Qwen Meetup, Function Calling Harness, turning 6.75% to 100%

Name: Qwen Meetup, Function Calling Harness, turning 6.75% to 100%
Availability: InStock
Author: samchon

by samchon·Mar 27, 2026·1 point·0 comments

Visit Project View on HN

AI Analysis

●●SolidBig BrainNiche Gem

Compiler-level validation turns Qwen's 6.75% structured output success rate into 100%.

Strengths

•Typia infrastructure compiles TypeScript types into validators instead of runtime JSON schema checks.
•Self-healing loops recover from double-stringify bugs that break most Qwen 3.5 function calls.
•Benchmarks specifically target recursive union types where industry consensus says function calling fails.

Weaknesses

•Iterative validation loops add latency and token costs compared to single-shot constrained decoding.
•LLM structured output validation is a crowded category with established tools like Instructor.

Post Description

I was personally invited by the Qwen team to speak at Qwen Meetup Korea, and got to present locally here in Korea yesterday — pretty honored to have been reached out to directly.

The talk was about how I got function calling to work reliably on deeply recursive union types — the stuff the industry generally says doesn't work. With `qwen3-coder-next`, first-try success rate was 6.75%. And the entire Qwen 3.5 model family was hitting 0% on union types due to a consistent double-stringify bug. Both ended up at 100%.

Slides are also available here: https://autobe.dev/seminars/20260326-qwen-meetup-korea.pptx — speaker notes are written inside as slide notes if you'd like the full narrative behind each slide.