yamaTK
/

dpo-qwen-y-v35-v3

Text Generation

text-generation-inference

Model card Files Files and versions

dpo-qwen-y-v35

DPO fine-tuned version of Qwen/Qwen3-4B-Instruct-2507. Full-merged 16-bit weights. No adapter loading required.

Training Configuration

Method: DPO
Epochs: 1
Learning rate: 1e-07
Beta: 0.1
Max sequence length: 1024

Downloads last month: 2

Safetensors

Model size

4B params

Tensor type

BF16

·

Model tree for yamaTK/dpo-qwen-y-v35-v3

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(1541)

this model

Dataset used to train yamaTK/dpo-qwen-y-v35-v3