miitarou
/

qwen3-4b-structeval-v12

Text Generation

structured-output

Model card Files Files and versions

qwen3-4b-structeval-v12

Aggressive tuning: α=2r, epochs=3, dropout=0.05, early stopping.

lr=2e-05, epochs=3, r=8, alpha=16, BS=32
dropout=0.05, seed=42
enable_thinking=False, strip_cot=True(strict)
load_best_model_at_end=True, early_stopping_patience=3

Downloads last month: 22

Safetensors

Model size

4B params

Tensor type

BF16

·

Model tree for miitarou/qwen3-4b-structeval-v12

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(894)

this model

Dataset used to train miitarou/qwen3-4b-structeval-v12