This model is a LoRA fine-tune of Qwen/Qwen3-0.6B for routing customer support tickets into the top 12 intents from bitext/Bitext-customer-support-llm-chatbot-training-dataset.

Task setup

Input column: instruction
Label column: intent
Data filter: top 12 intents by frequency in the source train split
Output format: strict JSON only

{"intent":"<label>","confidence":0.93,"reason":"<short reason>"}

Train / eval split

Train examples: 10194
Eval examples: 1799
Split method: stratified random split with seed 42

Evaluation metrics

accuracy: 0.9994
valid_json_rate: 1.0000
schema_pass_rate: 1.0000
accuracy_on_schema: 0.9994

Allowed intents

[ "check_invoice", "complaint", "contact_customer_service", "edit_account", "switch_account", "check_payment_methods", "contact_human_agent", "delivery_period", "get_invoice", "newsletter_subscription", "payment_issue", "registration_problems" ]

woojun-jung/qwen3-0.6b-bitext-ticket-router

Qwen3 0.6B Bitext Ticket Router

Task setup

Train / eval split

Evaluation metrics

Allowed intents

woojun-jung/
qwen3-0.6b-bitext-ticket-router