feat: route self-evolution through Hermes Codex OAuth#92
Open
stephenschoettler wants to merge 10 commits into
Open
feat: route self-evolution through Hermes Codex OAuth#92stephenschoettler wants to merge 10 commits into
stephenschoettler wants to merge 10 commits into
Conversation
- All dspy.LM() calls: num_retries=8, timeout=120 - LiteLLM backoff env vars: INITIAL_RETRY_DELAY=5, MAX_RETRY_DELAY=60 - Switch nightly from anthropic/sonnet to openai/gpt-4.1 (no rate limit conflicts) - Robust JSON parsing in dataset_builder (handles trailing commas, unescaped newlines) - Tested: gpt-4.1 optimizer yielded +10.6% improvement, o3 optimizer yielded 0%
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
openai-codex/*model stringsmax_full_evalsso policyiterationsmaps to a bounded run instead of GEPA's large hosted-model auto presetno-changeeven when judge scores improve, so no-op candidates are not reviewable/deployableCoordination with existing work
I found overlapping open work before opening this PR:
dspy-local/CodexLMand~/.codex/auth.json.chatgpt/*OAuth backend.This PR is narrower/different in two ways: it uses Hermes Agent's existing
openai-codexOAuth credential/runtime path directly, and it hardens the scheduled cron governance path so a successful score cannot mask a byte-identical no-op artifact.Validation
python -m pytest tests/core/test_dspy_lm_codex.py tests/test_nightly_evolve_cron.py tests/skills/test_evolve_skill_budget.py -qpython -m pytest -q->165 passed/usr/bin/python /home/w0lf/.hermes/scripts/nightly-self-evolution.py --profile babbage --skill goal-planning --dry-runmake_dspy_lm('openai-codex/gpt-5.4-mini')returnedOKstatus: no-change,gate: no-material-diff,review: rejected,applied: noSafety
auto_applyremainsfalsemainwere avoided; this is on a named branch