Shadowgraph
TrendingAbout
Sign in

WaltonFuture/geometry3k-direct-synthesizing

Dataset

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO to improve the reasoning abilities of multi-modal large language models (MLLMs). It contains image-text pairs, where each pair consists of an image, a problem described in text, and the corresponding answer. The dataset is designed for unsupervised post-training of MLLMs.

1

5 commits

1 linked in READMEs

updated Jun 2, 2025

See the code

README

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO to improve the reasoning abilities of multi-modal large language models (MLLMs). It contains image-text pairs, where each pair consists of an image, a problem described in text, and the corresponding answer. The dataset is designed for unsupervised post-training of MLLMs.

The dataset contains 2101 examples in the training split. Each example includes:

  • images: An image related to the problem.
  • problem: The problem statement (text).
  • answer: The correct answer (text).
multimodal
reasoning
unsupervised-learning

Contributors

WaltonFuture

4 commits

nielsr

1 commits

Similar repos

WaltonFuture/geometry3k-in-context-synthesizing

Dataset

This dataset is used for unsupervised post-training of multi-modal large language models (MLLMs).…

2

WaltonFuture/MMR1-direct-synthesizing

Dataset

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO…

1

WaltonFuture/MMR1-in-context-synthesizing

Dataset

This dataset is designed for unsupervised post-training of Multi-Modal Large Language Models…

0

WaltonFuture/GeoQA-8K-direct-synthesizing

Dataset

This dataset supports the unsupervised post-training of multi-modal large language models (MLLMs)…

1

WaltonFuture/GeoQA-8K-in-context-synthesizing

Dataset

* 🐙 GitHub Repo: waltonfuture/MM-UPT

0

WaltonFuture/Multimodal-RL-Data

Dataset

Multimodal-Cold-Start Dataset

8

WaltonFuture/Qwen2.5-VL-7B-MM-UPT-MMR1

Model

Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO

3

GML-FMGroup/VisNumBench

Dataset

This dataset is designed for research in **Deep Learning for Geometry Problem Solving (DL4GPS)**…

2

Linked in READMEs

waltonfuture/MM-UPT

[NeurIPS 2025] First SFT, Second RL, Third UPT: Continual Improving Multi-Modal LLM Reasoning via…

89

README badge

WaltonFuture/geometry3k-direct-synthesizing

Dataset

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO to improve the reasoning abilities of multi-modal large language models (MLLMs). It contains image-text pairs, where each pair consists of an image, a problem described in text, and the corresponding answer. The dataset is designed for unsupervised post-training of MLLMs.

1

5 commits

1 linked in READMEs

updated Jun 2, 2025

See the code

README

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO to improve the reasoning abilities of multi-modal large language models (MLLMs). It contains image-text pairs, where each pair consists of an image, a problem described in text, and the corresponding answer. The dataset is designed for unsupervised post-training of MLLMs.

The dataset contains 2101 examples in the training split. Each example includes:

  • images: An image related to the problem.
  • problem: The problem statement (text).
  • answer: The correct answer (text).
multimodal
reasoning
unsupervised-learning

Similar repos

WaltonFuture/geometry3k-in-context-synthesizing

Dataset

This dataset is used for unsupervised post-training of multi-modal large language models (MLLMs).…

2

WaltonFuture/MMR1-direct-synthesizing

Dataset

This dataset is used in the paper Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO…

1

WaltonFuture/MMR1-in-context-synthesizing

Dataset

This dataset is designed for unsupervised post-training of Multi-Modal Large Language Models…

0

WaltonFuture/GeoQA-8K-direct-synthesizing

Dataset

This dataset supports the unsupervised post-training of multi-modal large language models (MLLMs)…

1

WaltonFuture/GeoQA-8K-in-context-synthesizing

Dataset

* 🐙 GitHub Repo: waltonfuture/MM-UPT

0

WaltonFuture/Multimodal-RL-Data

Dataset

Multimodal-Cold-Start Dataset

8

WaltonFuture/Qwen2.5-VL-7B-MM-UPT-MMR1

Model

Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO

3

GML-FMGroup/VisNumBench

Dataset

This dataset is designed for research in **Deep Learning for Geometry Problem Solving (DL4GPS)**…

2

Linked in READMEs

waltonfuture/MM-UPT

[NeurIPS 2025] First SFT, Second RL, Third UPT: Continual Improving Multi-Modal LLM Reasoning via…

89

README badge

Contributors

WaltonFuture

4 commits

nielsr

1 commits