qwen3-4b-sft-lora-v2-20260207-1147
LoRA adapter fine-tuned from unsloth/Qwen3-4B-Instruct-2507 using QLoRA (4-bit).
Training Configuration
- LR: 5e-05, Epochs: 2, Effective BS: 32
- LoRA: r=64, alpha=128
- Max seq len: 1024
- CoT mask: enabled (after_marker)
Datasets
- u-10bei/structured_data_with_cot_dataset_512_v2
- daichira/structured-5k-mix-sft
- daichira/structured-hard-sft-4k
- Downloads last month
- 2
Model tree for yusei926/qwen3-4b-sft-lora-v2-20260207-1147
Base model
Qwen/Qwen3-4B-Instruct-2507 Finetuned
unsloth/Qwen3-4B-Instruct-2507