TimeLens2: Generalist Video Temporal Grounding with Multimodal LLMs
Paper โข 2607.17423 โข Published โข 76
How to use ZhuYuhan/TimeLens2-35B-A3B with Transformers:
# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM
processor = AutoProcessor.from_pretrained("ZhuYuhan/TimeLens2-35B-A3B")
model = AutoModelForMultimodalLM.from_pretrained("ZhuYuhan/TimeLens2-35B-A3B", device_map="auto")TimeLens2-35B-A3B is a video temporal grounding model based on Qwen/Qwen3.6-35B-A3B. Given a video and a text query, it predicts the relevant time intervals in seconds.
Paper ยท Code & Usage ยท Collection
| Benchmark | mIoU (%) |
|---|---|
| Charades-STA | 60.73 |
| ActivityNet Captions | 63.34 |
| QVHighlights | 70.53 |
| VUE-TR | 51.85 |
| VUE-TR-v2 | 51.90 |
| MomentSeeker | 30.79 |
| Ego4D-NLQ-v2 | 23.61 |
| Average | 50.39 |
The two VUE benchmarks use the available-video subsets. The average is the unweighted mean over the seven benchmarks.
For inference and evaluation, follow the official TimeLens2 repository and use ZhuYuhan/TimeLens2-35B-A3B as the model path.