fix(adk): bound summarization model input - #1165
Open
March-77 wants to merge 1 commit into
Open
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What type of PR is this?
fix
Check the PR title.
(Optional) Translate the PR title into Chinese.
fix(adk): 限制摘要模型的输入长度
(Optional) More detailed description for this PR(en: English/zh: Chinese).
en:
When conversation history crosses the summarization threshold, the default input builder currently forwards the complete over-limit history to the summarization model. If that model has the same or a smaller context window, the recovery mechanism fails with another context-length error.
This change uses
Trigger.ContextTokensas the input budget for the default summary request. After reserving space for the system and user summary instructions, it keeps the newest contiguous message window that fits and evicts older messages. Keeping a suffix preserves current user intent and tool-call/result ordering. CustomGenModelInputcallbacks remain untouched and retain full control.It also fixes a related trigger edge case: configuring only
ContextMessagespreviously leftContextTokensat zero, causing any non-empty history below the message threshold to trigger anyway. A disabled token condition now falls back to the documented default 160k threshold.User impact: default summarization no longer retries an already oversized history unchanged, and message-only trigger configurations behave as documented.
Compatibility/risk: no new API. The window applies only to the default model-input builder; users needing different retention semantics can continue using
GenModelInput.Validation:
go test ./adk/middlewares/summarization -run "TestMiddlewareShouldSummarize|TestMiddlewareBuildSummarizationModelInput|TestWindowMessagesByTokenLimit" -count=10go test -race ./adk/middlewares/summarization -run "TestMiddlewareShouldSummarize|TestMiddlewareBuildSummarizationModelInput|TestWindowMessagesByTokenLimit" -count=1go test ./adk/middlewares/summarizationgo test ./adkgo test ./...go vet ./...golangci-lintcould not be completed locally because the pinned runner repeatedly stalled while downloadinggithub.com/MirrexOne/unqueryvet v1.5.4.(Optional) Which issue(s) this PR fixes:
Fixes #940