Datasets:
task string | source_trajectory_id string | source_trajectory_sha256 string | lang string | category string | domain string | verifier string | split string | teacher_runtime string | teacher_model string | reasoning_effort string | provider string | observed_models list | model_attested bool | trace_format string | tools_used list | derivation string | assistant_step int64 | assistant_steps int64 | target_message_index int64 | original_n_messages int64 | n_messages int64 | messages list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
bash-filedrop | bash-filedrop:05cd4b5ba6aba12e53fd | 05cd4b5ba6aba12e53fdd813a5528675ab311f8756d8031799dfadf677caac61 | bash | refactor-argv | agent | acceptance-tests+protected-file-hash+independent-review | train | pi | anthropic/claude-fable-5 | xhigh | openrouter | [
"anthropic/claude-fable-5"
] | true | moonshiner-synthetic-correction-v1 | [
"edit",
"read"
] | cumulative-next-assistant-v1 | 1 | 2 | 1 | 6 | 2 | [
{
"role": "user",
"content": "Tech-debt ticket on runjobs.sh, our little filing-job runner (reads a tab-separated job list, copies/gzips files out of the drop folder, appends to filing.log). It works, and it has behavior tests that pass — but look at HOW it works: every job is assembled into a command STRIN... |
bash-filedrop | bash-filedrop:05cd4b5ba6aba12e53fd | 05cd4b5ba6aba12e53fdd813a5528675ab311f8756d8031799dfadf677caac61 | bash | refactor-argv | agent | acceptance-tests+protected-file-hash+independent-review | train | pi | anthropic/claude-fable-5 | xhigh | openrouter | [
"anthropic/claude-fable-5"
] | true | moonshiner-synthetic-correction-v1 | [
"edit",
"read"
] | cumulative-next-assistant-v1 | 2 | 2 | 4 | 6 | 5 | [
{
"role": "user",
"content": "Tech-debt ticket on runjobs.sh, our little filing-job runner (reads a tab-separated job list, copies/gzips files out of the drop folder, appends to filing.log). It works, and it has behavior tests that pass — but look at HOW it works: every job is assembled into a command STRIN... |
Model Synthetic Corrections
1 TRAJECTORIES · 2 TRAINING ROWS · 16 kB
Generated by moonshiner — an open harness for distilling verified instruction-following, tool-use, and agentic coding traces.
Synthetic Corrections companion dataset. The original dataset is greghavens/fable-5-coding-and-debugging-traces. These are narrowly, synthetically corrected, independently re-judged traces that never passed in the original dataset.
Behavior-preserving instruction-following, tool-use, and agent trajectories
from Claude Fable 5 (anthropic/claude-fable-5). The category and row-share tables
below describe the actual mix seen during training rather than assuming a
particular task domain.
This is an actively growing dataset. More is coming: additional training programs and substantially more sessions will be added to this same repo.
What makes it different
Narrowly corrected failed trajectories. Every row starts from a genuine model session that never passed the primary trace judge. The source reasoning is preserved unchanged; only a minimal, explicitly disclosed synthetic correction is retained.
One next step per row. A trajectory with N assistant turns produces N rows. Row k contains the complete context through assistant turn k; that final assistant message is the sole training target.
Runtime-normalized. Runtime plumbing, UI decoration, control sequences, and verbose success boilerplate are removed or canonicalized while causal context remains.
Independently verified. Coding sessions must pass deterministic tests and protected-file checks. Instruction-following sessions must pass deterministic tool-call, staging, argument, and response-constraint checks. Every retained trajectory also clears independent review.
Reasoning-effort step-down. Failed trace attempts proceed through
xhigh → medium → low(up to the configured attempt count) and stop at the first judge-accepted trace. If higher reasoning fails a task that lower reasoning succeeds on, the lower-effort trace is retained.
Task mix
High-level training programs, calculated from accepted trajectories using the same program mapping published in the seed catalog:
| kind | trajectories | share | row share | flavor |
|---|---|---|---|---|
| Refactoring & performance | 1 | 100.0% | 100.0% | Restructure safely and improve measured performance without behavior drift. |
Languages (current drop)
Bash
Schema
Each row:
| column | type | contents |
|---|---|---|
task |
string | stable task id |
lang |
string | English (en) or primary programming language |
category |
string | detailed recipe category |
split |
string | trajectory-disjoint train or val partition |
assistant_step |
int | 1-based target assistant turn |
assistant_steps |
int | assistant turns in the source trajectory |
target_message_index |
int | index of the final assistant target |
n_messages |
int | cumulative message count through the target |
messages |
list of objects | cumulative context ending at the target |
messages is native JSON.
Layout
Everything ships in one data file: traces.jsonl. It currently contains
2 cumulative next-step rows derived from 1 accepted
trajectories over disjoint train and validation tasks.
When training from the cumulative view, supervise only the final assistant message in each row. Supervising every assistant span would repeatedly overweight early steps because those spans recur as context in later prefixes.
Intended use
Supervised fine-tuning of instruction-following, tool-calling, and coding agents, plus analysis of multi-step planning, parallel calls, tool selection, state tracking, build-test-fix loops, and verification-driven completion.
Provenance
Source failures were generated with Claude Fable 5 (anthropic/claude-fable-5). The configured correction model made only narrowly eligible synthetic repairs, after which deterministic verification and the independent gpt-5.6-sol trace judge were run again. Source reasoning is preserved unchanged. Provider credentials, user keys, and host-identifying data are scrubbed before publication.
License
CC BY 4.0 — free for training, research, commercial products, modification, redistribution, and inclusion in other datasets or corpora, with attribution.
Suggested attribution:
Model Synthetic Corrections — https://huggingface.co/datasets/greghavens/fable-5-coding-and-debugging-traces-synthetic-corrections
- Downloads last month
- 38