Experiment · pace planner

Pace planner v6.1 four-adapter repair sweep

A four-way score collapse traced to prompt mismatch between training and serving. Pin the prompt, schema, and grammar triple across train, serve, and eval before believing any comparison. Retained outcome: worked with caveat.

Outcome
worked with caveat
Family
pace planner
Evidence confidence
exact
Record
training run

Evidence readout

10/19, 8/19, 2/19, 0/19, then 14/19 at 100 steps and 17/19 at 300 steps, reaching 19/19 after a serving-side minLength fix; 4/15 (26.7%) on fm-fixtures-v2

What the attempt taught

A four-way score collapse traced to prompt mismatch between training and serving. Pin the prompt, schema, and grammar triple across train, serve, and eval before believing any comparison.

Why it stopped or stayed bounded

The apparent collapse was two harness bugs, not training: SFT rows omitted the Pace system prompt that serve and eval both included, and a brace-counting JSON extractor was faulty.

Method and scope

  • Experiment family: pace planner.
  • Record kind: training run.
  • Objective: planner oos refusal.
  • Methods: sft, lora.
  • Base models: qwen3-0.6b.
  • Recorded data rows: 760.

Disposition

Recover the full sweep with `git show d1efcb8^:docs/prds/factory-pace-planner-v6_1.md` before repeating any step-count ablation; it is the only one in the series.

This retained disposition is historical evidence, not authorization to restart the experiment. A new run needs a fresh question, frozen evaluator, explicit resource budget, and scoped tracking issue.

How to read this dossier

This canonical page separates the retained repository decision from the material that informed it. A source may describe an external claim, historical measurement, local observation, or planned procedure; those evidence classes are not interchangeable. Follow the provenance links before reusing a number or method.

Status and disposition describe what PostTrainLLM retained when this record was indexed. They are not a live product promise, a newly run benchmark, or permission to restart historical work. Unknown or unmeasured fields remain unknown rather than being treated as zero.

To reuse the record, first name the exact claim you need and trace it to the linked source. Then check whether the original environment, model revision, data split, evaluator, hardware, and budget match the proposed use. If they do not, treat the record as a hypothesis or design reference and run the smallest fresh comparison that can falsify it. Preserve negative outcomes and regressions beside any improvement; a local win on one slice does not silently become a general capability claim.

Source provenance

The normalized record comes from docs/attempts.json. The links below are the tracked evidence and explanatory sources preserved with the record.