feat: v0.1.70 — AutoResearch optimization + eval infrastructure#19
Merged
kargarisaac merged 3 commits intomainfrom Mar 28, 2026
Merged
feat: v0.1.70 — AutoResearch optimization + eval infrastructure#19kargarisaac merged 3 commits intomainfrom
kargarisaac merged 3 commits intomainfrom
Conversation
…ipts - Removed the "explorer" section from the required configurations in `configure_dspy_from_eval` and related lifecycle functions. - Introduced new evaluation scripts: `run_dedup.py`, `run_maintain.py`, `run_search.py`, and `run_tool_selection.py` for assessing deduplication accuracy, maintenance quality, search relevance, and tool selection accuracy against golden datasets. - Enhanced the README documentation to include detailed descriptions of the new evaluation pipelines and their usage. - Updated scoring functions in `scores.py` to support new evaluation metrics and composite scoring for the LerimBench. - Added judge prompts for deduplication, maintenance, search, and tool selection evaluations to standardize quality assessments.
AutoResearch-style optimization loop (14 experiments, 7 kept): - ChainOfThought for DSPy extraction (Predict → CoT, biggest single win) - Explicit dedup thresholds (0.7/0.4) in sync prompt - Behavioral guidance in batch_dedup_candidates tool description - Improved MemoryCandidate schema field descriptions (title/body format) - Title format guidance in extraction signature - Raised post-extraction body filter from 30 to 50 chars Eval infrastructure extensions: - check_extraction_assertions() for deterministic golden-case scoring - check_summarization_assertions() for summary validation - _fuzzy_title_match() for dedup accuracy with Jaccard similarity - LerimBenchScore 7-dimension composite scoring - 59 scoring tests (all passing) Composite: 0.608 → 0.855 (+41%) Extraction: 0.693 → 0.877 (+27%) Dedup: 0.278 → 0.722 (+160%) Search: 0.905 (unchanged) Maintain: 1.000 (unchanged) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Dashboard: `lerim dashboard` shows transition message to lerim.dev - Ask prompt: replace stale Codex tool refs with memory_search/read_file - Clean up ResponsesProxy reference in src/lerim/README.md - Update test assertion for ask prompt (codex → memory_search) - Add v0.1.70 section to CHANGELOG.md - Update README dashboard section - Bump version 0.1.69 → 0.1.70 Co-Authored-By: Claude Opus 4.6 (1M context) <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
lerim dashboardshows "coming soon" message, web UI moves to lerim.devKey changes
Quality (AutoResearch Layer 1)
dspy.Predict→dspy.ChainOfThoughtfor extraction (single biggest win)Eval infrastructure
run_dedup.py,run_maintain.py,run_search.py,run_tool_selection.pyLerimBenchScore7-dimension composite with configurable weights_fuzzy_title_match)--golden-dirflagCleanup
lerim dashboard→ transition message (no more local dashboard)Test plan
lerim dashboardshows transition message🤖 Generated with Claude Code