | |
| "contrastive_loss": "clip", | |
| "data_paths": [ | |
| "lhallee/abstract_domain_parasitic" | |
| ], | |
| "domains": [ | |
| "[PAR]" | |
| ], | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.05, | |
| "hidden_size": 768, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "label_col": "label", | |
| "layer_norm_eps": 1e-12, | |
| "limits": false, | |
| "log_path": "./results.txt", | |
| "max_length": 512, | |
| "max_position_embeddings": 512, | |
| "model_path": "allenai/scibert_scivocab_uncased", | |
| "model_type": "bert", | |
| "moe_type": "topk", | |
| "new_special_tokens": false, | |
| "num_attention_heads": 12, | |
| "num_experts": 8, | |
| "num_hidden_layers": 12, | |
| "num_tasks": 5, | |
| "pad_token_id": 0, | |
| "patience": 3, | |
| "position_embedding_type": "absolute", | |
| "project_name": "SciMOE", | |
| "token_moe": false, | |
| "topk": 2, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.38.2", | |
| "type_vocab_size": 2, | |
| "use_cache": true, | |
| "vocab_size": 31090, | |
| "wBAL": 0.1, | |
| "wMI": 0.1, | |
| "wandb": true, | |
| "wandb_name": "parasite_mixed_full", | |
| "wandb_project": "SSPR", | |
| "weight_path": "parasitic_single.pt" | |
| } | |