Tensorops

TensorOps organization, TransformerX Python library.

Transformer-based models' building blocks for Tensorflow, Pytorch, and Jax.
Discord:

[Tensorops @tensorops]

TransformerX is now available on PyPi 🎉

pip install transformerx

GitHub: https://t.co/8XuZIIsOhm

[Towards Data Science @TDataScience]

How can Hamiltonian mechanics apply to generative modeling and Markov chain Monte Carlo? @soranghadri presents a thoughtful deep dive on a fascinating topic. https://t.co/Adpc7oYXn4

[Soran Ghaderi @soranghadri]

📌 Hi all! Here's my brief intro to Hamiltonian Monte Carlo and H-mechanics. It breaks down the math (fully explained) and includes bits of visualizations. #mcmc #hamiltonianMonteCarlo #generativeAI

[Soran Ghaderi @soranghadri]

Finally got around to deploying the first version of the TorchEBM docs! It’s a little library I’ve been tinkering with for energy-based models in PyTorch. Still very early stages—think rough draft vibes—but it’s open-source and meant for research tinkering. [1/n]

[Soran Ghaderi @soranghadri]

For the past month, I have been developing TorchEBM, a library for modeling and training energy-based models. I'm actively developing it. Any suggestions, insights, or contributions are welcome.
https://t.co/fTX1WEQ5zr

[Tensorops @tensorops]

📢 New update on #TransformerX! You can now explore the improved flexibility while implementing sparse attention mechanisms.

link: https://t.co/J2Doldo8KK

[Soran Ghaderi @soranghadri]

The new masking API is now merged into the master branch. Now, effortlessly integrate new attention masks with a simple implementation of the mask logic in the build_mask() method.
#transformerX
https://t.co/6dl17YBnEk

[Tensorops @tensorops]

🎉 Exciting news! The TransformerX library has just received a major update, bringing enhanced performance and new features to the table. 🚀 Stay tuned for the stable release in the coming days.💬 #TransformerX #NLP #research #Transformers #DeepLearning

[Towards Data Science @TDataScience]

Catch up with the recent history of Transformers research — @soranghadri's overview is comprehensive and accessible; it's a powerful resource to bookmark for future reference. https://t.co/uxiIY3rQEC

[Towards Data Science @TDataScience]

. @soranghadri explains transformer architecture through underlying math, Python code implementation, and visualization of different layers. https://t.co/sD97C5PljR

[Soran Ghaderi @soranghadri]

Scaled dot-product attention is a type of #attention_mechanism where the dot-products of tensors are scaled down by √dk which normalizes the q.k variance to 1.

Check out https://t.co/HjW5GfEi0r for more. #Transformers

Also in the #TransformerX library: https://t.co/Pm1FncDwNh

[Towards Data Science @TDataScience]

Transformers in Action: Attention Is All You Need by @soranghadri https://t.co/sD97C5PljR

[Tensorops @tensorops]

Join the TensorOps community on Discord!

https://t.co/wp7viEOkV4

[Tensorops @tensorops]

TransformerX is a Python library that comes with a range of building blocks you need to implement transformer-based models. It helps researchers implement their models with ease.

[Tensorops @tensorops]

We have made the TransformerX Python library public on GitHub. The library will be available on PyPI soon.

https://t.co/8XuZIIsOhm