dart_transformers library
Classes
- ByteTokenizer
- Embedding
- FeedForward
- GPTConfig
- GPTModel
- KVCache
- LayerNorm
- Linear
- LinearLMHead
- MultiHeadAttention
- Ops
- RoPE
- RoPEConfig
- Sampler
- SamplerConfig
- Tensor2D
- TransformerBlock
- TransformerBlock implements a single transformer block with pre-layer normalization.