Add processed MedMCQA splits and SFT files

.gitattributes

@@ -58,3 +58,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+sft/sft_train_20000.jsonl filter=lfs diff=lfs merge=lfs -text
+train_20000.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,110 @@
+---
+language:
+- en
+license: apache-2.0
+pretty_name: MedMCQA
+size_categories:
+- 10K<n<100K
+source_datasets:
+- openlifescienceai/medmcqa
+task_categories:
+- question-answering
+- multiple-choice
+tags:
+- medical
+- medmcqa
+dataset_info:
+- config_name: sft
+  features:
+  - name: prompt
+    dtype: string
+  - name: answer
+    dtype: string
+  splits:
+  - name: train
+    num_examples: 20000
+  - name: validation
+    num_examples: 200
+- config_name: train
+  splits:
+  - name: train
+    num_examples: 20000
+- config_name: validation
+  splits:
+  - name: validation
+    num_examples: 200
+- config_name: mia_members
+  splits:
+  - name: train
+    num_examples: 2000
+- config_name: mia_nonmembers
+  splits:
+  - name: train
+    num_examples: 2000
+- config_name: filtered
+  splits:
+  - name: train
+    num_examples: 84403
+configs:
+- config_name: sft
+  data_files:
+  - split: train
+    path: sft/sft_train_20000.jsonl
+  - split: validation
+    path: sft/sft_validation_200.jsonl
+- config_name: train
+  data_files:
+  - split: train
+    path: train_20000.jsonl
+- config_name: validation
+  data_files:
+  - split: validation
+    path: validation_200.jsonl
+- config_name: mia_members
+  data_files:
+  - split: train
+    path: mia_members_2000.jsonl
+- config_name: mia_nonmembers
+  data_files:
+  - split: train
+    path: mia_nonmembers_2000.jsonl
+- config_name: filtered
+  data_files:
+  - split: train
+    path: medmcqa_filtered.parquet
+---
+
+# MedMCQA
+
+由 [openlifescienceai/medmcqa](https://huggingface.co/datasets/openlifescienceai/medmcqa) 的 train split 清洗、去重后划分得到。上游许可证为 Apache-2.0。
+
+默认配置 `sft` 可直接用于监督微调。`answer` 是完整解答加正确选项字母，不是单独的选项标签。
+
+```python
+from datasets import load_dataset
+
+sft = load_dataset("HaolinRPI/medmcqa", "sft")
+train = load_dataset("HaolinRPI/medmcqa", "train", split="train")
+validation = load_dataset("HaolinRPI/medmcqa", "validation", split="validation")
+members = load_dataset("HaolinRPI/medmcqa", "mia_members", split="train")
+nonmembers = load_dataset("HaolinRPI/medmcqa", "mia_nonmembers", split="train")
+filtered = load_dataset("HaolinRPI/medmcqa", "filtered", split="train")
+```
+
+## 划分
+
+Training、MIA Non-member 和 Validation 的规范化 Question 互不重叠。MIA Member 是 Training 的子集。随机种子是 42。完整统计见 `split_manifest.json`。
+
+| 文件 | 条数 | 用途 |
+|---|---:|---|
+| `sft/sft_train_20000.jsonl` | 20,000 | SFT 训练集，只含 `prompt` 和 `answer` |
+| `sft/sft_validation_200.jsonl` | 200 | SFT 验证集，只含 `prompt` 和 `answer` |
+| `train_20000.jsonl` | 20,000 | 训练样本，保留原始字段 |
+| `mia_members_2000.jsonl` | 2,000 | Training 内部的 MIA Member |
+| `mia_nonmembers_2000.jsonl` | 2,000 | 不参与训练的 MIA Non-member |
+| `validation_200.jsonl` | 200 | 独立验证样本 |
+| `medmcqa_filtered.parquet` | 84,403 | 全部合格记录，含未使用样本 |
+
+## SFT 文本
+
+`prompt` 由题干、A–D 选项和指令 “Answer with the letter of the correct option.” 组成。`answer` 由解释和正确选项字母组成，字母单独占最后一行。

medmcqa_filtered.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87c3adeb51d3eae144ec229490855b3b9e92222dcd577efab575690466b4a868
+size 92325023

sft/sft_train_20000.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2a946cc7e173bdc422c109c6a47a08c9d98b30456823f35e9ae6e862a35d6eb
+size 17516590

split_manifest.json

@@ -0,0 +1,95 @@
+{
+  "seed": 42,
+  "rng": "random.Random (Mersenne Twister, Python stdlib)",
+  "sampling_procedure": [
+    "Filter and deduplicate in original dataset order, keeping the first question_key.",
+    "Shuffle all filtered rows with Random(seed).",
+    "Training is the first 20000 shuffled rows.",
+    "MIA members are a Random.sample of 2000 positions inside training.",
+    "MIA non-members are a Random.sample of 2000 positions from the non-training rows.",
+    "Validation is a Random.sample of 200 positions from the still-unused rows."
+  ],
+  "source": {
+    "dataset": "openlifescienceai/medmcqa",
+    "split": "train"
+  },
+  "prompt_template": "{question}\\nA. {opa}\\nB. {opb}\\nC. {opc}\\nD. {opd}\\nReturn the correct option letter.",
+  "reference_template": "{explanation}\\n{answer_letter}",
+  "english_word_pattern": "[A-Za-z]+(?:['’-][A-Za-z]+)*",
+  "question_key": "strip, collapse whitespace, casefold; original question is unchanged",
+  "answer_mapping": {
+    "0": "A",
+    "1": "B",
+    "2": "C",
+    "3": "D"
+  },
+  "filter_stats": {
+    "raw_train": 182822,
+    "dropped_missing": 21953,
+    "dropped_not_single_answer": 54499,
+    "dropped_short_explanation": 21939,
+    "dropped_duplicate_question": 28,
+    "duplicate_question_groups": 28,
+    "filtered": 84403,
+    "dropped_duplicate_id": 0
+  },
+  "split_counts": {
+    "filtered": 84403,
+    "train": 20000,
+    "mia_member": 2000,
+    "mia_nonmember": 2000,
+    "validation": 200,
+    "unused": 62203
+  },
+  "verification": {
+    "sizes": {
+      "train": 20000,
+      "mia_member": 2000,
+      "mia_nonmember": 2000,
+      "validation": 200
+    },
+    "normalized_question_overlaps": {
+      "train_x_mia_nonmember": 0,
+      "train_x_validation": 0,
+      "mia_nonmember_x_validation": 0,
+      "mia_member_x_mia_nonmember": 0,
+      "mia_member_x_validation": 0
+    },
+    "member_subset_of_train": true,
+    "unique_question_keys": {
+      "train": 20000,
+      "mia_member": 2000,
+      "mia_nonmember": 2000,
+      "validation": 200
+    },
+    "parquet_rows": 84403,
+    "parquet_split_counts": {
+      "train": 20000,
+      "unused": 62203,
+      "mia_nonmember": 2000,
+      "validation": 200
+    }
+  },
+  "files": {
+    "medmcqa_filtered.parquet": {
+      "path": "/home/luoh6/luoh4/LiSA/medmcqa_processed/medmcqa_filtered.parquet",
+      "sha256": "87c3adeb51d3eae144ec229490855b3b9e92222dcd577efab575690466b4a868"
+    },
+    "train_20000.jsonl": {
+      "path": "/home/luoh6/luoh4/LiSA/medmcqa_processed/train_20000.jsonl",
+      "sha256": "6aac76f589217f4e6f048b6b747014c2a3df3e6a74d1c5c84bf6a9707a64404c"
+    },
+    "mia_members_2000.jsonl": {
+      "path": "/home/luoh6/luoh4/LiSA/medmcqa_processed/mia_members_2000.jsonl",
+      "sha256": "5eafbb71983b2ed58be5d200aab8add7acb735811b4e178b33db5eca1ca052ba"
+    },
+    "mia_nonmembers_2000.jsonl": {
+      "path": "/home/luoh6/luoh4/LiSA/medmcqa_processed/mia_nonmembers_2000.jsonl",
+      "sha256": "dd7b600668e08fd439240b0de2699c0c10bba9d0d9d50501ed6610323b1bf7c9"
+    },
+    "validation_200.jsonl": {
+      "path": "/home/luoh6/luoh4/LiSA/medmcqa_processed/validation_200.jsonl",
+      "sha256": "31cfb2323781bd5f9126f342a3cf1a44658c42c02af2ee51f1f1893f95b20dd9"
+    }
+  }
+}

train_20000.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aac76f589217f4e6f048b6b747014c2a3df3e6a74d1c5c84bf6a9707a64404c
+size 39005867