contextboxai/Qwen3-1.7B-FC - Каталог нейросетей
Генерация текста

contextboxai/Qwen3-1.7B-FC

Добавлено:
contextboxai/Qwen3-1.7B-FC

A function calling model based on Qwen3-1.7B, fine-tuned using RLVR (Reinforcement Learning with Verifiable Rewards) to improve tool-use capabilities on the BFCL V3 benchmark. The fine-tuned model generates 36% fewer tokens while maintaining higher accuracy, thanks to: — Direct tool calls without verbose preambles — Concise refusal messages («None of the provided tools can answer this question») — Reduced reasoning blocks The model is trained to refuse appropriately using diverse negative samples: 1. Stage 1: Accuracy-focused training (V3) — Trained from Qwen3-1.7B base — Dataset: ~40K samples (stage2.parquet) — Reward: Correctness (1.0) + Format (0.1) + Efficiency (0.3) + Refusal (0.3) — Config: maxsteps=5000, LR=5e-7, temp=1.2 — Best checkpoint: step 100** (early stopping, highest accuracy) 2. Stage 2: Efficiency optimization (V4) — Loaded from Stage 1 checkpoint-100 — Focus: Reduce verbosity, discourage tags — Reward weights: Efficiency=1.0, Correctness=0.5, Format=0.1, Refusal=0.3 — Config: maxsteps=3000, LR=2e-7 — Selected checkpoint: step 1100 — Result**: 36% reduction in response tokens 1. Strong Refusal Penalty: -1.0 for calling tools when groundtruth = [] 2. Toucan…

Модальности:
Генерация текста

Области применения:
Диалог / чат Вызов функций (Tool use)


Задача: Генерация текста
Автор: contextboxai
Теги: qwen3, function-calling, tool-use, grpo, rl-fine-tuned, conversational, en, vi
Лайков: 4  |  Загрузок: 107

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.