Will AI take away all our jobs or will it augment and enable us to create something greater? Dell Tech World unfolded last week, a grand spectacle of AI prowess that drew over 10,000 eager minds. I found myself immersed in the heart of AI, rubbing shoulders with industry
GENERATIVE AI
-
Flash Praised as Excellent Small Language Model
By
–
Yes, I’ve said this before, but Flash is a crazy good small model
-
AI Capabilities Advancing Faster Than Evaluation Methods Can Track
By
–
The capabilities are improving so fast that evals can't keep up
-

NVIDIA’s Toy Jensen: Photorealistic AI Avatar Powered by Advanced Tech
By
–
Toy Jensen (TJ) was first introduced in the GTC 2021 November keynote. He's a photorealistic, lifelike autonomous toy avatar powered by NVIDIA AI technologies that responds to challenging domain-specific questions. Share your favorite keynote moment with #NVIDIAkeynote.
-

LLM Evaluation Improvements and the Challenge of Creating Good Benchmarks
By
–
Nice, a serious contender to @lmsysorg in evaluating LLMs has entered the chat. LLM evals are improving, but not so long ago their state was very bleak, with qualitative experience very often disagreeing with quantitative rankings. This is because good evals are very difficult
-
Jay Kreps Joins Anthropic’s Board of Directors
By
–
Welcoming @jaykreps to Anthropic's Board of Directors:
-
Optimizing batch size and throughput for GPT-3 training
By
–
Okay that's good to know 🙂
I was just following the GPT-3 paper numbers in the table, but like I mentioned it's possible the settings are way too conservative.
Few more things to try: we want to increase the batch size as much as possible to get higher tok/s. Are you using -r 1 -

GPT-4 Usage Cap and Fallback to GPT-4o for Pro Users
By
–
GPT-4 has a usage cap of 40 messages every 3 hours for Pro users. Once you run out, it'll put a message up that you're running on GPT-4o until the next time period.
-
Release Previous AI Model Open-Source on Hugging Face
By
–
very cool! you should release a previous model in open-source on HF, that would make the field go crazy!