search

huggingface / nanotron

Minimalistic large language model 3D-parallelism training
Apache License 2.0
1.14k stars 107 forks source link

[Refactor] DistributedOptimizer and FP32GradAccum #20

Open NouamaneTazi opened 8 months ago

NouamaneTazi commented 8 months ago

DistributedOptimizer and FP32GradAccum need to be refactored