| { | |
| "T": 0.1, | |
| "architectures": [ | |
| "BertModel" | |
| ], | |
| "attention_probs_dropout_prob": 0.1, | |
| "augmentation": "none", | |
| "beta1": 0.9, | |
| "beta2": 0.98, | |
| "bi_encoder": false, | |
| "chunk_length": 256, | |
| "classifier_dropout": null, | |
| "continue_training": false, | |
| "contrastive_mode": "moco", | |
| "dropout": 0.1, | |
| "eps": 1e-06, | |
| "eval_data": [ | |
| "minilm_denoised_T0_32_datasets_fixed_instruction_unfollowing_dev.jsonl" | |
| ], | |
| "eval_datasets": [], | |
| "eval_datasets_dir": "./", | |
| "eval_freq": 2000, | |
| "eval_normalize_text": false, | |
| "freeze_ctx_encoder": false, | |
| "global_rank": 0, | |
| "hard_order": false, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_size": 768, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "kd": false, | |
| "label_smoothing": 0.0, | |
| "layer_norm_eps": 1e-12, | |
| "loading_mode": "split", | |
| "local_rank": 0, | |
| "log_freq": 100, | |
| "loss_type": "kl", | |
| "lower_case": false, | |
| "lr": 1e-05, | |
| "lr_min_ratio": 0.0, | |
| "main_addr": "learnfair7603", | |
| "main_port": 15972, | |
| "max_position_embeddings": 512, | |
| "maxload": null, | |
| "moco_train_mode_encoder_k": false, | |
| "model_path": "/checkpoint/akariasai/contriever/contriever_wiki2020_ft_msmarco/checkpoint/step-10000/", | |
| "model_type": "bert", | |
| "momentum": 0.999, | |
| "n_context": 50, | |
| "negative_ctxs": 5, | |
| "negative_hard_min_idx": 0, | |
| "negative_hard_ratio": 0.1, | |
| "norm_doc": false, | |
| "norm_query": false, | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "num_workers": 5, | |
| "optim": "adamw", | |
| "output_dir": "/checkpoint/akariasai/contriever/instruction_unfollowing_full_data_from_new_checkpoints_5_0.1", | |
| "pad_token_id": 0, | |
| "per_gpu_batch_size": 16, | |
| "per_gpu_eval_batch_size": 16, | |
| "pooling": "average", | |
| "position_embedding_type": "absolute", | |
| "prob_augmentation": 0.0, | |
| "projection_size": 768, | |
| "queue_size": 65536, | |
| "random_init": false, | |
| "random_sort": false, | |
| "ratio_max": 0.5, | |
| "ratio_min": 0.1, | |
| "retriever_model_id": "bert-base-uncased", | |
| "rho": 0.05, | |
| "sampling_coefficient": 0.0, | |
| "save_freq": 2000, | |
| "scheduler": "linear", | |
| "score_function": "dot", | |
| "seed": 0, | |
| "temperature": 0.05, | |
| "torch_dtype": "float32", | |
| "total_steps": 20000, | |
| "train_data": [ | |
| "minilm_denoised_T0_32_datasets_fixed_instruction_unfollowing_train.jsonl" | |
| ], | |
| "transformers_version": "4.36.2", | |
| "type_vocab_size": 2, | |
| "use_cache": true, | |
| "vocab_size": 30522, | |
| "warmup_steps": 1000, | |
| "weight_decay": 0.01, | |
| "world_size": 64 | |
| } | |