Releases: betalgo/openai
7.3.0
- Updated Moderation categories as reported by @dmki.
- Breaking Changes:
- Introduced the use of
EnsureStatusCode()
after making requests.Please adjust your code accordingly for handling failure cases. Thanks to @miroljub1995 for reporting. - Previously, we used to override paths in the base domain, but this behavior has now changed. If you were using
abc.com/mypath
as the base domain, we used to ignore/mypath
. This will no longer be the case, and the code will now respect/mypath
. Thanks to @Hzw576816 for reporting.
- Introduced the use of
What's Changed
- removed finetnuning cleanup function by @kayhantolga in #385
- readme update by @kayhantolga in #386
- 362 missing moderation categories by @kayhantolga in #387
- closes#367 Fix: Prevent BaseDomain override in HttpClient requests by @kayhantolga in #388
- closes#277 Fix: implemented EnsureSuccessStatusCode after before read… by @kayhantolga in #389
- 7.3.0 by @kayhantolga in #392
Full Changelog: v7.2.0...v7.3.0
v7.2.0
Added Chatgpt Finetununig support thanks to @aghimir3
Default Azure Openai version increased thanks to @mac8005
Fixed Azure Openai Audio endpoint thanks to @mac8005
What's Changed
- Update Azure Endpoint Formatting for Audio and Azure API Version in OpenAI SDK (Whisper on Azure) by @mac8005 in #379
- Add support for new fine tuning job from Open AI by @aghimir3 in #368
- 7.2.0 by @kayhantolga in #382
New Contributors
Full Changelog: v7.1.5...v7.2.0
v7.1.5
Added error handling for PlatformNotSupportedException in PostAsStreamAsync when using HttpClient.Send, now falls back to SendRequestPreNet6 for compatibility on platforms like MAUI, Mac. Thanks to @Almis90
We now have a function caller describe method that automatically generates function descriptions. This method is available in the utilities library. Thanks to @vbandi
What's Changed
- readme update by @kayhantolga in #354
- version bump by @kayhantolga in #355
Full Changelog: v7.0.3...v7.1.5
What's Changed
- Function caller helpers - create definition from attributes, invoke the function as specified by LLM by @vbandi in #345
- Add fallback method for HttpClient.Send on unsupported platforms by @Almis90 in #316
- Betalgo.OpenAI 7.1.4 ---- Betalgo.OpenAI.Utilities 7.0.2 by @kayhantolga in #350
- Betalgo.OpenAI 7.1.4 -- Betalgo.OpenAI.Utilities 7.0.2 by @kayhantolga in #351
New Contributors
Full Changelog: v7.1.3...v7.1.4
v7.1.3
7.1.3
- This release was a bit late and took longer than expected due to a couple of reasons. The future was quite big, and I couldn't cover all possibilities. However, I believe I have covered most of the function definitions (with some details missing). Additionally, I added an option to build it manually. If you don't know what I mean, you don't need to worry. I plan to cover the rest of the function definition in the next release. Until then, you can discover this by playing in the playground or in the source code. This version also support using other libraries to export your function definition.
- We now have support for functions! Big cheers to @rzubek for completing most of this feature.
- Additionally, we have made bug fixes and improvements. Thanks to @choshinyoung, @yt3trees, @WeihanLi, @N0ker, and all the bug reporters. (Apologies if I missed any names. Please let me know if I missed your name and you have a commit.)
7.1.2-beta
7.1.0-beta
- Function Calling: We're releasing this version to bring in a new feature that lets you call functions faster. But remember, this version might not be perfectly stable and we might change it a lot later. A big shout-out to @rzubek for helping us add this feature. Although I liked his work, I didn't have enough time to look into it thoroughly. Still, the tests I did showed it was working, so I decided to add his feature to our code. This lets everyone use it now. Even though I'm busy moving houses and didn't have much time, seeing @rzubek's help made things a lot easier for me.
- Support for New Models: This update also includes support for new models that OpenAI recently launched. I've also changed the naming style to match OpenAI's. Model names will no longer start with 'chat'; instead, they'll start with 'gpt_3_5' and so on.
What's Changed
- Feature/plugins by @kayhantolga in #280
- Betalgo.OpenAI.Utilities by @kayhantolga in #281
- fix by @kayhantolga in #282
- Betalgo.OpenAI.Utilities by @kayhantolga in #283
- path fix by @kayhantolga in #284
- Updated libraries to latest versions. by @kayhantolga in #288
- v 7.0.1 by @kayhantolga in #289
- Added support for the new function calling capability in Chat Completions API by @rzubek in #300
- Update Readme.md by @N0ker in #297
- 7.1.0 -beta by @kayhantolga in #301
- sync by @kayhantolga in #309
- A couple of small fixes for the 7.1 beta by @rzubek in #302
- Dev v7.1.2 beta by @kayhantolga in #310
- Update ChatCompletionCreateRequest model comment by @WeihanLi in #332
- Some refactoring about functions by @kayhantolga in #334
- Several improvements / fixes for functions by @rzubek in #320
- Enable to set FunctionCall to AssistantRole's ChatMessage by @yt3trees in #313
- Version 7.1.3 by @kayhantolga in #337
- v7.1.3 by @kayhantolga in #341
New Contributors
- @rzubek made their first contribution in #300
- @N0ker made their first contribution in #297
- @yt3trees made their first contribution in #313
Full Changelog: v7.0.0...v7.1.3
v7.0.0
7.0.0
- The code now supports .NET 7.0. Big cheers to @BroMarduk for making this happen.
- The library now automatically disposes of the Httpclient when it's created by the constructor. This feature is thanks to @BroMarduk.
- New support has been added for using more than one instance at the same time. Check out this link for more details. Thanks to @remixtedi for bringing this to my attention.
- A lot of small improvements have been done by @BroMarduk.
- Breaking Changes 😢
- I've removed 'GPT3' from the namespace, so you might need to modify some aspects of your project. But don't worry, it's pretty simple! For instance, instead of writing
using OpenAI.GPT3.Interfaces
, you'll now writeusing OpenAI.Interfaces
. - The order of the OpenAI constructor parameters has changed. It now takes 'options' first, then 'httpclient'.
//Before var openAiService = new OpenAIService(httpClient, options); //Now var openAiService = new OpenAIService(options, httpClient);
- I've removed 'GPT3' from the namespace, so you might need to modify some aspects of your project. But don't worry, it's pretty simple! For instance, instead of writing
What's Changed
- Dispose HttpClient when created internally. by @BroMarduk in #243
- Fixes for Exception Warnings by @BroMarduk in #262
- Fixes for Warnings where CREFS are used instead of HREFS by @BroMarduk in #264
- DotNET 7 Project Support by @BroMarduk in #261
- Added support for multiple instance by @kayhantolga in #258
- removed GPT3 from namespace by @kayhantolga in #269
- nuget symbols by @kayhantolga in #272
- Version 7.0.0 by @kayhantolga in #271
- version bump for v7.0.0 by @kayhantolga in #275
New Contributors
- @BroMarduk made their first contribution in #243
Full Changelog: v6.8.6...v7.0.0
6.8.6
6.8.6
- Updated Azure OpenAI default API version to the preview version to support ChatGPT. thanks to all issue reporters
- Added support for an optional chat
name
field. thanks to @shanepowell - Breaking Change
FineTuneCreateRequest.PromptLossWeight
converto to float thanks to @JohnJ0808
What's Changed
- Add support for the optional "name" into the ChatMessage class. by @shanepowell in #164
- 240 finetunecreaterequest promptlossweight type float instead of int by @kayhantolga in #250
- 181 question get started with azureopenai by @kayhantolga in #249
- 6.8.6 by @kayhantolga in #252
Full Changelog: v6.8.5...v6.8.6
6.8.5
What's Changed
- Simplify NoWarn by @WeihanLi in #216
- Patch json field name of self-harm for moderation end point by @digitalvir in #210
- Use
BindConfiguration
and avoid building a service provider. by @Swimburger in #225 - Add
FileStream
property for audio file in addition to byte array by @Swimburger in #224 - Fix AudioCreateTranscriptionResponse JSON serialization by @Swimburger in #223
- v6.8.5 by @kayhantolga in #228
New Contributors
- @WeihanLi made their first contribution in #216
- @digitalvir made their first contribution in #210
- @Swimburger made their first contribution in #225
Full Changelog: v6.8.4...v6.8.5
6.8.3
Breaking Changes:
I am going to update library namespace from Betalgo.OpenAI.GPT3 to OpenAI.GPT3. This is the first time I am trying to update my nuget packageId. If something broken, please be patient. I will be fixing it soon. Reverted namespace change, maybe next time.
Small Typo change on model name Model.GPT4 to Model.GPT_4
ServiceCollection.AddOpenAIService(); now returns IHttpClientBuilder which means it allows you to play with httpclient object. Thanks for all the reporters and @LGinC. Here is a little sample
ServiceCollection.AddOpenAIService()
.ConfigurePrimaryHttpMessageHandler((s => new HttpClientHandler
{
Proxy = new WebProxy("1.1.1.1:1010"),
});
6.8.1
6.8.1
- Breaking Change: Typo fixed in Content Moderation CategoryScores, changing
Sexualminors
toSexualMinors
. Thanks to @HowToDoThis. - Tokenizer changes thanks to @IS4Code.
- Performance improvement
- Introduced a new method
TokenCount
that returns the number of tokens instead of a list. - Breaking Change: Removed overridden methods that were basically string conversions.
I think these methods were not used much and it is fairly easy to do these conversions outside of the method.
If you disagree, let me know and I can consider adding them back.
What's Changed
- CreateModerationResponse.Categories fixed Typo mistake by @kayhantolga in #184
- Tokenizer Improvements by @kayhantolga in #185
- v6.8.1 by @kayhantolga in #187
Full Changelog: v6.8.0...v6.8.1
6.8.0
6.8.0
What's Changed
- Add support for .Net Standard 2.0 by @pdcruze in #141
- 6.8.0 by @kayhantolga in #162
New Contributors
Full Changelog: v6.7.3...v6.8.0