attention_sparsity

Modules

modelopt.torch.sparsity.attention_sparsity.calibration

Calibration framework for sparse attention methods.

modelopt.torch.sparsity.attention_sparsity.config

Configuration classes for sparse attention optimization.

modelopt.torch.sparsity.attention_sparsity.conversion

Conversion and restoration utilities for sparse attention.

modelopt.torch.sparsity.attention_sparsity.methods

Sparse attention methods package.

modelopt.torch.sparsity.attention_sparsity.mode

Sparse attention mode descriptor for ModelOpt.

modelopt.torch.sparsity.attention_sparsity.model_sparsify

Main API functions for sparse attention optimization.

modelopt.torch.sparsity.attention_sparsity.plugins

Plugins for sparse attention integration with various frameworks.

modelopt.torch.sparsity.attention_sparsity.sparse_attention

Extensible sparse attention module.

modelopt.torch.sparsity.attention_sparsity.stats_manager

Statistics manager for sparse attention modules.

modelopt.torch.sparsity.attention_sparsity.utils

Utility functions for sparse attention module discovery.

Extensible sparse attention optimization for transformer models.