Home
Getting Started
Dataset Formats
Deployments
How To Guides
Core Concepts
Advanced Features
Troubleshooting
- FAQ
- Debugging
- NCCL

On this page

monkeypatch.llama_patch_multipack

monkeypatch.llama_patch_multipack

monkeypatch.llama_patch_multipack

Patched LlamaAttention to use torch.nn.functional.scaled_dot_product_attention