Piyush Agarwal

AI+ML Consulting | Python Dev | NLP | Blogger | Hiker

[Chip Huyen @chipro]

Summary of Gemini's 60-page technical report.

1. Written in Jax and trained using TPUs. The architecture, while not explained in details, seems similar to Flamigo's.

2. Gemini Pro's performance is similar to GPT-3.5 and Gemini Ultra is reported to be better than GPT-4. Nano-1 (1.8B params) and Nano-2 (3.25B params) are designed to run on-device.

3. 32K context length.

4. Very good at understanding vision and speech.

5. Coding ability: the big jump in HumanEval compared to GPT-4 (74.4% vs. 67%), if true, is awesome. However, the Natural2Code benchmark (no leakage on the Internet) shows a much smaller gap (74.9% vs. 73.9%).

6. On MMLU: using COT@32 (32 samples) to show that Gemini is better than GPT-4 seems forced. In 5-shot setting, GPT-4 is better (86.4% vs. 83.7%).

7. No information at all on the training data, other than they ensured "all data enrichment workers are paid at least a local living wage."

Full report: https://t.co/ZvgTUFPdH5

[Piyush Agarwal @impiyushag]

Another cancelled flight after 4+ hrs of wait @united #getyourshittogether #disappointedcustomers

[Piyush Agarwal @impiyushag]

Another day and another crew shortage at @united. Been on the airport with an infant for 24+ hrs. #dontflyunited #worstairline

[Piyush Agarwal @impiyushag]

@united we’ve been waiting over 6 hours. First it was the incoming flight delay, and then crew time out with no re-assignment. Very dissatisfied and disappointed! #unitedairlinedelays #dissapointedcustomers

[François Chollet @fchollet]

We're launching Keras Core, a new library that brings the Keras API to JAX and PyTorch in addition to TensorFlow.

It enables you to write cross-framework deep learning components and to benefit from the best that each framework has to offer.

Read more: https://t.co/xmmxBfSZgh

[Hugging Face @huggingface]

Today we are excited to announce a new partnership with @awscloud! 🔥

Together, we will accelerate the availability of open-source machine learning 🤝

Read the post 👉 https://t.co/fb77d1J2qX

[Andrej Karpathy @karpathy]

!!!! Ok I recorded a (new!) 2h25m lecture on "The spelled-out intro to neural networks and backpropagation: building micrograd" https://t.co/KQ23lQW1BT .
This is the culmination of about 8 years of obsessing about the best way to explain neural nets and backprop.

[Piyush Agarwal @impiyushag]

This course… 🙏🏾🙌🏾👏🏾

[Jeremy Howard @jeremyphoward]

After 2 years, Practical Deep Learning for Coders v5 is finally ready! 🎊

This is a from-scratch rewrite of our most popular course. It has a focus on interactive explorations, & covers @PyTorch, @huggingface, DeBERTa, ConvNeXt, @Gradio & other goodies 🧵
https://t.co/nzv7pek0iq

[Radek Osmulski @radekosmulski]

Would you like to make your life easier?

I certainly do! 😄

My brilliant colleagues have just published a blog post on how Merlin Models have been designed to do just that!

✅ DRY across the entire pipeline 🔥
✅ industry best practices
... and more!

https://t.co/FD7yx26diA

[Chitwan Saharia @Chitwan_Saharia]

We are thrilled to announce Imagen, a text-to-image model with unprecedented photorealism and deep language understanding. Explore https://t.co/mSplg4FlsM and Imagen!

A large rusted ship stuck in a frozen lake. Snowy mountains and beautiful sunset in the background. #imagen

[Piyush Agarwal @impiyushag]

@radekosmulski @NVIDIAAI @fastdotai @kaggle Congratulations! You are an inspiration…

[elvis @omarsar0]

🎉 Introducing ML Course Notes 🎉

A new repo to publish and share course notes related to machine learning, NLP, and AI.

I'll be releasing all my notes there. Contributions appreciated!

https://t.co/f3G6ARdH11

[Andrej Karpathy @karpathy]

New blog post!⬆️ Deep Neural Nets: 33 years ago and 33 years from now https://t.co/pbZvYh3Mck we reproduce what I think may be the earliest real-world application of a neural net trained end-to-end with backprop (LeCun et al. 1989), try improve it with time travel, and reflect.

[Piyush Agarwal @impiyushag]

Getting more and more into MLOps. Reading and learning recommendations welcome! #MLOps #MLEngineer

[Piyush Agarwal @impiyushag]

#learnsomethingneweveryday

[Hamel Husain @HamelHusain]

I've been using this site for years. I don't know how people do regex without something like this

https://t.co/FUJPYJZBA1

[Piyush Agarwal @impiyushag]

@A_K_Nain 😂😂

[Sanyam Bhutani @bhutanisanyam1]

An absolute masterclass by World's Top Data Scientists 🙏

The awesome Kaggle Grandmaster Team at NVIDIA shares their winning tips and tricks in this series:

https://t.co/o8bVaZB7Qq

[Piyush Agarwal @impiyushag]

Congrats Black #PythonCodeFormatter team. Very excited for these changes

[Łukasz Langa @llanga]

1/4 Black, your friendly #Python code formatter, is no longer beta! The maintainers just released the first ever version we're confident enough to call stable. I'm not crying. YOU'RE crying! 😭

The list of changes is pretty extensive:
https://t.co/k3MwKqKiDH

[Piyush Agarwal @impiyushag]

My reading for the weekend!!
#transformers #nlp

Thank you @PacktPub for sending this over

[abhishek @abhi1thakur]

🚀 In 2022, if you want to learn

🐍 Python
💻 Machine Learning / Deep Learning
📈 Data Science

Then follow your passion, stop slacking and start immediately. There are unlimited free resources available online. Choose carefully and happy learning 🙂