Releases: samchon/openapi
Releases · samchon/openapi
v2.2.1
v2.2.0
As I've understood that ChatGPT's strict mode is not perfect, and found much better way by giving validation feedback with typia.validate<T>()
function, I've changed the ChatGPT strict mode to be configurable, and also changed ChatGPT function calling (or structured output) composition strategy.
From now on, ChatGPT schema also supports the optional properties and dynamic key typed object unless the strict mode configured.
Advised by @AcrylicShrimp
What's Changed
- Enhance function calling test functions by @samchon in #111
- Make ChatGPT strict mode configurable. by @samchon in #112
Full Changelog: v2.1.2...v2.2.0
v2.1.2
v2.0.4
v2.0.3
v2.0.2
v2.0.1
v2.0.0
Middle layer ILlmSchema
has been divided to actual LLM provider models:
IChatGptSchema
: OpenAI ChatGPTIClaudeSchema
: Anthropic ClaudeIGeminiSchema
: Google GeminiILlamaSchema
: Meta Llama

What's Changed
- Wiriting Gemini schema and its type checker by @samchon in #58
- Add ChatGPT schema by @samchon in #59
GeminiConverter.schema()
function by @samchon in #60- Recursive level on LLM schemas by @samchon in #61
- LLM schemas of OpenAPI v3 and v3.1 by @samchon in #62
- Fix bug of recursive ChatGPT schema conversion by @samchon in #63
IChatGptSchema.IOneOf.discriminator
by @samchon in #64HttpLlm.application()
bans themultipart/form-data
type. by @samchon in #65- VSCode Debugger Configuration by @samchon in #70
- Adapt typia v7 and nestia v4 for testing. by @samchon in #71
- The exact
IChatGptSchema
and LLM function calling applilcation schemas. by @samchon in #72 @format uuid
like description forIChatGptSchema
and documetation. by @samchon in #73ILlmApplication.IChatGptOptions
by @samchon in #74- More example cases for the ChatGPT function calling. by @samchon in #75
- Enhance function callling prompt examples by @samchon in #76
- Enhance reference type description rule by @samchon in #78
- Completed Google Gemini schema by @samchon in #79
- Strict JSON/LLM schema of object type by @samchon in #81
- LLM schema conversion layer by @samchon in #82
- Export
ILlmSchema
by @samchon in #83 - Support new LLM provider, "Anthropic Claude" by @samchon in #84
- Enhance function calling test programs. by @samchon in #85
- Ban constraint fields
IChatGptSchema
. by @samchon in #86 - Started supporting Meta Llama schema
ILlmSchema
by @samchon in #87 - Update README with example codes by @samchon in #89
- Entire refactoring for concenstrate to the
ILlmSchema
. by @samchon in #90 - Allow
additionalProperties
in Claude and Llama. by @samchon in #91 - Add test cases of failures by @samchon in #92
- Fix
gemini
schema'soneOf
detection logic. by @samchon in #93 ILlmSchema.IParameters.additionalProperties
to befalse
. by @samchon in #94- Precise
ILlmFunction.ISeparated<Model>
by @samchon in #95 - Make
constraint
option to be defaulttrue
by @samchon in #96 - Strong test programs for LLM schema composers and refactorings by @samchon in #97
Try<T, E>
pattern to LLM related functions by @samchon in #98- Fix
IResult.ISuccess.value
. by @samchon in #99 - Completed v2 publishing by @samchon in #100
Full Changelog: v1.1.2...v2.0.0