Jevol-2B Drone Policy
VLA drone policy: image + mission in, flight command out
None defined yet.
VLA drone policy: image + mission in, flight command out
Camera image + mission → drone flight command
Chat model that folds proteins into 3D structures
Full-duplex audiovisual world model (Duo-1.0-distill)
Flexible-context probabilistic time series forecasting
Sanskrit TTS: type Devanagari, hear it spoken
Screenshot web agent — predicts the next browser action
Screenshot + request → one browser action, with reasoning
Screenshot to browser action with a Korean GUI web agent
Who said what — multi-talker transcription with timestamps
Speech-to-speech voice chat with sound tags (4B)
3.58B TTS: voice description + voice changer
Depth map + prompt to image with MOSS-Image LoRA
Streaming Turkish ASR with Damga ITN, int8 CPU
Training-free streaming diarization for meetings via e-means
Fine-grained visual perception VLM (ZoomGain-4B)
1.75M-param TTS decoder, 30x smaller than Kokoro's
Hindi text-to-speech, 62M params, 6 voices
One image to 3D Gaussian splats at any token budget
Radiology reports with outlined findings (X-ray, CT, MRI)
One-step two-speaker dialogue separation
4.5M-param English TTS, two voices, 24 kHz, runs on CPU
Vietnamese ASR with speaker diarization and timestamps
Streaming Chinese⇄English speech-to-speech translation