forge-neoforge-sft-v2.0.3 (full)

Supervised fine-tuning corpus for Forge 1.20.1 -> NeoForge 1.21.1 mod migration. Every task is grounded in exact pinned upstream snapshots (branch/tag + commit sha per record): the model is never asked about invisible code - all context files are serialized inline (inline-fenced-v2) and their presence is enforced by a blocking gate (G-CTX).

11216 records / 15 families / 841 repositories. Carries the entire v2.0.2 release (7030 records) byte-identical, expanded with a third discovery round: (1) the untapped remainder of the v2.0.2 re-scan (295 repositories with candidate pairs that were beyond the previous top-200 cutoff), (2) a fresh GitHub search-based discovery (100 additional valid pairs found outside the original registry altogether), and (3) a deeper PR-thread multiturn layer over all new repositories. multiturn_repair grows 90 -> 258.

Splits (repo-disjoint)

split	records	repos
train	10209	753
validation	674	49
eval	333	39

Repositories are disjoint across splits (no leakage). eval stays carryover-only at 333 (~ within the 200-300 target band, preserved per repository for purity).

Families

family	total	train / val / eval	description
api_version_lookup	547	507 / 34 / 6	symbol-level Forge->NeoForge replacement lookups validated literally in the repository own pre/post snapshots
code_generation	506	456 / 28 / 22	implement a new class from scratch consistent with attached conventions; the answer is the real upstream-added file
config_and_resources	1277	1148 / 80 / 49	lang/pack.mcmeta/toml/recipe/tag updates between pins, real diff notes only
context_verification	19	14 / 2 / 3	context adequacy checks (v2 family, extended)
grounded_qa	1144	1031 / 71 / 42	code comprehension Q&A grounded in pinned files (events, registrations, config, responsibilities)
insufficient_context	687	619 / 43 / 25	symmetric negative cases: missing internal dependencies, version-mismatch claims, unconfirmed third-party APIs - the correct behavior is to refuse and request exactly what is needed
issue_bugfix	459	424 / 14 / 21	real upstream bugfixes: prompt carries the exact pre-fix parent revision; the answer is the real fix commit result
issue_context_check	1069	988 / 54 / 27	real issue triage: version-conflict reports and insufficient bug reports (no logs) -> request specifics, do not act
multiturn_repair	258	241 / 13 / 4	real repair dialogues: PR-thread sequences (merged PRs, per-commit diffs, optional verbatim reviewer quotes) and fix -> follow-up commit chains; 4-message
porting	2603	2367 / 173 / 63	L1 file migration: migrate an exact Forge 1.20.1 snapshot of one file to NeoForge 1.21.1; the answer is the upstream 1.21.1 state of that file (pinned sha256)
porting_diff_review	1284	1179 / 79 / 26	L1: review a proposed single-file migration diff that IS the real upstream change; verify hunks and state what a full port additionally requires
pr_review	612	560 / 34 / 18	file-by-file review of real merged PR diffs with verifiable observations
project_porting	335	305 / 22 / 8	L2: subsystem-level migration; multiple dependent files planned, migrated and cross-referenced
review	262	225 / 19 / 18	L1 review-and-verify: analyze real 1.20.1 code / real migration artifacts, grounded quotes only
test_generation	154	145 / 8 / 1	game-test skeletons grounded in real registrations + NeoForge 1.21.1 gametest API surface; honest static_checked status

Methodology (honesty core)

• Pins. Each record carries pins (branch or tag + commit sha for Forge 1.20.1 and NeoForge 1.21.1) and upstream_truth (paths + sha256 of the exact snapshots it was derived from).
• Round-3 discovery. (a) 295 re-scan leftovers (candidate pairs found by v2.0.2's full ls-remote ref enumeration but not processed then) were taken as-is; (b) a fresh GitHub search round (10 queries; repos matching NeoForge/1.21.1/Forge-1.20.1 signals, forks and archives excluded) produced 390 previously unknown candidates, of which 100 had plausible pin pairs by ref matching. Every candidate then passed the same pin extractor.
• Loader gate + MC-version content check. Full-tree git grep at BOTH pins: 210 pass (98 multiloader with fabric/quilt subtrees excluded), 183 rejected (90 fabric/non-neo targets, 67 missing MC-version text, 26 forge-less sources), 2 without java diff -> 210 new valid pairs.
• PR-thread multiturn (deeper). Same provenance as v2.0.2: merged PRs -> per-commit .patch diffs (non-API endpoint, quota-free), exact git-object file states at PR commit shas (shallow fetch depth=2), dedup against every existing multiturn record; real reviewer comments quoted verbatim (author + URL) where the core API budget allowed.
• No fabricated builds. Statuses: static_checked (static verification against pinned snapshots). validation.method states exactly what was verified. No compile/run was performed in the producing environment for this release.
• Gates. G1 unique ids; G-CTX (every context file byte-present in a user turn - blocking, both user turns of 4-message dialogues included); G3 context sha256; G5 secret scan; G9 carried v2.0.2 answers byte-identical (7030/7030) and every carried v2.0.2 file an exact byte-prefix of the merged file; G10 upstream re-fetch sample (12 sha256 matches, 4 skipped by design, 0 mismatches; raw fetches are CRLF-normalized to match the git-object text extraction - one CRLF file otherwise compared unequally); repo-level split disjointness; exact-duplicate drop (0 dropped this round).
• Multiloader policy (v2 precedent): repos with Common/Forge/NeoForge trees are valid material; fabric/quilt subtrees are excluded from every selection.

Record schema (per record)

task_id, task_family, task_type, source_target, prompt, context_files[] (path, content, sha256, source_url), messages (self-contained inline-fenced-v2; for multiturn_repair a 4-message dialogue), expected_changes, evidence[] (claim, source_url, source_revision), build_environment, validation (honest status), repo, pins, split, serialization, upstream_truth (path/sha256/revision; PR dialogues carry pr_number + commit shas).

Limitations

• compile_verified statuses are absent: NeoForm decompile requires >=8 GB RAM (Vineflower -Xmx4G), unavailable in the producing environment; nothing was fabricated to hide this.
• Reviewer-quote multiturn subset remains a minority: GitHub core API (60/h, shared IP); PR-thread follow-ups quote real follow-up commit messages instead - labeled as such.
• test_generation records are grounded skeletons (API surface verified against NeoForge 1.21.1 tree), not compiled tests.
• Some discovered pairs pin release tags rather than branch tips (provenance keeps the exact ref + sha); tags were required when a repo had no dedicated 1.21.1 branch.
• Issue/PR texts are used as real user reports; issue-layer records are limited to repos with fetchable commit objects.

Source & provenance

• Base: ComerX/forge-neoforge-sft-v2.0.2 (7030 records, carried byte-identical) <- ComerX/forge-neoforge-sft-v2.0.1 (3615 records) <- ComerX/forge-neoforge-sft-v2 (1070 records).
• Expansion round 3: 210 new repository pairs (295 re-scan leftovers -> 98 valid after gates; 100 search-discovered valid pairs; totals per gate table above); issues/PRs via GitHub search for all new repos.
• Upstream licenses: per-repository (MIT/Apache-2.0/LGPL-2.1/BSD and similar); pins allow byte-exact verification of every derived answer.

Citation

@dataset{comerx_sft_v203_2026,
  title  = {forge-neoforge-sft-v2.0.3 (full): Forge 1.20.1 -> NeoForge 1.21.1 SFT corpus},
  author = {ComerX},
  year   = {2026},
  url    = {https://huggingface.co/datasets/ComerX/forge-neoforge-sft-v2.0.3}
}