Releases: WhiteMagic2014/gpt-magic
Releases · WhiteMagic2014/gpt-magic
v1.12.3
v1.12.2
Update: CreateChatCompletionRequest add new param store . Whether or not to store the output of this chat completion request for use in our model distillation or evals products.
v1.12.1
1.12.1
- New: gpt_o1 models have been added to the GptModel.class
v1.12.0
- Update: The URL field in all requests has been changed to the
final
type. - New: Support Uploads APIs
- New: Support invites APIs
- New: Support Users APIs
- New: Support Projects APIs
- New: Support ProjectUsers APIs
- New: Support ProjectServiceAccounts APIs
- New: Support ProjectApiKeys APIs
- New: Support AuditLogs APIs
1.11.0
v1.10.4
1.10.4
- Update: DefaultGptHttpUtil Class Changed to Singleton Pattern
Old
DefaultGptHttpUtil util = new DefaultGptHttpUtil();
New
DefaultGptHttpUtil util = DefaultGptHttpUtil.getInstance();
Old
DefaultGptHttpUtil util = new DefaultGptHttpUtil(true, 20000, 50000);
New
DefaultGptHttpUtil util = DefaultGptHttpUtil.getInstance();
util.init(true, 20000, 50000);
Additional Notes
- This change ensures that
DefaultGptHttpUtil
has only one instance throughout the application lifecycle, thereby
reducing memory usage and potential thread safety issues. - The
init
method provides flexibility to set the instance parameters, but it should be called before the first use of
theDefaultGptHttpUtil
instance.
v1.10.3
1.10.3
- Fixed Bug:Fixed a bug where CreateChatCompletionRequest, CreateRunRequest, and CreateThreadAndRunRequest could not be
properly requested when tools were not provided.
v1.10.1
1.10.1
- Update: Added support for file search customizations.
When using CreateVectorStoreFileRequest, you can specify the chunking_strategy parameter.
v1.10.0
1.10.0
- Update: Updated All Assistant-related APIs from version v1 to v2 (Assistant, Thread, Message, Run).
- Update: Updated RetrievalTool to FileSearchTool to adapt to the v2 version of Assistant.
- New: Added all APIs related to VectorStore required by
Assistant (VectorStores,VectorStoreFiles,VectorStoreFileBatches). - Update: Updated CreateChatCompletionRequest to sync with the latest official version (now supports seed, if specified,
our
system will make a best effort to sample deterministically. In stream mode, it supports returning usage, etc.). - Update: Updated RequestUtil utility class with a new method streamRequestV3, which supports using tool, usage,
multiple
choices return, logprob parameters, etc., in CreateChatCompletion in stream mode. This method can almost perfectly
convert the chunk object returned by CreateChatCompletionRequest into ChatCompletion.
v1.9.7
1.9.7
- Update: Enhanced the GptModel and removed some models that are no longer in use.
- New: Support Batch Apis
- Update: All POJOs have implemented the Serializable interface.
- Optimize: The assignment method for the purpose field in UploadFileRequest has been optimized.
- Update: In RetrieveFileContentRequest, the send method is no longer recommended. Instead, use the sendForContent method to directly obtain the file content.