Big Data Science
@bdscience
📊💡Dataset for setting up mathematical models
OpenMathInstruct-1 is a fresh synthetic dataset from NVIDIA, created for training mathematical models. This dataset includes 1.8 million problem-solution pairs intended for training.
As the developers note, this dataset was created by synthesizing code interpreter solutions for GSM8K and MATH, two popular calculus tests, using the recently released and permissively licensed Mixtral model.
OpenMathInstruct-1 is a fresh synthetic dataset from NVIDIA, created for training mathematical models. This dataset includes 1.8 million problem-solution pairs intended for training.
As the developers note, this dataset was created by synthesizing code interpreter solutions for GSM8K and MATH, two popular calculus tests, using the recently released and permissively licensed Mixtral model.
huggingface.co
nvidia/OpenMathInstruct-1 · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
3 2.4K
Обсуждение 0
Обсуждение не доступно в веб-версии. Чтобы написать комментарий, перейдите в приложение Telegram.
Обсудить в Telegram