OPEN SOURCE - AI Dynamics

Happy Birthday to C++ Creator Bjarne Stroustrup

By

@mit_csail

–

30 December 2022 18h01

Happy birthday to Bjarne Stroustrup, inventor of C++: http://
bit.ly/2E33g6O

→ View original post on X — @mit_csail

30 December 2022

LangChain v0.0.53 Release: Google Search Integration

By

@langchain

–

30 December 2022 17h06

LangChain v0.0.53 Doc fixes from @george_zhou Bug fix (highlighted by @yvesbalbas
) Also includes maybe the most requested feature to date… A google search alternative to SerpAPI (by @nickscamara_
) More on this in below

→ View original post on X — @langchain

30 December 2022

Using nlp.pipe() for faster batched predictions in spaCy

By

@honnibal

–

30 December 2022 13h18

Nice example! Here's a tip for next time as well: the nlp.pipe() function operates over a sequence of inputs, and lets the model make batched predictions, making it much faster:

→ View original post on X — @honnibal

30 December 2022

Google’s PaLM benchmark results and open-source RLHF implementation

By

@aibreakfast

–

30 December 2022 4h00

Google's PaLM beats GPT-3 on 28 out of 29 Natural Language Processing tasks (!) and was just open-sourced for reinforcement training as of today: https://
github.com/lucidrains/PaL
M-rlhf-pytorch
…

→ View original post on X — @aibreakfast

30 December 2022

FlashAttention Revert and torch.compile Integration Challenges

By

@karpathy

–

30 December 2022 2h14

Ah, I reverted FlashAttention in this run because it made code messier. Will look into incorporating it back, but yes not sure how nicely it plays with torch.compile. The usual problem with taking on large dependencies you don't understand ;(

→ View original post on X — @karpathy

30 December 2022

PyTorch Compile Optimization Reduces Iteration Time by 36%

By

@karpathy

–

30 December 2022 1h56

To follow up, I had a chance to try it btw:
before: 212ms / iter
>>> model = torch.compile(model)
after: 135ms / iter
nice 🙂

→ View original post on X — @karpathy

30 December 2022

LangChain v0.052 Adds Watch Mode Tests Agent Security

By

@langchain

–

29 December 2022 15h30

v0.052 Add watch mode for tests – @nfcampos Add schema property to SQL DB class – @diwanksingh Propagate verbose flag to combine-document chains – @parth_29 Add max_iterations kwarg to agent executor, in part to stop attacks like the below

→ View original post on X — @langchain

29 December 2022

PaLM 540B Architecture Code Released Open Source

By

@dotcsv

–

29 December 2022 13h26

Mola, pero… ¿quién tiene capacidad de entrenar un PaLM de 540 billions (x3 el tamaño de GPT-3)? Aquí están compartiendo el código de la arquitectura y no un modelo pre-entrenado. Con lo que es un entrénalo-tú-mismo de ChatGPT.

→ View original post on X — @dotcsv

29 December 2022

Organizations Getting AI and Data Privacy Right

By

@thegautamkamath

–

28 December 2022 19h08

Right, there are certainly bad actors out there. But nonetheless there remain a few organizations that seem to do it right (some that immediately come to mind include Microsoft, Google, @uscensusbureau
, @TumultLabs, and @opendp_org
).

→ View original post on X — @thegautamkamath

28 December 2022