TLDRocket
Sign in

Tools & Coding

975 summarised stories in Tools & Coding, each linking back to the original source. Browse all topics →

Tuesday, 28 June 2022

Announcing Evaluation on the Hub

Hugging Face 4 years ago 39

Hugging Face launched Evaluation on the Hub, a no-code tool powered by AutoTrain that allows users to evaluate any machine learning model on any dataset directly through the platform. The tool was used to evaluate hundreds of models on key datasets, with results automatically added to model cards in a standardized format. This enables researchers and practitioners to compare model performance, reproduce results, and select models for deployment without writing code or worrying about implementation inconsistencies.

Read the full story
Hugging Face Model Evaluation

Accelerate Large Model Training using DeepSpeed

Hugging Face 4 years ago 3

The Accelerate library enables training of large models using DeepSpeed's ZeRO optimization technique, which distributes optimizer states, gradients, and model parameters across GPUs and CPU memory. In a benchmark with a 900M-parameter DeBERTa model, DeepSpeed ZeRO Stage-2 achieved a batch size of 40 compared to DDP's maximum of 8, resulting in 3.5X faster training time without performance loss. This allows users to fit larger models and batch sizes on existing hardware without modifying their training code.

Read the full story
Model Training

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.