datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PerceptionTest_ValPerceptionTestCodalm_test_region_perceptionCodalm_test_general_perceptionPerceptionTest_valPerceptionTest_Valperception_test_val-event_recallfolding_nets_3d_perception_testfolding_nets_2d_perception_test3d_transformation_perception_test2d_transformation_perception_testperception_test_sfttangram_perception_testtangram_overlap_perception_testvideo-r1-perceptiontest-mc-v1 ---
language:
- en
license: apache-2.0
size_categories:
- 1K<n<10K
task_categories:
- video-text-to-text
tags:
- multimodal-rl
- qwen3-vl
- gspo
- grpo
---
# ngqtrung/video-r1-perceptiontest-mc-v1
Curated v1 dataset for multimodal RL fine-tuning of Qwen3-VL-4B-Instruct.
| Property | Value |
|---|---|
| Rows | 6348 |
| Modality | video |
| Split | train |
| Schema | verl-ready (prompt + images + videos +… See the full description on the dataset page: https://huggingface.co/datasets/ngqtrung/video-r1-perceptiontest-mc-v1.tangram_intermediate_perception_test