--- base_model: - QiWang98/VideoRFT-SFT-3B - Qwen/Qwen2.5-VL-3B-Instruct datasets: - QiWang98/VideoRFT-Data language: - en license: apache-2.0 metrics: - accuracy pipeline_tag: video-text-to-text library_name: transformers --- # 🎥 VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning This repository contains the **VideoRFT** model, presented in the paper [VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning](https://huggingface.co/papers/2505.12434).

  📖 Paper    │   💻 Code    │   📀 CoT Dataset    │   📀 RL Dataset    │   🤗 Models

## 📰 News - [2025/09/19] Our paper has been **accepted to NeurIPS 2025** 🎉! - [2025/06/01] We released our 3B Models ([🤗VideoRFT-SFT-3B](https://huggingface.co/QiWang98/VideoRFT-SFT-3B) and [🤗VideoRFT-3B](https://huggingface.co/QiWang98/VideoRFT-3B)) to huggingface. - [2025/05/25] We released our 7B Models ([🤗VideoRFT-SFT-7B](https://huggingface.co/QiWang98/VideoRFT-SFT) and [🤗VideoRFT-7B](https://huggingface.co/QiWang98/VideoRFT)) to huggingface. - [2025/05/20] We released our Datasets ([📀CoT Dataset](https://huggingface.co/datasets/QiWang98/VideoRFT-Data) and [📀RL Dataset](https://huggingface.co/datasets/QiWang98/VideoRFT-Data)) to huggingface. - [2025/05/18] Our paper is released on [ArXiv](https://arxiv.org/abs/2505.12434), and we have open-sourced our code on [GitHub](https://github.com/QiWang98/VideoRFT)! ## 🔎 Overview Reinforcement fine-tuning (RFT) has shown great promise in achieving humanlevel reasoning capabilities of Large Language Models (LLMs), and has recently been extended to MLLMs. Nevertheless, reasoning about videos, which is a fundamental aspect of human intelligence, remains a persistent challenge due to the complex logic, temporal and causal structures inherent in video data. To fill this gap, we propose VideoRFT, a novel approach that extends the RFT paradigm to cultivate human-like video reasoning capabilities in MLLMs. VideoRFT follows the standard two-stage scheme in RFT: supervised fine-tuning (SFT) with chain-of-thought (CoT) annotations, followed by reinforcement learning (RL) to improve generalization. A central challenge to achieve this in the video domain lies in the scarcity of large-scale, high-quality video CoT datasets. We address this by building a fully automatic CoT curation pipeline. First, we devise a cognitioninspired prompting strategy to elicit a reasoning LLM to generate preliminary CoTs based solely on rich, structured, and literal representations of video content. Subsequently, these CoTs are revised by a visual-language model conditioned on the actual video, ensuring visual consistency and reducing visual hallucinations. This pipeline results in two new datasets VideoRFT-CoT-102K for SFT and VideoRFT-RL-310K for RL. To further strength the RL phase, we introduce a novel semantic-consistency reward that explicitly promotes the alignment between textual reasoning with visual evidence. This reward encourages the model to produce coherent, context-aware reasoning outputs grounded in visual input. Extensive experiments show that VideoRFT achieves state-of-the-art performance on six video reasoning benchmarks.
## ✨ Methodology To overcome the scarcity of video CoTs, we develop a scalable, cognitively inspired pipeline for high-quality video CoT dataset construction.
To further strength the RL phase, we introduce a novel semantic-consistency reward that explicitly promotes the alignment between textual reasoning with visual evidence.
## 📀 Datasets Based on above pipeline, we construct two large-scale datasets, i.e., [📀VideoRFT-CoT-102K](https://huggingface.co/datasets/QiWang98/VideoRFT-Data) and [📀VideoRFT-RL-310K](https://huggingface.co/datasets/QiWang98/VideoRFT-Data).
## 🛠️ Set up ### Requirements * `Python >= 3.11` * `Pytorch >= 2.5.1` * `transformers == 4.51.3` * `vLLM == 0.7.3` * `trl == 0.16.0` ### Installation ```bash git clone https://github.com/QiWang98/VideoRFT cd VideoRFT # Create and activate environment conda create -n VideoRFT python=3.11 conda activate VideoRFT bash setup.sh # Install decord for improved video processing cd src/qwen-vl-utils pip install -e .[decord] ``` ## 🚀 Training ### Supervised Fine-Tuning (SFT) We begin with supervised fine-tuning on the VideoRFT-CoT dataset for one epoch: ```bash bash ./src/scripts/run_sft_video.sh ``` This step can be skipped by directly using our pretrained SFT models, available at [🤗VideoRFT-SFT-7B](https://huggingface.co/QiWang98/VideoRFT-SFT) or [🤗VideoRFT-SFT-3B](https://huggingface.co/QiWang98/VideoRFT-SFT-3B). ### Reinforcement Learning (RL) Next, perform reinforcement learning using the VideoRFT-RL dataset: ```bash bash ./src/scripts/run_grpo_video.sh ``` To enable faster training via vLLM acceleration: ```bash bash ./src/scripts/run_grpo_vllm_qwen25vl.sh ``` > **Note:** During training, we adopt the following settings for efficiency: * **VIDEO PIXELS**: 128 × 28 × 28 * **FPS FRAMES**: 16 All frame-related configurations can be adjusted in `src/qwen-vl-utils`. ## 📈 Inference & Evaluation > During inference, we increase the maximum frame resolution and length to boost performance: * **VIDEO PIXELS**: 256 × 28 × 28 * **FPS FRAMES**: 32 You can configure these parameters in `src/qwen-vl-utils`. > We evaluate all models under a unified decoding configuration following the official Qwen2.5-VL demo: * `top_p = 0.001` * `temperature = 0.01` ### Evaluation Procedure 1. Download preprocessed evaluation JSONs from: [[🤗 eval](https://huggingface.co/datasets/Video-R1/Video-R1-eval)] 2. Download the video data from the official sites of each benchmark and organize them as specified in the JSON files. 3. Run the evaluation across all benchmarks: ```bash bash ./src/eval_bench.sh ``` ## Quick Inference Code ```python import numpy as np import torch from longvu.builder import load_pretrained_model from longvu.constants import ( DEFAULT_IMAGE_TOKEN, IMAGE_TOKEN_INDEX, ) from longvu.conversation import conv_templates, SeparatorStyle from longvu.mm_datautils import ( KeywordsStoppingCriteria, process_images, tokenizer_image_token, ) from decord import cpu, VideoReader tokenizer, model, image_processor, context_len = load_pretrained_model( "./checkpoints/longvu_qwen", None, "cambrian_qwen", ) model.eval() video_path = "./examples/video1.mp4" qs = "Describe this video in detail" vr = VideoReader(video_path, ctx=cpu(0), num_threads=1) fps = float(vr.get_avg_fps()) frame_indices = np.array([i for i in range(0, len(vr), round(fps),)]) video = [] for frame_index in frame_indices: img = vr[frame_index].asnumpy() video.append(img) video = np.stack(video) image_sizes = [video[0].shape[:2]] video = process_images(video, image_processor, model.config) video = [item.unsqueeze(0) for item in video] qs = DEFAULT_IMAGE_TOKEN + " " + qs conv = conv_templates["qwen"].copy() conv.append_message(conv.roles[0], qs) conv.append_message(conv.roles[1], None) prompt = conv.get_prompt() input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0).to(model.device) stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 keywords = [stop_str] stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids) with torch.inference_mode(): output_ids = model.generate( input_ids, images=video, image_sizes=image_sizes, do_sample=False, temperature=0.2, max_new_tokens=128, use_cache=True, stopping_criteria=[stopping_criteria], ) pred = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ``` ## 🙏 Acknowledgements We gratefully acknowledge the contributions of the open-source community, particularly [DeepSeek-R1](https://github.com/deepseek-ai/DeepSeek-R1), [Open-R1](https://github.com/huggingface/open-r1), and [R1-V](https://github.com/Deep-Agent/R1-V). ## 📚 Citations If you find this work helpful, please consider citing: ``` @article{VideoRFT, title={VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning}, author={Wang, Qi and Yu, Yanrui and Yuan, Ye and Mao, Rui and Zhou, Tianfei}, journal={arXiv preprint arXiv:2505.12434}, year={2025} } ```