SOFTWARE - AI Dynamics

MosaicML’s Infrastructure Enables Seamless MPT-30B Model Training

By

–

27 June 2023 20h52

One of the big selling points for MosaicML was its infrastructure that allows enables companies to train complex models without any major hiccups. It launched its model MPT-30B with the same infrastructure and chief scientist @jefrankle said faced it saw no issues in training.

→ View original post on X — @mattlynley

27 June 2023

Databricks Acquires MosaicML to Lower ML Model Development Costs

By

@mattlynley

–

27 June 2023 20h48

Databricks, meanwhile, wants to build tools to enable companies to build machine learning models from scratch and serve them. That end-to-end is reinforced by its acquisition of MosaicML, which aimed to bring down the cost of building custom ML models from scratch for companies.

→ View original post on X — @mattlynley

27 June 2023

Snowflake launches Snowpark Container Services for ML applications

By

@mattlynley

–

27 June 2023 20h48

Snowflake said it was launching Snowpark Container Services, which enables its users to deploy native ML apps like Hex. It's leaning heavily on partners and building tools to try open the door to a variety of companies attacking problems it COULD attack, but isn't interested.

→ View original post on X — @mattlynley

27 June 2023

Harvard’s new CS50 teacher is a ChatGPT-based chatbot starting September

By

@aibreakfast

–

27 June 2023 20h19

Harvard’s new computer science teacher is a chatbot: Students enrolled in the university’s flagship CS50 course will be presented with the AI teacher based on ChatGPT in September.

→ View original post on X — @aibreakfast

27 June 2023

iPhone App Reliability Update Approaching Feature Parity

By

@aravsrinivas

–

27 June 2023 17h20

Thanks! The latest version of the Iphone app should be a lot more reliable and not freeze for certain queries. It will continue to get better this week and next and attain feature parity with the web soon.

→ View original post on X — @aravsrinivas

27 June 2023

High-speed tensor core inference with massive memory bandwidth

By

@id_aa_carmack

–

27 June 2023 17h10

So, you have a 200 TB/s fire hose of recycling information, which you would have to “pump up” over many cycles from much slower conventional memory, but it could then feed straight in to a field of tensor core registers for real time model inference. Fun to think about!

→ View original post on X — @id_aa_carmack

27 June 2023

Baby AGI: Task Automation Python Script with OpenAI

By

@marktabnet

–

27 June 2023 16h07

Baby AGI is a python script which uses OpenAI and Pinecone APIs, and the LangChain framework to create, organize, prioritize as well as the executing of tasks. https://
bit.ly/3qSgQac

→ View original post on X — @marktabnet

27 June 2023

TF Data: Powerful API Becoming Industry Standard for Deep Learning

By

@fchollet

–

27 June 2023 5h36

TF Data is an incredibly powerful data streaming and preprocessing API for deep learning (no reliance on the Python runtime!). It has become the standard not only for the TensorFlow ecosystem, but also for the JAX ecosystem. I could see it become an industry standard.

→ View original post on X — @fchollet

27 June 2023