Statements (32)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:infrastructure
|
| gptkbp:availableOn |
gptkb:Google_Cloud_Platform
|
| gptkbp:composedOf |
gptkb:Tensor_Processing_Units
|
| gptkbp:developedBy |
gptkb:Google
|
| gptkbp:enables |
large-scale distributed training
training of large models |
| gptkbp:firstReleased |
2017
|
| https://www.w3.org/2000/01/rdf-schema#label |
TPU Pods
|
| gptkbp:marketedAs |
Cloud TPU Pods
|
| gptkbp:offers |
high-speed interconnect
petaflops of compute |
| gptkbp:relatedTo |
gptkb:TPU_v4
gptkb:TPU_v5e gptkb:TPU_v2 gptkb:TPU_v3 |
| gptkbp:scalableTo |
thousands of chips
|
| gptkbp:supports |
gptkb:TensorFlow
gptkb:JAX gptkb:PyTorch distributed training |
| gptkbp:usedBy |
gptkb:researchers
enterprises |
| gptkbp:usedFor |
gptkb:artificial_intelligence
gptkb:machine_learning computer vision deep learning natural language processing transformer models BERT training GPT model training |
| gptkbp:bfsParent |
gptkb:TPUs
|
| gptkbp:bfsLayer |
6
|