Now we'll have to see if any of those API providers make a move to drop their product into something that works in a VPC. Until then, it looks like we're starting to see how the "business" of the current iteration of generative AI will play out.
SOFTWARE
-
API Providers Business Models and Enterprise Data Security Concerns
By
–
And as OpenAI puts together its business model, so too with other API providers likely follow. The challenge, though, is that companies that are skittish about using APIs don't want additional controls and a promise that their companies won't use their data—they want it in a VPC.
-
OpenAI GPT-4 API Rate Limiting Issues Impact Enterprise Performance
By
–
OpenAI is trying to pitch some controls and privacy additions for enterprise companies, but from companies I talk to, this is about performance and reliability. he GPT-4 API is really powerful, but users also end up running into issues with rate limiting and have to drop to 3.5T.
-
ChatGPT Pricing Tiers: Free to Enterprise Strategy
By
–
A "free" version that employees can use internally (w/ IT's blessing or not) by copy/pasting into the interface. A "base" version with the 3.5-Turbo API. A "plus" version with the fine-tuned 3.5-Turbo API. A "premium" version of the base GPT-4 API. And finally an enterprise tier.
-
Efficient LLM Training QLoRA Llama 2 Resource Optimization
By
–
Depends on your settings. But if you limit the context size to like 2048 (like in the NeurIPS competition) and use a microbatch size of 1 with gradient accumulation and qlora with llama 2 7B, that’s approx 20 GB RAM and shouldn’t take too long, maybe an hour.
-
Master Data Analytics Skills with SAS Free Learning Platform
By
–
To use software effectively, solve complex problems + build a career in data analytics, having skills to work with data is critical. ✅
— SAS Software (@SASsoftware) 31 août 2023
SAS makes it easy with a modern learning platform. Try it free for 7 days. #LearnSAS https://t.co/2emm8WoqFP pic.twitter.com/As0cZ8Gv8nTo use software effectively, solve complex problems + build a career in data analytics, having skills to work with data is critical. SAS makes it easy with a modern learning platform. Try it free for 7 days. #LearnSAS http://
2.sas.com/6016PtSwg -
Philippe’s Trolling Reshape Function Naming Convention
By
–
looks like phillippe wanted to troll everyone by calling what should've been `tl.reshape` as `tl.view`.
It's like those trolly C macros ( #define float int ) -
Google Cloud MultiSlice Scales ML Training Across TPU Pods
By
–
The @googlecloud MultiSlice system (see blog post in tweet below) is an external version of some internal software+hardware we've been using for a while to scale some of our largest ML training runs across multiple TPUv4 and TPUv5e pods. Scaling made easy!
-
BART Abstractive Text Summarization with KerasNLP and Keras Core
By
–
Abstractive text summarization with BART — using KerasNLP and Keras Core, with JAX
-
Microsoft Plans Office Web Apps Integration in Competing Services
By
–
'We will develop a new method for hosting the Office web applications within competing apps and services much like Microsoft accomplishes in Teams.' https://
blogs.microsoft.com/eupolicy/2023/
08/31/european-competition-teams-office-microsoft-365/
…