MANIFEST.in
README.md
pyproject.toml
setup.py
src/isagellm_backend.egg-info/PKG-INFO
src/isagellm_backend.egg-info/SOURCES.txt
src/isagellm_backend.egg-info/dependency_links.txt
src/isagellm_backend.egg-info/entry_points.txt
src/isagellm_backend.egg-info/requires.txt
src/isagellm_backend.egg-info/top_level.txt
src/sagellm_backend/__init__.py
src/sagellm_backend/__init__.pyc
src/sagellm_backend/_version.py
src/sagellm_backend/_version.pyc
src/sagellm_backend/base.pyc
src/sagellm_backend/benchmarking.pyc
src/sagellm_backend/dtype_conversion.pyc
src/sagellm_backend/feature_variants.pyc
src/sagellm_backend/pipeline.pyc
src/sagellm_backend/py.typed
src/sagellm_backend/registry.pyc
src/sagellm_backend/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/__pycache__/__init__.cpython-313.pyc
src/sagellm_backend/__pycache__/_version.cpython-311.pyc
src/sagellm_backend/__pycache__/base.cpython-310.pyc
src/sagellm_backend/__pycache__/base.cpython-311.pyc
src/sagellm_backend/__pycache__/benchmarking.cpython-311.pyc
src/sagellm_backend/__pycache__/capability.cpython-311.pyc
src/sagellm_backend/__pycache__/dtype_conversion.cpython-311.pyc
src/sagellm_backend/__pycache__/feature_variants.cpython-311.pyc
src/sagellm_backend/__pycache__/pipeline.cpython-311.pyc
src/sagellm_backend/__pycache__/registry.cpython-310.pyc
src/sagellm_backend/__pycache__/registry.cpython-311.pyc
src/sagellm_backend/attention/__init__.py
src/sagellm_backend/attention/__init__.pyc
src/sagellm_backend/attention/ascend.pyc
src/sagellm_backend/attention/base.pyc
src/sagellm_backend/attention/cpu.pyc
src/sagellm_backend/attention/flash_attention.pyc
src/sagellm_backend/attention/kunlun.pyc
src/sagellm_backend/attention/mthreads.pyc
src/sagellm_backend/attention/paged_attention.pyc
src/sagellm_backend/attention/registry.pyc
src/sagellm_backend/attention/rocm.pyc
src/sagellm_backend/attention/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/__init__.cpython-313.pyc
src/sagellm_backend/attention/__pycache__/ascend.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/base.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/base.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/base.cpython-313.pyc
src/sagellm_backend/attention/__pycache__/cpu.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/cpu.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/flash_attention.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/flash_attention.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/kunlun.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/mthreads.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/paged_attention.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/paged_attention.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/registry.cpython-310.pyc
src/sagellm_backend/attention/__pycache__/registry.cpython-311.pyc
src/sagellm_backend/attention/__pycache__/rocm.cpython-311.pyc
src/sagellm_backend/graph/__init__.py
src/sagellm_backend/graph/__init__.pyc
src/sagellm_backend/graph/config.pyc
src/sagellm_backend/graph/graph.pyc
src/sagellm_backend/graph/optimizer.pyc
src/sagellm_backend/graph/pass_base.pyc
src/sagellm_backend/graph/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/graph/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/graph/__pycache__/config.cpython-310.pyc
src/sagellm_backend/graph/__pycache__/config.cpython-311.pyc
src/sagellm_backend/graph/__pycache__/graph.cpython-310.pyc
src/sagellm_backend/graph/__pycache__/graph.cpython-311.pyc
src/sagellm_backend/graph/__pycache__/optimizer.cpython-310.pyc
src/sagellm_backend/graph/__pycache__/optimizer.cpython-311.pyc
src/sagellm_backend/graph/__pycache__/pass_base.cpython-310.pyc
src/sagellm_backend/graph/__pycache__/pass_base.cpython-311.pyc
src/sagellm_backend/graph/passes/__init__.py
src/sagellm_backend/graph/passes/__init__.pyc
src/sagellm_backend/graph/passes/constant_folding.pyc
src/sagellm_backend/graph/passes/cuda_hardware_optimization.pyc
src/sagellm_backend/graph/passes/data_format_selection.pyc
src/sagellm_backend/graph/passes/dead_node_elimination.pyc
src/sagellm_backend/graph/passes/identity_elimination.pyc
src/sagellm_backend/graph/passes/layout_conversion.pyc
src/sagellm_backend/graph/passes/memory_reuse_analysis.pyc
src/sagellm_backend/graph/passes/operator_fusion.pyc
src/sagellm_backend/graph/passes/static_precompute.pyc
src/sagellm_backend/graph/passes/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/constant_folding.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/constant_folding.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/cuda_hardware_optimization.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/cuda_hardware_optimization.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/data_format_selection.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/data_format_selection.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/dead_node_elimination.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/dead_node_elimination.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/identity_elimination.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/identity_elimination.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/layout_conversion.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/layout_conversion.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/memory_reuse_analysis.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/memory_reuse_analysis.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/operator_fusion.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/operator_fusion.cpython-311.pyc
src/sagellm_backend/graph/passes/__pycache__/static_precompute.cpython-310.pyc
src/sagellm_backend/graph/passes/__pycache__/static_precompute.cpython-311.pyc
src/sagellm_backend/hardware/__init__.py
src/sagellm_backend/hardware/__init__.pyc
src/sagellm_backend/hardware/base.pyc
src/sagellm_backend/hardware/cache.pyc
src/sagellm_backend/hardware/profile.pyc
src/sagellm_backend/hardware/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/hardware/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/hardware/__pycache__/base.cpython-310.pyc
src/sagellm_backend/hardware/__pycache__/base.cpython-311.pyc
src/sagellm_backend/hardware/__pycache__/cache.cpython-310.pyc
src/sagellm_backend/hardware/__pycache__/cache.cpython-311.pyc
src/sagellm_backend/hardware/__pycache__/profile.cpython-310.pyc
src/sagellm_backend/hardware/__pycache__/profile.cpython-311.pyc
src/sagellm_backend/hardware/__pycache__/sm90.cpython-310.pyc
src/sagellm_backend/hardware/inspectors/__init__.py
src/sagellm_backend/hardware/inspectors/__init__.pyc
src/sagellm_backend/hardware/inspectors/ascend.pyc
src/sagellm_backend/hardware/inspectors/cpu.pyc
src/sagellm_backend/hardware/inspectors/cuda.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/ascend.cpython-310.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/ascend.cpython-311.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/cpu.cpython-310.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/cpu.cpython-311.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/cuda.cpython-310.pyc
src/sagellm_backend/hardware/inspectors/__pycache__/cuda.cpython-311.pyc
src/sagellm_backend/kernels/__init__.py
src/sagellm_backend/kernels/__init__.pyc
src/sagellm_backend/kernels/activation.pyc
src/sagellm_backend/kernels/base.pyc
src/sagellm_backend/kernels/cuda_graph.pyc
src/sagellm_backend/kernels/embedding.pyc
src/sagellm_backend/kernels/fused_ops.py
src/sagellm_backend/kernels/fused_ops.pyc
src/sagellm_backend/kernels/linear.pyc
src/sagellm_backend/kernels/normalization.pyc
src/sagellm_backend/kernels/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/activation.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/activation.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/base.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/base.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/cuda_graph.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/cuda_graph.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/embedding.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/embedding.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/fused_ops.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/fused_ops.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/linear.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/linear.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/normalization.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/normalization.cpython-311.pyc
src/sagellm_backend/kernels/__pycache__/sm90_gmma.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/sm90_mbarrier.cpython-310.pyc
src/sagellm_backend/kernels/__pycache__/sm90_tma.cpython-310.pyc
src/sagellm_backend/kernels/ascend/__init__.py
src/sagellm_backend/kernels/ascend/__init__.pyc
src/sagellm_backend/kernels/ascend/attention.pyc
src/sagellm_backend/kernels/ascend/matmul.pyc
src/sagellm_backend/kernels/ascend/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/kernels/ascend/__pycache__/attention.cpython-311.pyc
src/sagellm_backend/kernels/ascend/__pycache__/matmul.cpython-311.pyc
src/sagellm_backend/kernels/kunlun/__init__.py
src/sagellm_backend/kernels/kunlun/__init__.pyc
src/sagellm_backend/kernels/kunlun/gemm.pyc
src/sagellm_backend/kernels/kunlun/matmul.pyc
src/sagellm_backend/kernels/kunlun/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/kernels/kunlun/__pycache__/gemm.cpython-311.pyc
src/sagellm_backend/kernels/kunlun/__pycache__/matmul.cpython-311.pyc
src/sagellm_backend/kernels/rocm/__init__.py
src/sagellm_backend/kernels/rocm/__init__.pyc
src/sagellm_backend/kernels/rocm/attention_kernels.pyc
src/sagellm_backend/kernels/rocm/gemm_kernels.pyc
src/sagellm_backend/kernels/rocm/normalization_kernels.pyc
src/sagellm_backend/kernels/rocm/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/kernels/rocm/__pycache__/attention_kernels.cpython-311.pyc
src/sagellm_backend/kernels/rocm/__pycache__/gemm_kernels.cpython-311.pyc
src/sagellm_backend/memory/__init__.py
src/sagellm_backend/memory/__init__.pyc
src/sagellm_backend/memory/allocator.pyc
src/sagellm_backend/memory/memory_handle.pyc
src/sagellm_backend/memory/memory_pool.pyc
src/sagellm_backend/memory/pin_memory.pyc
src/sagellm_backend/memory/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/memory/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/memory/__pycache__/allocator.cpython-310.pyc
src/sagellm_backend/memory/__pycache__/allocator.cpython-311.pyc
src/sagellm_backend/memory/__pycache__/memory_handle.cpython-310.pyc
src/sagellm_backend/memory/__pycache__/memory_handle.cpython-311.pyc
src/sagellm_backend/memory/__pycache__/memory_pool.cpython-310.pyc
src/sagellm_backend/memory/__pycache__/memory_pool.cpython-311.pyc
src/sagellm_backend/memory/__pycache__/pin_memory.cpython-310.pyc
src/sagellm_backend/memory/__pycache__/pin_memory.cpython-311.pyc
src/sagellm_backend/providers/__init__.py
src/sagellm_backend/providers/__init__.pyc
src/sagellm_backend/providers/ascend.pyc
src/sagellm_backend/providers/cpu.pyc
src/sagellm_backend/providers/cuda.pyc
src/sagellm_backend/providers/dcu.pyc
src/sagellm_backend/providers/kunlun.pyc
src/sagellm_backend/providers/mthreads.pyc
src/sagellm_backend/providers/__pycache__/__init__.cpython-310.pyc
src/sagellm_backend/providers/__pycache__/__init__.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/ascend.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/cpu.cpython-310.pyc
src/sagellm_backend/providers/__pycache__/cpu.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/cuda.cpython-310.pyc
src/sagellm_backend/providers/__pycache__/cuda.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/dcu.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/kunlun.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/mock.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/mthreads.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/pytorch_ascend.cpython-310.pyc
src/sagellm_backend/providers/__pycache__/pytorch_ascend.cpython-311.pyc
src/sagellm_backend/providers/__pycache__/pytorch_cuda.cpython-310.pyc
src/sagellm_backend/providers/__pycache__/pytorch_cuda.cpython-311.pyc
tests/test_attention_backends.py
tests/test_attention_registry.py
tests/test_config.py
tests/test_cpu_backend.py
tests/test_cuda_graph.py
tests/test_cuda_hardware_optimization.py
tests/test_dcu_provider_rocm.py
tests/test_dtype_conversion.py
tests/test_entry_points.py
tests/test_feature_variants.py
tests/test_feature_variants_provider.py
tests/test_fused_rmsnorm.py
tests/test_fused_silu_mul.py
tests/test_graph.py
tests/test_kernel_registry.py
tests/test_kunlun_attention.py
tests/test_kunlun_kernels.py
tests/test_kunlun_provider.py
tests/test_memory.py
tests/test_memory_layout_optimization.py
tests/test_optimizer.py
tests/test_pass_base.py
tests/test_passes.py
tests/test_pipeline.py
tests/test_precision_validation.py
tests/test_providers.py
tests/test_rocm_attention.py