Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
### Features * [PrivateAggregation] Add PrivateAggregateReportingResponse in Auction Service response * [PrivateAggregation] Append private aggregation wrapper to existing seller's wrapper * Add `common.privateAggregation.createContribution` function for creating a contribution object. * Add `common.privateAggregation.ReservedEvent` object that holds reserved events string constants. * Add auction_service.privateAggregation.contributeToHistogram/OnEvent * Add bucket protos (Bucket128Bit, BucketOffset, SignalBucket, PrivateAggregationBucket) and add bucket and value fields to PrivateAggregateContribution * Add checksum capability to model loading * Add config flag enable_private_aggregate_reporting to bidding service's runtime_config * Add ConvertSignalValueToInt for calculating final value from contribution's base value, scale, and offset. * Add enable_private_aggregate_reporting for GetSellerWrappedCode and unit tests that uses GetSellerWrappedCode * Add event code and private aggregation objects for contributeToHistogramOnEvent, also added tests for event field * Add helper functions convertEventToInt and convertBaseValueToInt and tests for them for contributeToHistogramOnEvent for bidding service * Add Inference Metrics like count, duration, error and size to TF & Pytorch sidecars. * add option to switch DebugInfo in secure invoke encrypt * Add PrivateAggregateContribution proto with event field and enum EventType to the .proto file * Add proto messages for SignalValue, PrivateAggregationValue, and BaseValue enum. * Add wrapper and test files with method headers for contributeToHistogramOnEvent for bidding service * Added `isValidCustomEvent` validation function for common.privateAggregation * Added isValidValue validation function for common.privateAggregation * Added new files for JS private aggregation util for bidding service * Change js helper convertEventToInt to mapEventToEnum to return string corresponding to enum instead of int * contributeToHistogram with test for bidding service * contributeToHistogramOnEvent with tests, using common private aggregation helpers * Convert Private Aggregation wrapper functions from .js to C++ string * Create a periodic model fetcher library for inference * Declare inference JS error schema * Declare schema for model fetching metadata file * Enable blob fetcher to only fetch blobs with given prefixes * Enable dynamic model loading for inference * Enable Service Mesh on AWS by Default * export non-privacy log through otel with safe system context * Force chaffing to enabled for prod builds on BFE + update common repo dep * Instrument inference JS error for the PyTorch sidecar * Load models using model metadata config for the periodic model fetcher * Make chaffing flag configurable via Terraform * move ig metric to bfe * Propagate inference error back to JS caller * Unit test for loading parsing fake contributions in Roma for bidding service * Update Demo Terraform Configs with values for integrating TEE KV Servers into Mesh * Write a fake generateBid() script which calls privateAggregation.contributeToHistogramOnEvent(event, <fake contribution>) ### Bug Fixes * Add createContribution inside contributeToHistogram/OnEvent and fixed typo with createContribution's value conditional statement. * Allow up to 15 buyers when chaffing enabled * chaffing bugs ([027d5bb]( )), closes [/github.com/abseil/abseil-cpp/blob/master/absl/container/internal/raw_hash_set.h#L1572]( ) * Changed raw string delimiter JSCODE to JS_CODE for consistency * Correct output_filter typo * Don't set chaff size for non-chaff requests * Fix bugs in reportWin URL validation in auction service * Fix release notes by adding a dedicated tag on main to generate the changelog * truncate curl failure info for fail count metric * Update SUTs to include and test Experiment Group ID ### Documentation * Add PAS input example to BFE * Document how to deployment B&A without inference * Update inference_sidecar README with model size limits * Updated expected format for INFERENCE_MODEL_BUCKET_PATHS in README Bug: N/A GitOrigin-RevId: 0de077cc59c002638794990cf88a91191fbf1e97 Change-Id: If1185f7873c3afd74c6e78406c6ec084110f89d6
- Loading branch information