zghhui
/

Janus-Pro-7B-GCPO-HPS

Model card Files Files and versions

This repository hosts a [Janus-Pro 7B] trained by GCPO. The reward model is HPS.

The training code is available at GCPO

Downloads last month: 9

Safetensors

Model size

7B params

Tensor type

BF16

·

Model tree for zghhui/Janus-Pro-7B-GCPO-HPS

Base model

deepseek-ai/Janus-Pro-7B

Finetuned

(34)

this model

Collection including zghhui/Janus-Pro-7B-GCPO-HPS

GCPO

6 items • Updated Sep 26 • 1