Upload prepare_sft.py with huggingface_hub

prepare_sft.py

@@ -0,0 +1,90 @@
+from datasets import Dataset
+import sys
+import os
+import numpy as np
+
+
+DRY_RUN: bool = os.getenv("DRY_RUN", "0") == "1"
+
+
+__doc__ = rf"""
+A simple preprocessing script for preparing SFT data.
+Usage:
+    python {sys.argv[0]} /path/to/VTCTrain # results will be saved to /path/to/VTCTrain/sft
+    DRY_RUN=1 python {sys.argv[0]} /path/to/VTCTrain # to inspect the output format without saving
+
+Output arrow schema looks like:
+```json
+{{
+    'instruction': 'str',
+    'problem': 'str',
+    'answer': 'str',
+}}
+```
+
+You may customize instruction templates to mitigate overfitting, 
+or adjust problem/answer mildly to avoid exact matches.
+"""
+
+
+# normally this would be /path/to/VTCTrain
+parent_folder = sys.argv[1]
+assert "VTC" in parent_folder, "Please provide a VTC path."
+
+
+RULER_INSTRUCTION = [
+    "sample instruction 1",
+    "sample instruction 2",
+]
+NOLIMA_INSTRUCTION = [
+    "sample instruction 1",
+    "sample instruction 2",
+]
+
+
+def ruler_map(e: dict, idx: int) -> dict:
+    rng = np.random.default_rng(idx)
+    first_gt = e["answers"][0]
+    if isinstance(first_gt, str) and not first_gt.isdigit():
+        problem = f'What is one of the magic words for {e["problem"]}?'
+    else:
+        # isdigit or int, treat as number
+        assert isinstance(first_gt, (str, int))
+        problem = f'What is one of the magic numbers for {e["problem"]}?'
+
+    return {
+        "instruction": rng.choice(RULER_INSTRUCTION),
+        "problem": "<image>" * len(e["images"]) + problem,
+        "answer": ",".join(e["answers"]),
+    }
+
+
+def nolima_map(e: dict, idx: int) -> dict:
+    rng = np.random.default_rng(idx)
+    return {
+        "instruction": rng.choice(NOLIMA_INSTRUCTION),
+        "problem": "<image>" * len(e["images"]) + e["problem"],
+        "answer": e["answers"][0],
+    }
+
+
+subfolders = ["NoLiMa", "RULER-MK", "RULER-MQ", "RULER-MV", "RULER-S"]
+mapping_fns = [nolima_map, ruler_map, ruler_map, ruler_map, ruler_map]
+max_samples = [6600, 2200, 2200, 2200, 2200]
+
+for subfolder, mapping_fn, max_sample in zip(subfolders, mapping_fns, max_samples):
+    d = Dataset.load_from_disk(f"{parent_folder}/{subfolder}")
+    d = d.select(range(min(len(d), max_sample)))
+    d: Dataset = d.map(mapping_fn, with_indices=True, batch_size=100, writer_batch_size=100, num_proc=16)
+    sample = d[0]
+    print(
+    f"""
+instruction: {sample["instruction"]}
+    problem: {sample["problem"]}
+     answer: {sample["answer"]}
+    """
+    )
+
+    if DRY_RUN:
+        continue
+    d.save_to_disk(f"{parent_folder}/sft/{subfolder}")