Views
No views yet
Non-commercial weights. These model weights are released by SenseTime under CC BY-NC 4.0 and may be used for non-commercial purposes only. The LibreYOLO integration code is MIT; the vendored architecture code is Apache-2.0. The license of the weights does not change by being mirrored here.
LibreVLM tier. A unified
multimodal model (Bagel-MoT architecture: Qwen2.5-7B MoT decoder, SigLIP
vision tower, FLUX autoencoder) that serves many vision tasks from one
checkpoint: symbolic outputs (boxes, points, keypoints, OCR words) are
generated as tagged text, dense outputs (depth maps, segmentation masks,
panoptic maps) are generated as images decoded by the VAE.79548fcc5b954598799b9317f8d3ec5e347d5c0e.
Copyright (c) 2026 SenseTime Group Inc. and/or its affiliates.12ccd96e32b32967a11cacb6c5bd5fe3a555fc0c, Apache-2.0).
Paper: Vision as Unified Multimodal Generation.ema.safetensors 96f29abd98791288c5a24087322e964bb9bcabfc2f185ece71543f827bc2b11e
ae.safetensors afc8e28272cd15db3919bacdb6918ce9c1ed22e96cb12c4d5ed0fba823529e38tokenizer.json assigns the chat/vision
special tokens ids beyond the checkpoint's embedding table, while
tokenizer_config.json's added_tokens_decoder records the layout the model
was trained with (structured tokens overriding ids 149632-151664). LibreYOLO
reconstructs the trained layout at load time; other consumers should do the
same or use a legacy slow tokenizer built from vocab.json/merges.txt.1from libreyolo import LibreVLM
2
3model = LibreVLM("sensenova-vision", task="detect")
4model.set_classes(["person", "bicycle"])
5results = model.predict("image.jpg") # Results.boxes
6
7model.set_task("depth")
8results = model.predict("image.jpg") # Results.depth_map
9
10model.set_task("segment").set_classes(["the person on the left"])
11results = model.predict("image.jpg") # Results.masksdetect, point, pose, ocr, depth,
segment (referring), panoptic. Free-form access via model.chat(...)
and model.generate(...).