| # Attribution and license scope | |
| Phoenix-LIBERO is a modified policy derived from Wan2.2-TI2V-5B through Phoenix Robot37 pretraining (step 125000) and LIBERO post-training (step 25000). Its release checkpoint also changes the Joint96 metadata schema to the Phoenix name; the schema migration does not change inference tensors. | |
| The Wan Team's upstream model is distributed under Apache License 2.0. A copy is included in [LICENSES/Apache-2.0.txt](LICENSES/Apache-2.0.txt). Upstream source and model terms: [Wan-AI/Wan2.2-TI2V-5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B). This notice preserves the upstream license and records the modifications; it does not assign an additional license to Phoenix-owned weight contributions. | |
| LIBERO is acknowledged as the post-training and evaluation benchmark: [Lifelong Robot Learning / LIBERO](https://github.com/Lifelong-Robot-Learning/LIBERO). The accompanying configuration, normalization statistics and initialization records document this policy. Historical absolute paths identify source artifacts; they do not grant access to, or redistribute, those datasets or private pretraining checkpoints. | |
| The separately distributed [Phoenix-Encoders](https://huggingface.co/OpenPhoenix/Phoenix-Encoders) repository contains the required DINO/CSVAE encoders and their notices. Public Wan VAE, T5 and tokenizer dependencies retain their upstream terms and are obtained separately. | |
| Phoenix-MWA's code license is not a blanket license for model weights, datasets or third-party dependencies. No additional Phoenix weight license is inferred from that code license. Retained third-party notices do not imply endorsement by the upstream authors. | |