graphcore-research/out-of-the-box-fp8-training
Demo of the unit_scaling library, showing how a model can be easily adapted to train in FP8.
Jupyter NotebookMIT
Stargazers
- abcp4
- accesine
- balancap@Graphcore
- Banguiskode
- bilelomrani1@DynamoFL
- bobbercheng
- C-TCZurich, Switzerland
- carmocca@Lightning-AI
- dumpmemory
- ethansmith2000florida
- f-dangel@ProbabilisticNumerics @VectorInstitute
- FarisHijaziking fahd university of petroleum and minerals
- fly51flyPRIS
- isamu-isozakiKulicke & Soffa
- jmargetaKardioMe
- justinchuby@microsoft
- kyegomezSwarms
- Pent
- progerSupercomputer City
- s-maddrellmanderDayhoff Labs
- SandalotsVolcanak
- sheng-qin
- simonJJJDAMO Academy
- Siris-LiPeking University
- spencerfreiUC Davis
- Sudo42bhome
- tantara@bytedance
- Thalituto
- thecharlieblakeLondon
- vahbuna
- vvvm23Cohere
- xrsrke@huggingface
- zauberresonatorSan Jose, CA, USA
- zigforge