Yep! It’s SoTA/ near-SoTA atm! This serves as the key piece in a TTS/ TTA pipeline. As it allows us to encode an audio into discrete representations and then perform language modelling on it!
MACHINE LEARNING
-
GeneCIS: Benchmark for Model Adaptation to Similarity Conditions
By
–
GeneCIS: we propose a benchmark, which measures models’ ability to adapt to a range of similarity conditions. Paper https://
bit.ly/3qQCiwq 6/7 -
HierVL: Hierarchical Video-Language Embedding for Temporal Associations
By
–
HierVL is a novel hierarchical video-language embedding that simultaneously accounts for both long-term and short-term associations. Paper https://
bit.ly/3qSmEk6 7/7 -
Galactic: Fast RL Framework for Robotic Mobile Manipulation
By
–
🚀 Galactic is a large-scale simulation and RL framework for robotic mobile manipulation in indoor environments that achieves an incredible 54x speed increase over Habitat 2.0 on an 8-GPU node.
— AI at Meta (@AIatMeta) 20 juin 2023
Paper ➡️ https://t.co/2z7JrmdyHK
5/7 pic.twitter.com/hIXjWa9l3kGalactic is a large-scale simulation and RL framework for robotic mobile manipulation in indoor environments that achieves an incredible 54x speed increase over Habitat 2.0 on an 8-GPU node. Paper https://
bit.ly/42UkNbI 5/7 -
Learning Video Representations from Large Language Models
By
–
📺 Learning Video Representations from Large Language Models
— AI at Meta (@AIatMeta) 20 juin 2023
This work repurposed pre-trained LLMs to be conditioned on visual input, and finetune them to create automatic video narrators.
Paper ➡️ https://t.co/OsZt3AnAEM
3/7 pic.twitter.com/DLunz5bcgPLearning Representations from Large Language Models This work repurposed pre-trained LLMs to be conditioned on visual input, and finetune them to create automatic video narrators. Paper https://
bit.ly/3NDCWpR 3/7 -
EgoTask Translation: Multi-Task Model Output Integration
By
–
EgoTask Translation (EgoT2) takes a collection of models optimized on separate tasks and learns to translate their outputs for improved performance on any or all of them at once. Paper https://
bit.ly/3PmIJBr 2/7 -
Why AI Must Thrive: Our World Needs It
By
–
Why AI Must Thrive: Our World Needs It
#AI #AIio #BigData #ML #NLU #Futureofwork @oluskayacan @richardkimphd @MariaFariello @johnpearcenews5 @ResurgentAV @rohanmuralee @globaliqx http://
ow.ly/hZaC30svHTC -
EnCodec Model Now Available in Transformers Library
By
–
Want to train your own Bark/MusicGen-like TTS/TTA models? 👀
— Vaibhav (VB) Srivastav (@reach_vb) 20 juin 2023
The SoTA Encodec model by @MetaAI has now landed in 🤗Transformers!
It supports compression up to 1.5KHz and produces discrete audio representations. ⚡️
Model: https://t.co/Hq8rDHBfjw
Colab: https://t.co/MaWVEAMCXs pic.twitter.com/tDpPAdlYHUWant to train your own Bark/MusicGen-like TTS/TTA models? The SoTA Encodec model by @MetaAI has now landed in Transformers! It supports compression up to 1.5KHz and produces discrete audio representations. Model: https://
huggingface.co/docs/transform
ers/main/en/model_doc/encodec#overview
…
Colab: https://
github.com/Vaibhavs10/not
ebooks/blob/main/use_encodec_w_transformers.ipynb
… -
Transformers Integration Enables Large-Scale Text-to-Speech Music Models
By
–
Okay, but why is it a big deal!? Transformers integration allows us to use any LM and dataset from the ecosystem seamlessly to train Text-to-Speech and Text-to-Music models at scale! More exciting announcements on this front soon! https://
github.com/Vaibhavs10/not
ebooks/blob/main/use_encodec_w_transformers.ipynb
… -
Infinigen: Procedural Generation for Photorealistic 3D Worlds
By
–
Infinite Photorealistic Worlds using Procedural Generation
— AK (@_akhaliq) 20 juin 2023
paper page: https://t.co/HiSl313uEN
introduce Infinigen, a procedural generator of photorealistic 3D scenes of the natural world. Infinigen is entirely procedural: every asset, from shape to texture, is generated from… pic.twitter.com/K1iGHwRahtInfinite Photorealistic Worlds using Procedural Generation paper page: https://
huggingface.co/papers/2306.09
310
… introduce Infinigen, a procedural generator of photorealistic 3D scenes of the natural world. Infinigen is entirely procedural: every asset, from shape to texture, is generated from