feat(summarize): hierarchical multi-level conversation summarization (#2)#63
Merged
Merged
Conversation
Implements issue #2. Compresses conversation turns progressively as they age using extractive techniques (no LLM required): - LevelFull (0): original content, recent turns - LevelParagraph (1): first paragraph + code blocks preserved - LevelSentence (2): first 1-2 sentences - LevelKeywords (3): top-12 significant words - LevelEvicted (4): dropped entirely when budget is exhausted Key behaviours: - PreserveRecent: N most recent turns always kept at LevelFull - ImportanceThreshold: turns above threshold never exceed LevelParagraph - ScoreImportance: heuristic scoring (code blocks, error keywords, role) - enforceTokenBudget: second pass with progressive eviction when over budget - estimateTokens: 4-chars-per-token approximation Co-authored-by: Ona <[email protected]>
Siddhant-K-code
force-pushed
the
feat/2-hierarchical-summarization
branch
from
fccd972 to
04cb5a1
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes #2
Summary
Adds
pkg/summarize— a rule-based hierarchical summarizer that compresses conversation turns progressively as they age, without requiring an LLM.Compression levels
LevelFullLevelParagraphLevelSentenceLevelKeywordsLevelEvictedKey behaviours
PreserveRecent: N most recent turns always kept atLevelFullImportanceThreshold: turns scoring above threshold never exceedLevelParagraphScoreImportance: heuristic scoring — code blocks (+0.4), error keywords (+0.3), decision keywords (+0.2), system role (1.0)enforceTokenBudget: second pass with progressive eviction when still over budget after first passBuildFromGoFiles(inpkg/graph): 4-chars-per-token approximation for token estimationFiles
pkg/summarize/summarize.go—Turn,Level,SummarizeOptions,Summarizerinterfacepkg/summarize/importance.go—ScoreImportance,ScoreTurns,estimateTokenspkg/summarize/hierarchy.go—HierarchicalSummarizerwith all compression logicpkg/summarize/summarize_test.go— 10 tests covering all major behaviours