You can deploy this model and start earning money today!

rhindsight/
lab21-2A202601903-qwen35-triage-vi

Name: lab21-2A202601903-qwen35-triage-vi
Availability: InStock

Views

No views yet

Lab 21 ? Qwen3.5-4B LoRA for Vietnamese ticket triage

Student: D??ng Minh Qu?n ? MSSV: 2A202601903

Public Option B submission for Track 3 ? Lab 21. This repository contains the primary LoRA adapter, every grader-readable evaluation artifact, and the completed report.

Measured results

Run	Target	Regression	Format	Latency (ms/sample)
Base + optimized prompt	0.765	0.7578	1.000	1026.6
LoRA fine-tune	0.980	0.6778	1.000	1452.5

Base model: unsloth/Qwen3.5-4B
Hardware: Tesla T4 14.6 GB, fp16
LoRA: text-linear, rank 16, alpha 32, 30 steps
Scientific verdict: FAILED because regression decreased by 0.080, beyond the 0.020 tolerance

The failed deployment gate is intentional and analyzed honestly in submission/REPORT.md. Grader-readable files are in results/, and the PEFT adapter is in adapters/correct/.

rhindsight/lab21-2A202601903-qwen35-triage-vi

Lab 21 ? Qwen3.5-4B LoRA for Vietnamese ticket triage

Measured results

rhindsight/
lab21-2A202601903-qwen35-triage-vi