AI - AI Dynamics

→ View original post on X — @jordannovet

Company Revenue Beats Expectations: German Word for Overperformance

By

@jordannovet

–

23 August 2023 22h36

is there a German word for when company says it will grow revenue 64% but ends up growing revenue by 101%

→ View original post on X — @jordannovet

Trillion-Dollar Company Milestone Achievement

By

@jordannovet

–

23 August 2023 22h28

this is a trillion-dollar company

LoRA Compared to Llama-Adapter and Llama-Adapter v2

By

–

23 August 2023 22h19

LoRA is a parameter-efficient finetuning technique, yes. I recently compared to Llama-Adapter and Llama-Adapter v2:

→ View original post on X — @jeande_d

Deep Learning Indaba 2023 in Accra: African AI Community Meeting

By

@jeande_d

–

23 August 2023 22h10

Deep Learning Indaba 2023 is happening in the next few days. It's the annual meeting of African AI community, happening in Accra this time. Many students from different countries in Africa wish to attend the event, but it's not easy for them to get the travel expenses. There is

Alternative Precision Flag for Non-GPU Machines

By

–

23 August 2023 22h04

I haven't tried on non-GPU machines, but maybe the following works:
"–precision 16-true" instead of "–precision bf16-true"

BFloat16 Support in Ampere GPUs and M1 Chips

By

–

23 August 2023 22h02

BFloat16 is a format introduced in Ampere architectures (Nvidia A100 cards) or newer. M1 chips for some reason have also bfloat16 in PyTorch, but I think this MPS bfloat16 is something very different.

→ View original post on X — @groqinc

Enterprise AI ROI: Master LLM Inference Strategy for Maximum Impact

By

@groqinc

–

23 August 2023 21h48

Get your inference strategy right & your enterprise will achieve a generational leap in the ROI of AI solutions for LLMs & other revolutionary workloads. For guidance, get our white paper, Key Enterprise Considerations for Inference Deployment of LLMs. http://
groq.com/inference

BFloat16 Weight Clipping Limits and Extreme Value Representation

By

–

23 August 2023 21h42

Hm, so that means all weights above values 65504 and below -65504 will be clipped. Does anyone know of there are numbers that large/small in the original bfloat16 representation?

→ View original post on X — @jeande_d

Natural Language Processing Course UT Austin Lecture Videos and Materials

By

@jeande_d

–

23 August 2023 21h39

Natural Language Processing – UT Austin Lecture videos: https://
youtube.com/playlist?list=
PLofp2YXfp7TZZ5c7HEChs0_wfEfewLDs7
… Website(for slides, readings): https://
cs.utexas.edu/~gdurrett/cour
ses/online-course/materials.html
…