wchai/AuroraCap-trainset
AuroraCap Trainset Resources Website arXiv: Paper GitHub: Code Huggingface: AuroraCap Model Huggingface: VDC Benchmark Huggingface: Trainset Features We use over 20 million high-quality image/video-text pairs to train AuroraCap in three stages. Pretraining stage. We first align visual features with the word embedding space of LLMs. To achieve this, we freeze the pretrained ViT and LLM, training solely the vision-language connector. Vision… See the full description on the dataset page: https://huggingface.co/datasets/wchai/AuroraCap-trainset.
| Author / organization | wchai |
|---|---|
| License (as declared) | apache-2.0 |
| Downloads | 1.6K (as reported by Hugging Face) |
| Likes | 9 |
| Created | 3 October 2024 |
| Last updated at source | 13 October 2024 |
Popularity, last 3 days
1,581 now (+80 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
- mradermacher/AuroraCap-7B-IMG-xtuner-i1-GGUFModel · used to train
- mradermacher/AuroraCap-7B-VID-xtuner-GGUFModel · used to train
- mradermacher/AuroraCap-7B-IMG-xtuner-GGUFModel · used to train
- mradermacher/AuroraCap-7B-VID-xtuner-i1-GGUFModel · used to train
Recent changes
- 17 Oct 2026 Description: AuroraCap Trainset Resources Website arXiv: Paper GitHub: Code Huggingface: AuroraCap Model Huggingface: VDC Benchmark Huggingface: Trainset Features We use over 20 million high-quality image/video-text pairs to train AuroraCap in three sta → AuroraCap Trainset Resources Website arXiv: Paper GitHub: Code Huggingface: AuroraCap Model Huggingface: VDC Benchmark Huggingface: Trainset Features We use over 20 million high-quality image/video-text pairs to train AuroraCap in three sta
Changes detected by Silytics between refreshes of the source.
More from wchai
- wchai/Video-Detailed-CaptionDataset
- wchai/lmms_VDC_testDataset
- wchai/imagenet-wdsDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.