SOURCE-REVIEWED MODEL PROFILE
EgoCaptioner
Qwen3.5-9B fine-tune for video overviews and timed first-person activity steps.

Metadata reviewed 2026-11-21 · Original creator: Vickyinmyheart824
PARAMETERSApproximately 9.41B BF16 parameters in Hub metadata
CONTEXTSee task-specific source settings; no independent context-limit validation
LICENSEApache-2.0
MODALITYVideo and text to captions
What it does
- Video overview
- Timestamped steps
- Event-conditioned captions
Local fit
Linux and suitable NVIDIA BF16 GPU; context and video processing add memory overhead.. Source reviewed only; no weights downloaded or inference run.
Publisher documentation, not a measured local limit.
Evidence and limits
No independent local inference, benchmark or memory-fit test.
Source reviewed
License and source
Card, weight-repository license and code license declare Apache-2.0. This is a derivative Qwen fine-tune.
Read the original model card ↗
Reviewed source revision: cc0a58c49a32ac8fc7a5838428579ffc84306618