Transformer Lens

Cons

• Quantization and not flash attention not supported which cause more memory

Supported model

Model Properties Table - TransformerLens Documentation

https://transformerlensorg.github.io/TransformerLens/generated/model_properties_table.html

Sites