About
static quants of https://huggingface.co/internlm/CapRL-Qwen3VL-4B
weighted/imatrix quants are available at https://huggingface.co/internlm/CapRL-Qwen3VL-4B-GGUF
Usage
If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files.
Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|---|---|---|---|
| GGUF | mmproj-Q8_0 | 0.6 | multi-modal supplement |
| GGUF | mmproj-f16 | 0.9 | multi-modal supplement |
| GGUF | Q2_K | 1.8 | |
| GGUF | Q4_K_S | 2.6 | fast, recommended |
| GGUF | Q4_K_M | 2.7 | fast, recommended |
| GGUF | Q6_K | 3.6 | very good quality |
| GGUF | Q8_0 | 4.7 | fast, best quality |
| GGUF | f16 | 8.8 | 16 bpw, overkill |
Citation
If you find this project useful, please cite:
@article{xing2025caprl,
title={{CapRL}: Stimulating Dense Image Caption Capabilities via Reinforcement Learning},
author={Xing, Long and Dong, Xiaoyi and Zang, Yuhang and Cao, Yuhang and Liang, Jianze and Huang, Qidong and Wang, Jiaqi and Wu, Feng and Lin, Dahua},
journal={arXiv preprint arXiv:2509.22647},
year={2025}
}