Per-category gate ladders and 4-round training split - #175
Open
lukeinglis wants to merge 1 commit into
Open
Conversation
Replace the single overall-index ladder with two 10-rung ladders (dataset + market) so a candidate cannot pass by improving one category while regressing the other — the failure mode that made gen-0 calibration look like an improvement. - gate_baseline.json stores overall/dataset/market indices; old single-value format skips gracefully - GATE_ROUNDS_OVERRIDE env var lets holdout rounds be evaluated without modifying committed gate config - Expand training rounds to 4 (2026-03-01 .. 2026-04-12), all post-Luna-cutoff - Baseline values are placeholders pending re-baseline with Luna Co-Authored-By: Claude Code <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
GATE_ROUNDS_OVERRIDEenv var so holdout rounds can be evaluated without touching committed gate configgate_baseline.jsonnow stores per-category baselines (currently 50/50/50 placeholders — real values require re-baseline with Luna)Context
Issue #174 (gen-1 gate redesign). Part of the task/benchmark design work before building a factory mode for tournament-track forecaster improvement.
Note
Baseline re-baseline is blocked on OpenAI credits (Sept 4 attempt failed silently — all forecasts missing, scored 50.0). An empty-eval assertion will follow in a separate commit so dead API keys fail loudly.
🤖 Generated with Claude Code