COMP90042 26S1 — New-Methodology data (ZHUOCHENCN/26S1COMP90042)

Training and derived datasets for the climate claim verification pipeline.

Retrieval (data/)

File	Description
reranker_train_pairs.jsonl	CE finetune pairs (claim, passage, label)
reranker_train_pairs.meta.json	Hard-negative mining metadata
train_bge_top150_candidates.json	Per-claim BGE-large top-150 pools (train)

Classification (classification_datasets/)

| classification_datasets/dev_claim_only_k1.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k1.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k2.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k3.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k4.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k5.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k6.jsonl | Classification pipeline | | classification_datasets/dev_gold_evidence_k8.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k1.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k2.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k3.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k4.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k5.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k6.jsonl | Classification pipeline | | classification_datasets/dev_miss_to_nei_relabel_k8.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k1.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k2.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k3.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k4.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k5.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k6.jsonl | Classification pipeline | | classification_datasets/dev_mixed_gold_retrieved_k8.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k1.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k2.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k3.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k4.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k5.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k6.jsonl | Classification pipeline | | classification_datasets/dev_retrieved_official_label_k8.jsonl | Classification pipeline | | classification_datasets/test_claim_only_k1.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k1.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k2.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k3.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k4.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k5.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k6.jsonl | Classification pipeline | | classification_datasets/test_retrieved_official_label_k8.jsonl | Classification pipeline | | classification_datasets/train_claim_only_k1.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k1.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k2.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k3.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k4.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k5.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k6.jsonl | Classification pipeline | | classification_datasets/train_gold_evidence_k8.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k1.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k2.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k3.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k4.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k5.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k6.jsonl | Classification pipeline | | classification_datasets/train_miss_to_nei_relabel_k8.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k1.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k2.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k3.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k4.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k5.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k6.jsonl | Classification pipeline | | classification_datasets/train_mixed_gold_retrieved_k8.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k1.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k2.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k3.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k4.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k5.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k6.jsonl | Classification pipeline | | classification_datasets/train_retrieved_official_label_k8.jsonl | Classification pipeline |

Intermediate artifacts (artifacts/)

HF path	Local	Purpose
artifacts/data_cache/	outputs/data_cache/	Preprocessed claims/evidence
artifacts/retrieval/candidates/	outputs/retrieval/candidates/	BM25/dense/hybrid candidate pools
artifacts/retrieval/cache/	outputs/retrieval/cache/	BM25 index, dense embeddings, CE score caches
artifacts/retrieval/metrics/	per-claim CSVs	Fine-grained eval tables
artifacts/ablation/	outputs/ablation/	Preprocessing ablation run
artifacts/reranker/	outputs/reranker/	MMR / fusion / calibration figures & predictions (excl. pairwise weights)

Pairwise CE weights: HF Model → retrieval/pairwise_reranker/. See REPRODUCIBILITY.md.

Mining config

{
  "output": "/hpc2hdd/home/jyang577/joe/New-Methodology/data/reranker_train_pairs.jsonl",
  "pool_cache": "/hpc2hdd/home/jyang577/joe/New-Methodology/data/train_bge_top150_candidates.json",
  "dense_model": "BAAI/bge-large-en-v1.5",
  "candidate_top_n": 150,
  "hard_neg_min": 4,
  "hard_neg_max": 8,
  "bf16_encode": false,
  "num_pairs": 13946,
  "claims": 1228,
  "positives": 4122,
  "hard_negatives": 9824,
  "claims_with_fewer_than_min_neg": 0,
  "claims_with_no_gold": 0
}