📦 Release: v0.1.0-rc.1 #280
Merged
Conversation
This file contains 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
- Initing a new type of workflow, a streaming (async) routing workflow using the Streaming Chat API as an example - Updated the Bruno collection - Updated the LanguageModel API to include `ChatStream()` and `SupportChatStream()` methods - Get the streaming router working - Implemented SSE event parsing to be able to work with OpenAI streaming chat API - Integrated OpenAI chat streaming into the Glide's streaming chat API - Covered the happy workflow by tests
…tStream (#166) - Separated sync and streaming chat schemas - Extracted assumptions on where to find latency from routing strategies to a separate `LatencyGetters` that can be different for different models/workflows - Elaborated the client provider `chatStream()` interface. Clients now expose a response channel instead of being provided with by caller - Connected the stream chat workflow to latency & health tracking - Refined the `chatStream()` method of clients to return a stream struct - Separated latency tracking of the streaming workflow from the sync chat workflow - defined a new `HealthTracker` to incorporate all health tracking logic
- Separated chat & chat stream request schemas - introduced a new finish reason field - added metadata to stream chat response - allow to attach some metadata to a chat stream request and then attach it to each chat stream chunk - adjusted error message schema to include request ID and metadata
…age (#184) - Fixed the header where Anthropic API key is passed - Started propagating token usage of Anthropic requests - Corrected the TokenUsage interface by changing count field to integers from floats
…penAI, Azure and Cohere (#194) - text length bound passed in request params - content moderation/toxicity - Cohere streaming workflow doesn't seem to be working as errMapper was not really initialized. I have fixed that in this PR - Cohere now ignores stream chunk types that Glide doesn't support like citation related stuff - Cohere stream chunks are not set with the correct model name (e.g. some placeholder was used before)
…re chat streams correctly (#201) - implementing a custom stream reader to correctly handle Cohere streams - Start handling the stream-start event to propagate generationID to all following chunks
…n case of some errors (#203) - Passed RouterID and ModelID information in the chat stream messages - Introduced a new ChatStreamMessage type that joins both chunk and error messages. Removed unneeded context from provider chatStream structs - defined a set of possible error codes during chat streaming - started simplifying logging by using context-based loggers - Introduced finish_reason on the error schema
- Fixed validation of nested arrays, so it can now reach all structures including provider params - Removed ChatHistory & ConversationID fields from the params - Added a bunch of other params like max_tokens, penalties, k, p, etc. - Added validations to some params
…g swagger.yaml file (#211) This change fixes panics like "./docs/swagger.yaml is not found"
# Conflicts: # README.md # docs/docs.go # docs/swagger.json # docs/swagger.yaml # go.mod # go.sum # pkg/api/http/handlers.go # pkg/api/http/server.go # pkg/api/schemas/chat_stream.go # pkg/gateway.go # pkg/providers/azureopenai/chat_stream.go # pkg/providers/azureopenai/client.go # pkg/providers/cohere/chat.go # pkg/providers/cohere/chat_stream.go # pkg/providers/cohere/chat_stream_test.go # pkg/providers/cohere/client.go # pkg/providers/cohere/config.go # pkg/providers/cohere/schemas.go # pkg/providers/cohere/testdata/chat_stream.success.txt # pkg/providers/lang.go # pkg/providers/openai/chat.go # pkg/providers/openai/chat_stream.go # pkg/providers/openai/client.go # pkg/providers/provider.go # pkg/providers/testing/lang.go # pkg/providers/testing/models.go # pkg/routers/config.go # pkg/routers/router.go # pkg/routers/router_test.go
…rics and OTEL Collector (#225)
We use go.opentelemetry.io/contrib/exporters/autoexport for standard loading exporter configurations via env variables.
… secrets in a secure way (#244)
…s to give clients more context around the error (#236) - Introduced a new error type to hold useful context like HTTP response status, error name, message - If all providers are unavailable, we are not throwing 500 error anymore but 503 - Start throwing unknown_error with 500 status on unexpected exceptions - Predefined all static HTTP errors instead of creating them every time they occur - Introduced the name field on the error schema - Changed the req/response schema to snake_case (hopefully, to stick with it forever) - Removed Bruno collections (it doesn't cover all our needs like websocket or gRPC protocol) - Moved all schemas to `api/schema` package - Made router list API opaque - Changed the field name for overrides not to clash with defined statements in some languages
- Introduced a new concept/struct `ChatParams` that contains all param overrides for the specific modelName/modelID - Adjusted the LangModel interface to rely on `ChatParams` rather than the original request schema for both sync and stream chat API - Standardize on the chat message structure with two fields. Removed all duplicated structures - Fixed Ollama's broken/half-backed tests
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #280 +/- ##
==========================================
- Coverage 66.98% 65.98% -1.00%
==========================================
Files 78 83 +5
Lines 3577 3634 +57
==========================================
+ Hits 2396 2398 +2
- Misses 1054 1114 +60
+ Partials 127 122 -5 ☔ View full report in Codecov by Sentry. |
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.
The first major update with breaking changes to the language chat schemas
and begging of work on instrumenting the gateway with OpenTelemetry.
Added
Changed
Breaking Changes
Fixed
Security
Miscellaneous