Add files using upload-large-folder tool

README.md

@@ -0,0 +1,57 @@
+---
+license: mit
+language:
+- en
+task_categories:
+- text-generation
+tags:
+- instruction-tuning
+- tool-calling
+- math
+- dfm11
+pretty_name: dfm11-mathagentic-tinygsm-python
+---
+
+# dfm11-mathagentic-tinygsm-python
+
+English arithmetic word problems converted into native Python tool-call trajectories with precomputed tool responses and boxed final answers.
+
+## Contents
+
+- Rows: 367,749
+- Shards: 4
+- Format: deterministic gzip JSON Lines in `data/train-*.jsonl.gz`
+- Schema: `tools`, four-message native tool trajectory, execution metadata,
+  stable source ID, source revision, and admission status
+- Intended repository: `schneiderkamplab/dfm11-mathagentic-tinygsm-python`
+
+Each row contains a user problem, an assistant tool call, a precomputed tool
+response, and a terminal assistant response containing exactly one boxed final
+answer. It is directly consumable by Mathagentic/HRM-Text's Gemma 4 native
+tool-call rendering path; no executor is used during training.
+
+## Upstream datasets
+
+- `TinyGSM/TinyGSM`
+
+## Selection and verification
+
+Deterministic 1-in-20 source sample; restricted-Python validation and execution; independent Gemma 4 26B A4B semantic audit; accepted verdicts only.
+
+The semantic audit covered 500,000 candidates and accepted 367,749 (73.55%). Rejected rows are not packaged.
+
+The package contains source trajectories, not tokenized arrays or sampled epoch
+indices. Packaging does not itself admit the dataset into a DFM11 data mix.
+The approved DFM11 sampling policy is to deterministically cap to 50,000 distinct rows per epoch, with repeat
+`1`. This policy does not remove rows from the published
+source package.
+
+## License and citation review
+
+The upstream TinyGSM dataset declares the MIT license. Preserve its attribution.
+
+## Validate
+
+```bash
+python validate_dataset.py
+```

data/train-00000.jsonl.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a676fa3969dcabeee142bf2b37e29d074d593a404ee5601f90311adf80136e6e
+size 16257859

data/train-00001.jsonl.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:596789e6be5e4e2603329fe1e2d991ae52563efe7bc31e973ea86f4eb4e04a1e
+size 15562648

data/train-00002.jsonl.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d49ebbe480484a09c87c520d444ed27c4a03fd8aa04d8031de97ab0845b280d0
+size 16724818

data/train-00003.jsonl.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:318ae8339303c05170259b89d57fa4abda8109884ff831fc7cd457b224d68327
+size 10788975

metadata/manifest.json

@@ -0,0 +1,60 @@
+{
+  "admission_status": "semantic_verified",
+  "data_files": [
+    {
+      "bytes": 16257859,
+      "file": "data/train-00000.jsonl.gz",
+      "sha256": "a676fa3969dcabeee142bf2b37e29d074d593a404ee5601f90311adf80136e6e"
+    },
+    {
+      "bytes": 15562648,
+      "file": "data/train-00001.jsonl.gz",
+      "sha256": "596789e6be5e4e2603329fe1e2d991ae52563efe7bc31e973ea86f4eb4e04a1e"
+    },
+    {
+      "bytes": 16724818,
+      "file": "data/train-00002.jsonl.gz",
+      "sha256": "d49ebbe480484a09c87c520d444ed27c4a03fd8aa04d8031de97ab0845b280d0"
+    },
+    {
+      "bytes": 10788975,
+      "file": "data/train-00003.jsonl.gz",
+      "sha256": "318ae8339303c05170259b89d57fa4abda8109884ff831fc7cd457b224d68327"
+    }
+  ],
+  "dfm11_sampling": {
+    "repeat": 1,
+    "row_cap_per_epoch": 50000
+  },
+  "format": "mathagentic-native-tool-sft-v1; deterministic gzip JSON Lines",
+  "intended_hf_id": "schneiderkamplab/dfm11-mathagentic-tinygsm-python",
+  "license": "mit",
+  "name": "dfm11-mathagentic-tinygsm-python",
+  "rows": 367749,
+  "selection": "Deterministic 1-in-20 source sample; restricted-Python validation and execution; independent Gemma 4 26B A4B semantic audit; accepted verdicts only.",
+  "semantic_audit": {
+    "acceptance_rate": 0.735498,
+    "accepted_rows": 367749,
+    "audit_models": [
+      "/work/mimir/.home/.cache/huggingface/hub/models--google--gemma-4-26B-A4B-it/snapshots/4d7ae4984b7db7de8f8457170b3f1a419ee76d52"
+    ],
+    "audited_rows": 500000,
+    "rejected_rows": 132251,
+    "rejection_reasons": {
+      "ambiguous": 658,
+      "malformed_question": 9900,
+      "unit_error": 1924,
+      "unsafe_or_irrelevant": 3028,
+      "wrong_operation": 110774,
+      "wrong_quantity": 5967
+    },
+    "verdict_sha256": "9badd56a7c71a29c74cffc5ccc478a744b961b0a071a3657cabdfb7dfb048f97"
+  },
+  "shards": 4,
+  "source_revision": "5049a7840fa546985eaae4c698d5daf6fdf7b47d",
+  "source_shards": 50,
+  "upload_performed": false,
+  "upstream": [
+    "TinyGSM/TinyGSM"
+  ]
+}

validate_dataset.py

@@ -0,0 +1,32 @@
+#!/usr/bin/env python3
+import gzip, hashlib, json
+from pathlib import Path
+
+root = Path(__file__).parent
+manifest = json.loads((root / "metadata/manifest.json").read_text())
+rows = 0
+seen = set()
+for item in manifest["data_files"]:
+    path = root / item["file"]
+    digest = hashlib.sha256(path.read_bytes()).hexdigest()
+    if digest != item["sha256"]:
+        raise ValueError(f"checksum mismatch: {path}")
+    with gzip.open(path, "rt", encoding="utf-8") as handle:
+        for line in handle:
+            row = json.loads(line)
+            sid = row.get("source_id")
+            if not sid or sid in seen:
+                raise ValueError(f"missing or duplicate source_id: {sid}")
+            seen.add(sid)
+            messages = row.get("messages")
+            if [message.get("role") for message in messages] != ["user", "assistant", "tool", "assistant"]:
+                raise ValueError(f"invalid roles: {sid}")
+            call = messages[1]["tool_calls"][0]
+            if call["id"] != messages[2].get("tool_call_id"):
+                raise ValueError(f"call/result mismatch: {sid}")
+            if not str(messages[3].get("content") or "").startswith("\\boxed{"):
+                raise ValueError(f"missing boxed answer: {sid}")
+            rows += 1
+if rows != manifest["rows"]:
+    raise ValueError(f"row count mismatch: {rows} != {manifest['rows']}")
+print(json.dumps({"rows": rows, "valid": True}, sort_keys=True))