Search Results for "hours"

...The project includes dense and MoE model variants for text-to-image, text-to-video, and text-image-to-video workflows. Its training combines large-scale web video data with more than 70,000 hours of embodied data. A multi-reward system emphasizes aesthetics, physical rationality, and task completion. The repository includes models, inference code, prompt rewriting tools, refiner workflows, and single-GPU or multi-GPU scripts.

Qwen2.5 is a series of large language models developed by the Qwen team at Alibaba Cloud, designed to enhance natural language understanding and generation across multiple languages. The models are available in various sizes, including 0.5B, 1.5B, 3B, 7B, 14B, 32B, and 72B parameters, catering to diverse computational requirements. Trained on a comprehensive dataset of up to 18 trillion tokens, Qwen2.5 models exhibit significant improvements in instruction following, long-text generation...

...Built on advanced deep learning architectures that combine Diffusion Transformer (DiT) and flow matching techniques, HY-Motion scales these approaches to the billion-parameter level, resulting in strong instruction-following capabilities and richer motion outputs compared to existing open-source models. The training strategy for the HY-Motion series includes extensive pre-training on thousands of hours of varied motion data, fine-tuning on curated high-quality datasets, and reinforcement learning with human feedback, which improves both the plausibility and adaptability of generated motion sequences.

Surya is an open‑source, AI‑based foundation model for heliophysics developed collaboratively by NASA (via the IMPACT AI team) and IBM. Named after the Sanskrit word for “sun,” Surya is trained on nine years of high‑resolution solar imagery from NASA’s Solar Dynamics Observatory (SDO). It is designed to forecast solar phenomena—such as flares, solar wind, irradiance, and active region behavior—by predicting future solar images with a sophisticated long–short vision transformer architecture,...

LingBot-VLA is an open-source Vision-Language-Action (VLA) foundational AI model designed to serve as a general “brain” for real-world robotic manipulation by grounding multimodal perception and language into actionable motions. It has been pretrained on tens of thousands of hours of real robotic interaction data across multiple robot platforms, which enables it to generalize well to diverse morphologies and tasks without needing extensive retraining on each new bot. The model aims to bridge vision, language understanding, and motor control within one unified architecture, making it capable of understanding high-level instructions and generating coherent low-level actions in physical environments. ...

...Vidi targets applications like intelligent video editing, automated video search, content analysis, and editing assistance, enabling users to efficiently locate relevant segments and objects in hours-long footage. The system is built with open-source release in mind, giving developers access to model code, inference scripts, and evaluation pipelines so they can reproduce research results or integrate Vidi into their own video-processing workflows.

...Specifically, we train a neural codec language model (called VALL-E) using discrete codes derived from an off-the-shelf neural audio codec model, and regard TTS as a conditional language modeling task rather than continuous signal regression as in previous work. During the pre-training stage, we scale up the TTS training data to 60K hours of English speech which is hundreds of times larger than existing systems. VALL-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt. Experiment results show that VALL-E significantly outperforms the state-of-the-art zero-shot TTS system in terms of speech naturalness and speaker similarity. ...