Skip to content
ModelListedExternal
Browse files

INFRL Qwen2.5 VL 72B Preview Ggufs Fully Quantized

INFRL-Qwen2.5-VL-72B-Preview improves visual reasoning upon Qwen2.5-VL-72B-Instruct model.

Sourcehuggingface.co/GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized/tree/e424ecbe1d959280e6a05b066898612e9577bb4a
KindModelBase modelQwen/Qwen2.5-VL-72B-InstructVersionRevision e424ecbBgrade

Model source

Kind
Model
File size
1.31–46.5 GB
Base model
Qwen/Qwen2.5-VL-72B-Instruct
Version
vRevision e424ecb
Updated
Oct 30, 2026
Downloads
980
Source
Hugging Face

Source description

INFRL-Qwen2.5-VL-72B-Preview

Model Overview

  • INFRL-Qwen2.5-VL-72B-Preview improves visual reasoning upon Qwen2.5-VL-72B-Instruct model.

  • As of March 25th, 2025, INFRL-Qwen2.5-VL-72B-Preview is the best-performing open-sourced VL model on various visual reasoning benchmarks (MathVision,MathVista, MathVerse).

Evaluation

ModelsMathVision (test)MathVista (testmini)MathVerse (testmini)
GPT4o30.66041.2
Gemini-2.0-Flash41.370.150.6
Claude 3.5 Sonnet33.567.747.8
QvQ-72B35.971.448.6
InternVL2.5-78B34.972.351.7
Qwen-VL-2.5-72B38.174.857.18
INFRL-VL-Preview41.977.858.84

We will release a code repository for VLM evaluation. It supports RL training with simple rule-based rewards, meanwhile aligning with LLM-Judge results.

Stay tuned!

Contributors

Supervisors

Wei Chu • Yuan Qi

VL Team

Haozhe Wang • Zuming Huang

RL Team

Haozhe Wang • Chao Qu • Long Li

Thanks

Thanks to Jiaran Hao, Liuyihan Song for supports in the RL infrastructure.

Citation

If you find our model useful, please consider citing:

@misc {INFRL_VL_Preview,
	author       = { {Wang, Haozhe and Huang, Zuming and Qu, Chao and Chu, Wei and Qi, Yuan} },
	title        = { INFRL-Qwen2.5-VL-72B-Preview },
	year         = 2025,
	url          = { https://huggingface.co/infly/INFRL-Qwen2.5-VL-72B-Preview},
	publisher    = { Hugging Face }
}

Sources

1 source
Hugging FacePrimaryVersion e424ecbe1d959280e6a05b066898612e9577bb4a
huggingface.co/GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized/tree/e424ecbe1d959280e6a05b066898612e9577bb4a

Verified Oct 30

Download files

Repository files

File inventory only; completeness is unverified.

Model artifacts

23 artifacts

F16/INFRL-Qwen2.5-VL-72B-Preview-f16-00001-of-00003.gguf

gguf · 46.3 GB · SHA-256 6b42a9a5b1af…1345 · Hugging Face

Download shard

F16/INFRL-Qwen2.5-VL-72B-Preview-f16-00002-of-00003.gguf

gguf · 46.2 GB · SHA-256 e40289d241bb…b8f0 · Hugging Face

Download shard

F16/INFRL-Qwen2.5-VL-72B-Preview-f16-00003-of-00003.gguf

gguf · 43.0 GB · SHA-256 5e9deef7fee9…ae9d · Hugging Face

Download shard

INFRL-Qwen2.5-VL-72B-Preview-IQ3_XS.gguf

gguf · 30.6 GB · SHA-256 948d33bace47…1f08 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-IQ4_NL.gguf

gguf · 38.5 GB · SHA-256 d4d97f2a7598…6315 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-IQ4_XS.gguf

gguf · 37.0 GB · SHA-256 cf2a6e63969f…9b17 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q2_K_S.gguf

gguf · Q2_K_S (unverified) · 27.5 GB · SHA-256 66c263e7b858…fd82 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q2_K.gguf

gguf · Q2_K (unverified) · 27.8 GB · SHA-256 9bca1d50f1fe…cce2 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q3_K_L.gguf

gguf · Q3_K_L (unverified) · 36.8 GB · SHA-256 96f9c5538e19…e157 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q3_K_M.gguf

gguf · Q3_K_M (unverified) · 35.1 GB · SHA-256 55c476163d67…b132 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q3_K_S.gguf

gguf · Q3_K_S (unverified) · 32.1 GB · SHA-256 87884f216a12…fdb1 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q3_K.gguf

gguf · Q3_K (unverified) · 35.1 GB · SHA-256 55c476163d67…b132 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q4_K_M.gguf

gguf · Q4_K_M (unverified) · 44.2 GB · SHA-256 d6f8b66bcb6e…b479 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q4_K_S.gguf

gguf · Q4_K_S (unverified) · 40.9 GB · SHA-256 535b85b7d4dc…6282 · Hugging Face

Download file

INFRL-Qwen2.5-VL-72B-Preview-Q4_K.gguf

gguf · Q4_K (unverified) · 44.2 GB · SHA-256 d6f8b66bcb6e…b479 · Hugging Face

Download file

Q5_K_M/INFRL-Qwen2.5-VL-72B-Preview-Q5_K_M-00001-of-00002.gguf

gguf · Q5_K_M (unverified) · 46.5 GB · SHA-256 1ed8ac6b957f…df80 · Hugging Face

Download shard

Q5_K_M/INFRL-Qwen2.5-VL-72B-Preview-Q5_K_M-00002-of-00002.gguf

gguf · Q5_K_M (unverified) · 4.20 GB · SHA-256 7648da3aa6fa…b554 · Hugging Face

Download shard

Q5_K_S/INFRL-Qwen2.5-VL-72B-Preview-Q5_K_S-00001-of-00002.gguf

gguf · Q5_K_S (unverified) · 46.5 GB · SHA-256 a207a72d3f92…92fd · Hugging Face

Download shard

Q5_K_S/INFRL-Qwen2.5-VL-72B-Preview-Q5_K_S-00002-of-00002.gguf

gguf · Q5_K_S (unverified) · 1.31 GB · SHA-256 aa5bcc417bb0…1c44 · Hugging Face

Download shard

Q6_K/INFRL-Qwen2.5-VL-72B-Preview-Q6_K-00001-of-00002.gguf

gguf · Q6_K (unverified) · 46.5 GB · SHA-256 b1dfdb1c06e9…0e6d · Hugging Face

Download shard

Q6_K/INFRL-Qwen2.5-VL-72B-Preview-Q6_K-00002-of-00002.gguf

gguf · Q6_K (unverified) · 13.4 GB · SHA-256 c6aafdaecc81…acdf · Hugging Face

Download shard

Q8_0/INFRL-Qwen2.5-VL-72B-Preview-Q8_0-00001-of-00002.gguf

gguf · Q8_0 (unverified) · 46.5 GB · SHA-256 9119562ce0bb…d7af · Hugging Face

Download shard

Q8_0/INFRL-Qwen2.5-VL-72B-Preview-Q8_0-00002-of-00002.gguf

gguf · Q8_0 (unverified) · 25.4 GB · SHA-256 d266b23d8428…89ba · Hugging Face

Download shard

Source excerpts

3 excerpts
ReadmeAug 14huggingface.co/GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized

--- base_model: - Qwen/Qwen2.5-VL-72B-Instruct language: - en license: apache-2.0 tags: - transformers - multimodal pipeline_tag: visual-question-answering --- # INFRL-Qwen2.5-VL-72B-Preview ## Model Overview - **INFRL-Qwen2.5-VL-72B-Preview** improves visual reasoning upon [Qwen2.5-VL-72B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-72B-Instruct) model. - As of March 25th, 2025, **INFRL-Qwen2.5-VL-72B-Preview** is the best-performing open-sourced VL model on various visual reasoning benchmarks ([MathVision](https://mathllm.github.io/mathvision/),[MathVista](https://mathvista.github.io/), [MathVerse](https://mathverse-cuhk.github.io/)). ## Evaluation | Models | MathVision (test) | MathVista (testmini) | MathVerse (testmini) | |-------------------|-------------------|----------------------|----------------------| | GPT4o | 30.6 | 60 | 41.2 | | Gemini-2.0-Flash | 41.3 | 70.1 | 50.6 | | Claude 3.5 Sonnet | 33.5 | 67.7 | 47.8 | | QvQ-72B | 35.9 | 71.4 | 48.6 | | InternVL2.5-78B | 34.9 | 72.3 | 51.7 | | Qwen-VL-2.5-72B | 38.1 | 74.8 | 57.18 | | INFRL-VL-Preview | 41.9 | 77.8 | 58.84 | We will release a code repository for VLM evaluation. It supports RL training with simple rule-based rewards, meanwhile aligning with LLM-Judge results. Stay tuned! ## Contributors ### Supervisors Wei Chu • Yuan Qi ### VL Team Haozhe Wang • Zuming Huang ### RL Team Haozhe Wang • Chao Qu • Long Li ## Thanks Thanks to Jiaran Hao, Liuyihan Song for supports in the RL infrastructure. ## Citation If you find our model useful, please consider citing: ``` @misc {INFRL_VL_Preview, author = { {Wang, Haozhe and Huang, Zuming and Qu, Chao and Chu, Wei and Qi, Yuan} }, title = { INFRL-Qwen2.5-VL-72B-Preview }, year = 2025, url = { https://huggingface.co/infly/INFRL-Qwen2.5-VL-72B-Preview}, publisher = { Hugging Face } } ```

NoteOct 30huggingface.co/GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized

Source context: 980 downloads · 0 likes · Pipeline visual-question-answering · Library transformers · Repo GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized

NoteAug 14huggingface.co/GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized

Source context: 124 downloads · 0 likes · Pipeline visual-question-answering · Library transformers · Repo GeorgyGUF/INFRL-Qwen2.5-VL-72B-Preview-ggufs-fully-quantized