| { |
| "graph_path": "${DATA_ROOT}/Mathlib/mathlib4-v4.22.0.json", |
| "batch_size": 16, |
| "epochs": 80, |
| "lr": 0.0006, |
| "lr_scheduler": "constant", |
| "warmup_ratio": 0.05, |
| "warmup_steps": 0, |
| "min_lr": 0.0, |
| "max_nodes": 300, |
| "subgraph_mode": "random_2hop", |
| "select_top_k": 64, |
| "hidden_dim": 512, |
| "proj_dim": 256, |
| "local_layers": 3, |
| "heads": 4, |
| "global_dim": null, |
| "dropout": 0.1, |
| "device": "cuda", |
| "text_encoder_device": "cuda", |
| "num_workers": 4, |
| "max_train_queries": 0, |
| "max_val_queries": 0, |
| "max_test_queries": 0, |
| "train_repeat": 2, |
| "dataset_mode": "mathlib", |
| "combined_jsonl_path": null, |
| "combined_val_count": 1000, |
| "combined_test_count": 1000, |
| "split_mode": "fixed_holdout", |
| "val_count": 2000, |
| "test_count": 2000, |
| "holdout_per_split": 1000, |
| "out_dir": "runs/phase2_subgraph", |
| "embedding": "alibaba-gte-modernbert-ft", |
| "seed": 42, |
| "max_length": 1024, |
| "sub_node_max_length": 1024, |
| "proof_pattern_mode": "none", |
| "proof_pattern_embedding": "alibaba-gte-modernbert-ft", |
| "proof_pattern_max_length": 768, |
| "pin_memory": true, |
| "prefetch_factor": 1, |
| "max_candidates_per_batch": 1500, |
| "persistent_workers": false, |
| "eval_batch_size": 8, |
| "eval_num_workers": 2, |
| "eval_prefetch_factor": 1, |
| "eval_pin_memory": false, |
| "eval_dynamic_fraction": 0.05, |
| "feature_cache_dir": "runs/phase2_subgraph/feature_cache", |
| "precompute_batch_size": 512, |
| "force_precompute_features": false, |
| "ddp_backend": "nccl", |
| "resume_path": null, |
| "memory_bank_dir": "runs/phase2_subgraph/memory_bank", |
| "memory_bank_topk": 512, |
| "memory_bank_chunk_size": 2048, |
| "memory_bank_per_query": 0, |
| "memory_bank_build_device": "cuda:0", |
| "random_negatives_per_query": 0, |
| "random_hard_negatives_per_query": 30, |
| "hard_negatives_per_query": 0, |
| "global_negatives_per_query": 30, |
| "fixed_retrieval_negatives": false, |
| "fixed_retrieval_topk": 10, |
| "fixed_retrieval_chunk_size": 256, |
| "fixed_retrieval_build_device": null, |
| "fixed_retrieval_cache_dir": null, |
| "fixed_retrieval_force_rebuild": false, |
| "fixed_retrieval_strict": false, |
| "mathlib_fixed_retrieval_checkpoint": null, |
| "mathlib_fixed_premise_embeddings": null, |
| "combined_fixed_retrieval_checkpoint": null, |
| "combined_fixed_premise_embeddings": null, |
| "cross_batch_negs": 32, |
| "cross_batch_queue": 4096, |
| "resource_monitor_interval": 60.0, |
| "wandb_project": null, |
| "wandb_entity": null, |
| "wandb_run_name": null, |
| "loss_reduction": "positive", |
| "loss_type": "positive_infonce", |
| "diversity_weight": 0.0, |
| "diversity_temperature": 1.0, |
| "loss_scale": 20.0, |
| "loss5_neg_source": null, |
| "loss5_neg_count": 3, |
| "loss5_pair_batch_size": null, |
| "grad_accum_steps": 8, |
| "subgraph_cache_dir": "runs/phase2_subgraph/subgraph_cache", |
| "force_precompute_subgraphs": false, |
| "use_amp": true |
| } |
|
|
