Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe) header
Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe) profile picture

Creator of ViTSTR and EfficientSpeech (ICASSP2023) and co-creator of PARSeq. Professor & Scientist at the University of the Philippines.

University of the Philippines
Joined February 2007
Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)Photo by Rowel Atienza πŸ‡΅πŸ‡­ (@jacobe)
AI agents shouldn't be complex, heavy, or costly. baby-onit: one file, ~1,000 lines. Everything is traceable. Use @tavilyai and @OpenRouter or @vercel to start using agents without incurring initial costs. github.com/sibyl-oracles/bab…
2
130
Most ARM chips can't run decent AI models. Introducing EfficientSpeech, a 266k-param TTS model. Low cost ARM chips like in RPi4 can generate 104sec of speech mel spec in 1sec. Here's an AI-generated video w/ voice from EfficientSpeech. Info: github.com/roatienza/efficie… #ICASSP2023
2
2
1
659
We are lucky to have this paper accepted at ECCV 2022! More details later ... #ECCV2022
1
1
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
.@Gradio & @huggingface for Rapid Deep Learning App Development by @jacobe link: medium.com/@rowel/gradio-hug…
2
11
1
52
Idea: If data augmentation improves model generalization, why not use it to generate 2 new inputs and force the representations to agree. Result: Additional model performance improvement. Comparison: Unlike Label Smoothing, the performance of our method, AgMax, is consistent.
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
I am excited to share my latest work: 8-bit optimizers – a replacement for regular optimizers. Faster πŸš€, 75% less memory πŸͺΆ, same performanceπŸ“ˆ, no hyperparam tuning needed πŸ”’. 🧡/n Paper: arxiv.org/abs/2110.02861 Library: github.com/facebookresearch/… Video: youtube.com/watch?v=IxrlHAJt…
18
269
29
1,381
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
We’re introducing GSLM, the first language model that breaks free completely of the dependence on text for training. This β€œtextless NLP” approach learns to generate expressive speech using only raw audio recordings as input. Learn more and get the code: ai.facebook.com/blog/textles…
14
332
57
1,210
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
4
5
My 1st github project that reached the 1k milestone. @PacktPub github.com/roatienza
4
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
GOO: A Dataset for Gaze Object Prediction in Retail Environments deepai.org/publication/goo-a… by Henri Tomas et al. including @jacobe #Estimator #Statistics
1
4
1
14
Yesterday, my former grad student Daryl gave a talk at Sony CSL Paris about his thesis on Next View Policy for 3D Reconstruction. Youtube: youtu.be/KdyDj3bjU0I
2
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
After the Christmas break, we come back with our #enticelle seminar series! We welcome Daryl Peralta from @upsystem with his talk about β€˜Next-Best View Policy for 3D Reconstruction’. Register here to attend on Wednesday 13 at 11am (CET) csl.sony.fr/seminars/
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
Next-Best View Policy for 3D Reconstruction pdf: arxiv.org/pdf/2008.12664.pdf abs: arxiv.org/abs/2008.12664
11
36
Our paper Next-Best View Policy for 3D Reconstruction was presented (oral) Aug 28 at ECCV 2020 Workshops. Key contribs: 1) 3K 3D House Models dataset 2)Learning algo to efficiently scan 3D houses. Paper: arxiv.org/abs/2008.12664 Youtube: youtu.be/OXynAHTDTTA
2
Rowel Atienza πŸ‡΅πŸ‡­ retweeted
Advanced Deep Learning with TensorFlow 2 and Keras 2nd Ed (@PacktPub) has been added to bookwatch. Author Rowel Atienza (@jacobe) introduces the practical side of deep learning i-programmer.info/book-watch…
2
6