Staging - #376
Merged
Merged
Conversation
This commit introduces a node filtering feature to enhance the retrieval process. Key additions include configuration options for enabling node filtering, setting filter parameters, and updating relevant functions to support filtering logic. The changes also include updates to the decision trace to log node filter actions and their outcomes, ensuring better tracking of retrieval decisions. Additionally, tests have been added to validate the new functionality.
…eatures This commit refactors the navigation configuration by removing unused parameters and features, including `navigate_max_steps`, `enable_depth0_oversize_to_dispatch`, and `depth0_oversize_char_limit`. The `RegionReport` class has also been eliminated, simplifying the codebase. Additionally, adjustments have been made to the decision trace and related functions to reflect these changes, ensuring a cleaner and more efficient navigation process.
This commit refines the TOC anchoring logic by ensuring that only resolvable outline destinations are consumed, improving the handling of null-page entries. The `_find_toc_text_matches` function is updated to match TOC keywords based on line-level containment rather than whole-line equality, enhancing accuracy. Additionally, the outline processing functions are restructured to retain no-page destinations, ensuring they are correctly represented in the output. Debugging and persistence mechanisms for outline roots are also introduced to support Stage-2 processing.
This commit updates the instructions for deriving the semantic title body from the given title. The changes emphasize the importance of stripping leading prefixes and trailing metadata qualifiers while ensuring that the semantic body is prioritized over any metadata-only queries. This refinement aims to enhance the accuracy of title processing in the document agent's structure.
…stat-readers fix(retrieval): use token hashes for map-unit frequency lookup
feat: TOC outline keep/null-page react + nav node filter
…cy-query perf(retrieval): drive frequency lookup from token hash
This commit removes the `filter_submap_char_limit` parameter from the navigation configuration and related functions, streamlining the retrieval process. Additionally, updates to the `render_submap_observation` function reflect this change by eliminating the character limit logic, ensuring that all matched nodes are displayed without truncation. Tests have been adjusted accordingly to validate the new behavior.
…-feedback fix: node-filter scope feedback and prompt cleanup
…y query logic This commit removes the `n_chunks` attribute from various classes and functions related to navigation and retrieval, streamlining the codebase. Additionally, it updates the frequency query logic to drive lookups from `document_map_unit_tokens`, enhancing performance and clarity in the retrieval process. Tests have been adjusted to reflect these changes.
Defer corpus map scoring until after the planner emits subgoals, cache those scores for harvest, and keep refine map-aware after lighting. Co-authored-by: Cursor <cursoragent@cursor.com>
…en-score refactor(retrieval): plan query-only then light from retrieval_query
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
…memory fix(deploy): keep production API at 2 GiB
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
3 participants
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.
No description provided.