.gitignore
CONTRIBUTING.md
CONTRIBUTING.zh.md
LICENSE
MANIFEST.in
NOTICE
README.md
README.zh.md
pyproject.toml
docs/ascend-adaptation-vs-triattention-vllm.md
docs/ascend-adaptation-vs-triattention-vllm.zh.md
docs/benchmarking.md
docs/benchmarking.zh.md
docs/calibration-artifacts.md
docs/calibration-artifacts.zh.md
docs/environment-installation.md
docs/environment-installation.zh.md
docs/kv-compress-test-requirements.md
docs/kv-compress-test-requirements.zh.md
docs/methods.md
docs/methods.zh.md
docs/ownership-and-licensing.md
docs/ownership-and-licensing.zh.md
docs/packaging-and-release.md
docs/packaging-and-release.zh.md
docs/resuts.md
docs/resuts.zh.md
docs/validation.md
docs/validation.zh.md
docs/dev/ab-validation-2026-08-27.md
docs/dev/ab-validation-2026-08-27.zh.md
docs/dev/deep-optimization-2026-08-28.md
docs/dev/deep-optimization-2026-08-28.zh.md
docs/dev/kv-compress-v3.8-test-report-2026-08-28.md
docs/dev/kv-compress-v3.8-test-report-2026-08-28.zh.md
docs/dev/kv-compress-v3.8-test-requirements-2026-08-28.md
docs/dev/kv-compress-v3.8-test-requirements-2026-08-28.zh.md
docs/dev/reports/weekly/2026-09-11.md
docs/dev/results/ab-2026-08-27/knorm-kv-compression-longctx/baseline/raw.json
docs/dev/results/ab-2026-08-27/knorm-kv-compression-longctx/triattention/raw.json
docs/dev/results/ab-2026-08-27/kv-pressure-online/baseline/raw.json
docs/dev/results/ab-2026-08-27/kv-pressure-online/triattention/raw.json
docs/dev/results/ab-2026-08-27/prefix-repetition-online/baseline/raw.json
docs/dev/results/ab-2026-08-27/prefix-repetition-online/triattention/raw.json
docs/dev/results/ab-2026-08-27/random-online/baseline/raw.json
docs/dev/results/ab-2026-08-27/random-online/triattention/raw.json
docs/dev/results/deep-optimization-2026-08-28/knorm-kv-compression-longctx/baseline/raw.json
docs/dev/results/deep-optimization-2026-08-28/knorm-kv-compression-longctx/triattention/raw.json
docs/dev/results/deep-optimization-2026-08-28/kv-pressure-online/baseline/raw.json
docs/dev/results/deep-optimization-2026-08-28/kv-pressure-online/triattention/raw.json
docs/dev/results/deep-optimization-2026-08-28/prefix-repetition-online/baseline/raw.json
docs/dev/results/deep-optimization-2026-08-28/prefix-repetition-online/triattention/raw.json
docs/dev/results/deep-optimization-2026-08-28/quality/baseline.json
docs/dev/results/deep-optimization-2026-08-28/quality/triattention.json
docs/dev/results/deep-optimization-2026-08-28/random-online/baseline/raw.json
docs/dev/results/deep-optimization-2026-08-28/random-online/triattention/raw.json
docs/dev/results/kv-compress-v3.8-2026-08-28/summary.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-1/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-1/quality.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-2/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-2/quality.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-3/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c0/repeat-3/quality.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-1/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-1/quality.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-2/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-2/quality.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-3/pressure.json
docs/dev/results/kv-compress-v3.8-2026-08-28/c1/repeat-3/quality.json
docs/evidence/kvcompress-v0.4.0-a2-16k-c4-summary.json
docs/template/weekly-report-template.md
examples/triattention.json
scripts/kvcompress_acceptance_compare.py
scripts/kvcompress_long_context_run.py
scripts/kvcompress_m3_analyze.py
scripts/kvcompress_m3_quality.py
scripts/kvcompress_prepare_dataset.py
src/vllm_ascend_kvcompress/__init__.py
src/vllm_ascend_kvcompress/config.py
src/vllm_ascend_kvcompress/model.py
src/vllm_ascend_kvcompress/plugin.py
src/vllm_ascend_kvcompress/provider.py
src/vllm_ascend_kvcompress/stateful.py
src/vllm_ascend_kvcompress/manifests/__init__.py
src/vllm_ascend_kvcompress/manifests/vllm-hust-extension-v0.2.json
src/vllm_ascend_kvcompress/methods/__init__.py
src/vllm_ascend_kvcompress/methods/base.py
src/vllm_ascend_kvcompress/methods/registry.py
src/vllm_ascend_kvcompress/methods/triattention/__init__.py
src/vllm_ascend_kvcompress/methods/triattention/cache.py
src/vllm_ascend_kvcompress/methods/triattention/config.py
src/vllm_ascend_kvcompress/methods/triattention/kernels.py
src/vllm_ascend_kvcompress/methods/triattention/method.py
src/vllm_ascend_kvcompress/methods/triattention/scoring.py
src/vllm_ascend_kvcompress/methods/triattention/stats.py
src/vllm_ascend_kvcompress_hust.egg-info/PKG-INFO
src/vllm_ascend_kvcompress_hust.egg-info/SOURCES.txt
src/vllm_ascend_kvcompress_hust.egg-info/dependency_links.txt
src/vllm_ascend_kvcompress_hust.egg-info/entry_points.txt
src/vllm_ascend_kvcompress_hust.egg-info/requires.txt
src/vllm_ascend_kvcompress_hust.egg-info/top_level.txt
tests/run_npu_kernel_benchmark.py
tests/run_npu_kernel_smoke.py
tests/test_cache.py
tests/test_config.py
tests/test_current_host_contract.py
tests/test_manifest.py
tests/test_method_registry.py
tests/test_package_layout.py
tests/test_plugin.py
tests/test_provider_integration.py
tests/test_scoring.py
tests/test_selection.py
tests/test_stateful.py
tests/test_stats.py