dsalt 0.1.0__tar.gz

dsalt-0.1.0/CHANGELOG.md

@@ -0,0 +1,33 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+### Added
+- Initial release of DSALT (Dynamic Sparse Attention with Landmark Tokens)
+- Triton-accelerated sparse attention kernels
+- Dynamic window sizing with learned predictors
+- Hybrid energy-based landmark token selection
+- Complete transformer implementation with DSALTAttention
+- Language model wrapper (DSALTLMHeadModel)
+- Training harness with mixed precision and DDP support
+- Comprehensive test suite
+- Documentation and examples
+
+### Changed
+- Migrated to modern Python packaging with pyproject.toml
+
+### Fixed
+- Various bug fixes in attention kernels and training loop
+
+## [0.1.0] - 2024-05-02
+
+### Added
+- Core DSALT implementation
+- Basic training infrastructure
+- CPU fallback for attention kernels
+- Initial test coverage

dsalt-0.1.0/LICENSE

@@ -0,0 +1,200 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the terms of this License, as
+      indicated by a copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based upon (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and derivative works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to use, reproduce, modify, merge, publish,
+      distribute, sublicense, and/or sell copies of the Work, and to
+      permit persons to whom the Work is furnished to do so, subject to
+      the following conditions:
+
+      The above copyright notice and this permission notice shall be
+      included in all copies or substantial portions of the Work.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, trademark, patent,
+          attribution and other notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright notice to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Support, Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, making the sure the file contains the path to the
+      Apache License. You may also attach the full text of the Apache License
+      instead of referencing it.
+
+   Copyright 2024 Leonardo Cofone
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

dsalt-0.1.0/MANIFEST.in

@@ -0,0 +1,26 @@
+include README.md
+include LICENSE
+include CHANGELOG.md
+include pyproject.toml
+include requirements*.txt
+
+# Include test files
+recursive-include tests *.py
+
+# Include documentation
+recursive-include docs *.md
+recursive-include docs *.rst
+recursive-include docs *.txt
+
+# Include examples
+recursive-include examples *.py
+recursive-include examples *.ipynb
+recursive-include examples *.md
+
+# Exclude development files
+exclude .gitignore
+exclude .pre-commit-config.yaml
+exclude tox.ini
+global-exclude *.pyc
+global-exclude __pycache__
+global-exclude .DS_Store

dsalt-0.1.0/PKG-INFO

@@ -0,0 +1,243 @@
+Metadata-Version: 2.4
+Name: dsalt
+Version: 0.1.0
+Summary: Dynamic Sparse Attention with Landmark Tokens — High-performance Triton implementation
+Author-email: Leonardo Cofone <leonardo.cofone@example.com>
+Maintainer-email: Leonardo Cofone <leonardo.cofone@example.com>
+License: Apache-2.0
+Project-URL: Homepage, https://github.com/yourusername/dsalt-pytorch
+Project-URL: Documentation, https://dsalt-pytorch.readthedocs.io/
+Project-URL: Repository, https://github.com/yourusername/dsalt-pytorch
+Project-URL: Issues, https://github.com/yourusername/dsalt-pytorch/issues
+Project-URL: Changelog, https://github.com/yourusername/dsalt-pytorch/blob/main/CHANGELOG.md
+Keywords: deep-learning,transformers,attention,sparse-attention,triton,pytorch
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: torch>=2.0.0
+Requires-Dist: numpy>=1.21.0
+Provides-Extra: triton
+Requires-Dist: triton>=2.0.0; extra == "triton"
+Provides-Extra: flash-attn
+Requires-Dist: flash-attn>=2.0.0; extra == "flash-attn"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
+Requires-Dist: black>=22.0.0; extra == "dev"
+Requires-Dist: isort>=5.10.0; extra == "dev"
+Requires-Dist: flake8>=4.0.0; extra == "dev"
+Requires-Dist: mypy>=1.0.0; extra == "dev"
+Requires-Dist: pre-commit>=2.20.0; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx>=5.0.0; extra == "docs"
+Requires-Dist: sphinx-rtd-theme>=1.2.0; extra == "docs"
+Requires-Dist: myst-parser>=0.18.0; extra == "docs"
+Provides-Extra: all
+Requires-Dist: dsalt[triton]; extra == "all"
+Requires-Dist: dsalt[flash-attn]; extra == "all"
+Requires-Dist: dsalt[dev]; extra == "all"
+Requires-Dist: dsalt[docs]; extra == "all"
+Dynamic: license-file
+
+# DSALT – Dynamic Sparse Attention with Landmark Tokens
+
+[![PyPI](https://img.shields.io/pypi/v/dsalt)](https://pypi.org/project/dsalt/)
+[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
+
+A high-performance PyTorch library implementing **DSALT** (Dynamic Sparse Attention with Landmark Tokens), a novel sparse attention mechanism that addresses noise accumulation and rank collapse in dense self-attention transformers.
+
+## 🚀 Key Features
+
+- **Efficient Sparse Attention**: Triton-accelerated kernels for GPU-optimized sparse causal self-attention
+- **Dynamic Window Sizing**: Adaptive local attention windows that grow with sequence position
+- **Landmark Token Selection**: Global landmark tokens selected via hybrid energy scoring
+- **Mixed Precision Training**: Full support for BF16/FP16 training with gradient scaling
+- **Distributed Training**: DDP (DistributedDataParallel) support for multi-GPU training
+- **Production Ready**: Complete training harness with checkpointing, logging, and validation
+
+## 📋 Table of Contents
+
+- [Installation](#installation)
+- [Quick Start](#quick-start)
+- [Architecture](#architecture)
+- [Training](#training)
+- [API Reference](#api-reference)
+- [Benchmarks](#benchmarks)
+- [Citation](#citation)
+- [License](#license)
+
+## 🛠️ Installation
+
+### Requirements
+- Python 3.8+
+- PyTorch 2.0+
+- CUDA 11.0+ (for GPU acceleration)
+- Triton 2.0+ (optional, for GPU kernels)
+
+### Install from source
+```bash
+git clone https://github.com/LeonardoCofone/dsalt-pytorch.git
+cd dsalt-pytorch
+pip install -e .
+```
+
+### Optional dependencies
+```bash
+pip install triton  # For GPU acceleration
+pip install flash-attn  # For Flash Attention 2 fallback
+```
+
+## 🚀 Quick Start
+
+```python
+import torch
+from dsalt.model import DSALTLMHeadModel
+
+# Create a DSALT language model
+model = DSALTLMHeadModel(
+    vocab_size=32000,
+    d_model=1024,
+    n_layers=24,
+    n_heads=16,
+    n_min=32,      # Minimum window size
+    n_max=512,     # Maximum window size
+    k_lmk=64,      # Number of landmark tokens
+)
+
+# Forward pass
+input_ids = torch.randint(0, 32000, (1, 1024))
+logits = model(input_ids)
+print(f"Output shape: {logits.shape}")  # [1, 1024, 32000]
+```
+
+## 🏗️ Architecture
+
+DSALT combines **local causal windows** with **global landmark tokens**:
+
+- **Local Attention**: Each token attends to a dynamic window of recent tokens
+- **Landmark Selection**: Top-k informative tokens selected globally via energy scoring
+- **Sparse Computation**: Only compute attention for relevant token pairs
+
+### Key Components
+
+- `DSALTTransformer`: Main transformer architecture
+- `DSALTAttention`: Multi-head sparse attention layer
+- `WindowSizePredictor`: Learned adaptive window sizing
+- `HybridEnergyScorer`: Landmark token selection
+- `SparseAttentionKernel`: Triton-accelerated attention computation
+
+## 🎯 Training
+
+### Single GPU Training
+```python
+from dsalt.training import DSALTTrainer
+from torch.utils.data import DataLoader
+
+trainer = DSALTTrainer(
+    model=model,
+    train_loader=train_dataloader,
+    val_loader=val_dataloader,
+    lr=3e-4,
+    total_steps=100000,
+    save_dir="checkpoints",
+    dtype=torch.bfloat16,
+)
+
+trainer.train()
+```
+
+### Multi-GPU Distributed Training
+```python
+import torch.distributed as dist
+
+# Initialize process group
+dist.init_process_group(backend='nccl')
+
+trainer = DSALTTrainer(
+    model=model,
+    train_loader=train_dataloader,
+    val_loader=val_dataloader,
+    ddp=True,  # Enable DDP
+    # ... other args
+)
+```
+
+## 📚 API Reference
+
+### Core Classes
+
+- `DSALTLMHeadModel`: Language model wrapper with LM head
+- `DSALTTransformer`: Base transformer architecture
+- `DSALTAttention`: Sparse attention module
+- `DSALTTrainer`: Training harness
+
+### Kernel Functions
+
+- `dsalt_attention()`: Main sparse attention function
+- `compute_hybrid_energy_scores()`: Landmark scoring
+- `select_landmarks()`: Landmark selection
+
+## 🧪 Testing
+
+Run the full test suite:
+```bash
+python tests/test.py
+```
+
+Run specific tests:
+```bash
+python tests/test_sparse_attn.py  # Attention kernels
+python tests/test_dsalt_lm.py     # LM wrapper
+```
+
+## 📊 Benchmarks
+
+DSALT achieves significant speedups over dense attention:
+
+- **Memory**: O(n) vs O(n²) for dense attention
+- **Speed**: 2-5x faster training on long sequences
+- **Quality**: Maintains perplexity comparable to dense models
+
+*Detailed benchmarks coming soon*
+
+## 📖 Citation
+
+If you use DSALT in your research, please cite our paper:
+
+```bibtex
+@article{dsalt2024,
+  title={Noise Accumulation and Rank Collapse in Dense Self-Attention: DSALT},
+  author={Your Name et al.},
+  journal={arXiv preprint},
+  year={2024}
+}
+```
+
+Paper: [https://zenodo.org/records/19312827](https://zenodo.org/records/19312827)
+
+## 🤝 Contributing
+
+We welcome contributions! Please see our [contributing guidelines](CONTRIBUTING.md).
+
+## 📄 License
+
+This project is licensed under the Apache 2.0 License - see the [LICENSE](LICENSE) file for details.
+
+## 🙏 Acknowledgments
+
+- Built on top of [Triton](https://github.com/openai/triton) for GPU kernels
+- Inspired by [Flash Attention](https://github.com/Dao-AILab/flash-attention)
+- Thanks to the PyTorch team for the excellent deep learning framework

dsalt-0.1.0/README.md

@@ -0,0 +1,189 @@
+# DSALT – Dynamic Sparse Attention with Landmark Tokens
+
+[![PyPI](https://img.shields.io/pypi/v/dsalt)](https://pypi.org/project/dsalt/)
+[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
+
+A high-performance PyTorch library implementing **DSALT** (Dynamic Sparse Attention with Landmark Tokens), a novel sparse attention mechanism that addresses noise accumulation and rank collapse in dense self-attention transformers.
+
+## 🚀 Key Features
+
+- **Efficient Sparse Attention**: Triton-accelerated kernels for GPU-optimized sparse causal self-attention
+- **Dynamic Window Sizing**: Adaptive local attention windows that grow with sequence position
+- **Landmark Token Selection**: Global landmark tokens selected via hybrid energy scoring
+- **Mixed Precision Training**: Full support for BF16/FP16 training with gradient scaling
+- **Distributed Training**: DDP (DistributedDataParallel) support for multi-GPU training
+- **Production Ready**: Complete training harness with checkpointing, logging, and validation
+
+## 📋 Table of Contents
+
+- [Installation](#installation)
+- [Quick Start](#quick-start)
+- [Architecture](#architecture)
+- [Training](#training)
+- [API Reference](#api-reference)
+- [Benchmarks](#benchmarks)
+- [Citation](#citation)
+- [License](#license)
+
+## 🛠️ Installation
+
+### Requirements
+- Python 3.8+
+- PyTorch 2.0+
+- CUDA 11.0+ (for GPU acceleration)
+- Triton 2.0+ (optional, for GPU kernels)
+
+### Install from source
+```bash
+git clone https://github.com/LeonardoCofone/dsalt-pytorch.git
+cd dsalt-pytorch
+pip install -e .
+```
+
+### Optional dependencies
+```bash
+pip install triton  # For GPU acceleration
+pip install flash-attn  # For Flash Attention 2 fallback
+```
+
+## 🚀 Quick Start
+
+```python
+import torch
+from dsalt.model import DSALTLMHeadModel
+
+# Create a DSALT language model
+model = DSALTLMHeadModel(
+    vocab_size=32000,
+    d_model=1024,
+    n_layers=24,
+    n_heads=16,
+    n_min=32,      # Minimum window size
+    n_max=512,     # Maximum window size
+    k_lmk=64,      # Number of landmark tokens
+)
+
+# Forward pass
+input_ids = torch.randint(0, 32000, (1, 1024))
+logits = model(input_ids)
+print(f"Output shape: {logits.shape}")  # [1, 1024, 32000]
+```
+
+## 🏗️ Architecture
+
+DSALT combines **local causal windows** with **global landmark tokens**:
+
+- **Local Attention**: Each token attends to a dynamic window of recent tokens
+- **Landmark Selection**: Top-k informative tokens selected globally via energy scoring
+- **Sparse Computation**: Only compute attention for relevant token pairs
+
+### Key Components
+
+- `DSALTTransformer`: Main transformer architecture
+- `DSALTAttention`: Multi-head sparse attention layer
+- `WindowSizePredictor`: Learned adaptive window sizing
+- `HybridEnergyScorer`: Landmark token selection
+- `SparseAttentionKernel`: Triton-accelerated attention computation
+
+## 🎯 Training
+
+### Single GPU Training
+```python
+from dsalt.training import DSALTTrainer
+from torch.utils.data import DataLoader
+
+trainer = DSALTTrainer(
+    model=model,
+    train_loader=train_dataloader,
+    val_loader=val_dataloader,
+    lr=3e-4,
+    total_steps=100000,
+    save_dir="checkpoints",
+    dtype=torch.bfloat16,
+)
+
+trainer.train()
+```
+
+### Multi-GPU Distributed Training
+```python
+import torch.distributed as dist
+
+# Initialize process group
+dist.init_process_group(backend='nccl')
+
+trainer = DSALTTrainer(
+    model=model,
+    train_loader=train_dataloader,
+    val_loader=val_dataloader,
+    ddp=True,  # Enable DDP
+    # ... other args
+)
+```
+
+## 📚 API Reference
+
+### Core Classes
+
+- `DSALTLMHeadModel`: Language model wrapper with LM head
+- `DSALTTransformer`: Base transformer architecture
+- `DSALTAttention`: Sparse attention module
+- `DSALTTrainer`: Training harness
+
+### Kernel Functions
+
+- `dsalt_attention()`: Main sparse attention function
+- `compute_hybrid_energy_scores()`: Landmark scoring
+- `select_landmarks()`: Landmark selection
+
+## 🧪 Testing
+
+Run the full test suite:
+```bash
+python tests/test.py
+```
+
+Run specific tests:
+```bash
+python tests/test_sparse_attn.py  # Attention kernels
+python tests/test_dsalt_lm.py     # LM wrapper
+```
+
+## 📊 Benchmarks
+
+DSALT achieves significant speedups over dense attention:
+
+- **Memory**: O(n) vs O(n²) for dense attention
+- **Speed**: 2-5x faster training on long sequences
+- **Quality**: Maintains perplexity comparable to dense models
+
+*Detailed benchmarks coming soon*
+
+## 📖 Citation
+
+If you use DSALT in your research, please cite our paper:
+
+```bibtex
+@article{dsalt2024,
+  title={Noise Accumulation and Rank Collapse in Dense Self-Attention: DSALT},
+  author={Your Name et al.},
+  journal={arXiv preprint},
+  year={2024}
+}
+```
+
+Paper: [https://zenodo.org/records/19312827](https://zenodo.org/records/19312827)
+
+## 🤝 Contributing
+
+We welcome contributions! Please see our [contributing guidelines](CONTRIBUTING.md).
+
+## 📄 License
+
+This project is licensed under the Apache 2.0 License - see the [LICENSE](LICENSE) file for details.
+
+## 🙏 Acknowledgments
+
+- Built on top of [Triton](https://github.com/openai/triton) for GPU kernels
+- Inspired by [Flash Attention](https://github.com/Dao-AILab/flash-attention)
+- Thanks to the PyTorch team for the excellent deep learning framework

dsalt-0.1.0/dsalt/__init__.py

@@ -0,0 +1,5 @@
+from .modules.dsalt_attention import DSALTAttention
+from .modules.dsalt_transformer import DSALTTransformer
+from .kernels.sparse_attn import dsalt_attention
+
+__version__ = "0.1.0"