Andrew Ng Praises Marin Project Openness in AI
Highlights Percy Liang's lab releasing code, data, recipes, and results for the project.
Andrew Ng called the Marin project a key demonstration of openness in model training. He noted its release of code, data, recipes, and experimental results under Percy Liang's lab approach, which he said used to be the norm in AI research. Jesse Dodge replied with enthusiasm for the open science and asked how the team forecast loss for intermediate checkpoints, offering to review any related paper.
Combined views
172.9K
2 posts, first seen 18h ago
Andrew Ng Praises Marin Project Openness in AI
Highlights Percy Liang's lab releasing code, data, recipes, and results for the project.
Andrew Ng called the Marin project a key demonstration of openness in model training. He noted its release of code, data, recipes, and experimental results under Percy Liang's lab approach, which he said used to be the norm in AI research. Jesse Dodge replied with enthusiasm for the open science and asked how the team forecast loss for intermediate checkpoints, offering to review any related paper.
Reactions from ranked influencers
2 posts
In the fight to defend openness in AI, the Marin project is a precious demonstration of openness in model training, with open code, data, recipes, even experimental results. Releasing AI research openly used to be the norm; I'm grateful for @percyliang's open lab approach.
🚢 Marin 535B-A23B started training this week! As usual, the whole process is open. Voyage plan: pretraining (80%) + midtraining (20%) on 18.75T tokens on 11 x GB200 NVL72 for ~3 months (2.7e24 FLOPs). Post-training will follow. Before kicking off the run, we trained a 4-rung…
@percyliang very cool! love to see the open science 😄 how did you forecast the loss of intermediate checkpoints? happy to check out a paper if there is one!
