dl-backtrace 0.1.7__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (237) hide show
  1. dl_backtrace-0.1.7/.github/scripts/release.py +61 -0
  2. dl_backtrace-0.1.7/.github/workflows/publish.yml +94 -0
  3. dl_backtrace-0.1.7/.github/workflows/release.yml +30 -0
  4. dl_backtrace-0.1.7/.gitignore +30 -0
  5. dl_backtrace-0.1.7/LICENSE.md +91 -0
  6. dl_backtrace-0.1.7/MANIFEST.in +8 -0
  7. dl_backtrace-0.1.7/PKG-INFO +544 -0
  8. dl_backtrace-0.1.7/README.md +417 -0
  9. dl_backtrace-0.1.7/assets/images/AryaXAI-light.jpg +0 -0
  10. dl_backtrace-0.1.7/assets/images/AryaXAI.png +0 -0
  11. dl_backtrace-0.1.7/assets/images/arya-fav.png +0 -0
  12. dl_backtrace-0.1.7/assets/images/arya-logo-mini.png +0 -0
  13. dl_backtrace-0.1.7/assets/images/dlb_logo.png +0 -0
  14. dl_backtrace-0.1.7/assets/images/lexsi_logo.svg +12 -0
  15. dl_backtrace-0.1.7/assets/images/speedup_comparison.png +0 -0
  16. dl_backtrace-0.1.7/assets/images/throughput_comparison.png +0 -0
  17. dl_backtrace-0.1.7/assets/images/total_dlb_time_comparison.png +0 -0
  18. dl_backtrace-0.1.7/dl_backtrace/__init__.py +6 -0
  19. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/__init__.py +1 -0
  20. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/__init__.py +4 -0
  21. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/backtrace.py +825 -0
  22. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/config.py +41 -0
  23. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/gpt_oss.py +385 -0
  24. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/helper.py +247 -0
  25. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/jetmoe.py +459 -0
  26. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/model_utils.py +155 -0
  27. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/moe_auto_sampler.py +821 -0
  28. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/olmoe.py +393 -0
  29. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/core/qwen3_moe.py +401 -0
  30. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/__init__.py +1 -0
  31. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/GPT_oss/helper_cuda/lm_head.py +162 -0
  32. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/GPT_oss/original_version.py +803 -0
  33. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/GPT_oss/pytorch_version.py +851 -0
  34. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/GPT_oss/refactored_version.py +757 -0
  35. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/JetMoE/original_version.py +698 -0
  36. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/JetMoE/pytorch_version.py +826 -0
  37. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/JetMoE/refactored_version.py +580 -0
  38. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/MoE_utils/nonneg_conserve.py +196 -0
  39. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/MoE_utils/relevance_gated_proj.py +245 -0
  40. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/MoE_utils/relevance_proj.py +212 -0
  41. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/MoE_utils/relevance_single.py +168 -0
  42. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/MoE_utils/wt_router_logits.py +194 -0
  43. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/OLMoE/original_version.py +401 -0
  44. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/OLMoE/pytorch_version.py +577 -0
  45. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/OLMoE/refactored_version.py +518 -0
  46. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/Qwen_MoE/helper_cuda/lm_head.py +157 -0
  47. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/Qwen_MoE/original_version.py +636 -0
  48. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/Qwen_MoE/pytorch_version.py +860 -0
  49. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/Qwen_MoE/refactored_version.py +674 -0
  50. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/cuda_utils/__init__.py +0 -0
  51. dl_backtrace-0.1.7/dl_backtrace/moe_pytorch_backtrace/backtrace/utils/default_v2.py +1066 -0
  52. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/__init__.py +1 -0
  53. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/__init__.py +12 -0
  54. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/activation.py +47 -0
  55. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/aten_operations.py +210 -0
  56. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/__init__.py +0 -0
  57. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/config.py +12 -0
  58. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/dlb_auto_sampler.py +1002 -0
  59. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/execution_engine_noncache.py +4232 -0
  60. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/graph_builder.py +388 -0
  61. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/io_utils.py +240 -0
  62. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/relevance_propagation.py +1201 -0
  63. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/reproducibility.py +287 -0
  64. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/token_relevance_visuals.py +211 -0
  65. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/trace_utils.py +124 -0
  66. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/visualization.py +453 -0
  67. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/core/visualization_module_aware.py +117 -0
  68. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/dlbacktrace.py +1305 -0
  69. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/__init__.py +1 -0
  70. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/__init__.py +0 -0
  71. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/__init__.py +0 -0
  72. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/calculate_gavgpool2d_kernel.cu +143 -0
  73. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/calculate_gavgpool2d_op.cpp +21 -0
  74. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/include/__init__.py +0 -0
  75. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/include/utils.h +12 -0
  76. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/setup.py +29 -0
  77. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/cuda_version/wt_gavgpool_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  78. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/original_version.py +36 -0
  79. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/pytorch_version.py +74 -0
  80. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AdaptiveAvgPool2D/refactored_version.py +76 -0
  81. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AveragePool2D/cuda_padding.py +240 -0
  82. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AveragePool2D/cuda_version.py +80 -0
  83. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AveragePool2D/cuda_wt_avg_unit.py +205 -0
  84. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/AveragePool2D/original_version.py +78 -0
  85. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/__init__.py +0 -0
  86. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_padding.py +240 -0
  87. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/__init__.py +0 -0
  88. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/calculate_wt_conv2d.cpp +43 -0
  89. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/conv2d_kernel.cu +203 -0
  90. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/include/__init__.py +0 -0
  91. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/include/utils.h +22 -0
  92. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/setup.py +31 -0
  93. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_version/wt_conv_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  94. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/cuda_wt_conv_unit.py +390 -0
  95. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/launch_cuda_version.py +145 -0
  96. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/original_version.py +160 -0
  97. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/pytorch_version.py +375 -0
  98. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Conv2D/refactored_version.py +299 -0
  99. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/ConvUnit2D/__init__.py +0 -0
  100. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/ConvUnit2D/original_version.py +102 -0
  101. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/ConvUnit2D/pytorch_version.py +111 -0
  102. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/__init__.py +0 -0
  103. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_v2.py +285 -0
  104. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/__init__.py +0 -0
  105. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/calculate_wt_embedding_kernel.cu +216 -0
  106. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/calculate_wt_embedding_op.cpp +17 -0
  107. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/include/__init__.py +0 -0
  108. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/include/utils.h +14 -0
  109. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/setup.py +29 -0
  110. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/cuda_version/wt_embedding_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  111. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/original_version.py +32 -0
  112. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/pytorch_version.py +65 -0
  113. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Embedded/refactored_version.py +57 -0
  114. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/__init__.py +0 -0
  115. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/__init__.py +0 -0
  116. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/calculate_wt_fc_kernel.cu +245 -0
  117. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/calculate_wt_fc_op.cpp +96 -0
  118. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/include/__init__.py +0 -0
  119. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/include/utils.h +22 -0
  120. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/setup.py +29 -0
  121. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/cuda_version/wt_fc_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  122. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/original_version.py +61 -0
  123. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/pytorch_version.py +158 -0
  124. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear/refactored_version.py +188 -0
  125. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/__init__.py +0 -0
  126. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/__init__.py +0 -0
  127. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/cal_wt_fc_kernel.cu +202 -0
  128. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/cal_wt_fc_ops.cpp +85 -0
  129. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/include/__init__.py +0 -0
  130. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/include/utils.h +42 -0
  131. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/setup.py +29 -0
  132. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_layer/wt_fc_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  133. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_v2.py +353 -0
  134. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/cuda_version.py +68 -0
  135. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/original_version.py +100 -0
  136. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v2/pytorch_version.py +154 -0
  137. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v3/__init__.py +0 -0
  138. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Linear_v3/cuda_v3.py +414 -0
  139. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/MaxPool2D/__init__.py +0 -0
  140. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/MaxPool2D/cuda_version.py +142 -0
  141. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/MaxPool2D/original_version.py +31 -0
  142. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/MaxPool2D/pytorch_version.py +98 -0
  143. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/MaxPool2D/refactored_version.py +87 -0
  144. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Padding/__init__.py +0 -0
  145. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Padding/original.py +34 -0
  146. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Padding/pytorch.py +97 -0
  147. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/__init__.py +0 -0
  148. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_v2.py +508 -0
  149. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_v3.py +451 -0
  150. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/__init__.py +0 -0
  151. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/calculate_wt_selfattention_kernel.cu +260 -0
  152. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/calculate_wt_selfattention_op.cpp +81 -0
  153. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/include/__init__.py +0 -0
  154. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/include/utils.h +28 -0
  155. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/setup.py +29 -0
  156. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/cuda_version/wt_selfattention_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  157. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/original_version.py +120 -0
  158. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/pytorch_v2.py +150 -0
  159. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/SelfAttention/pytorch_version.py +109 -0
  160. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/__init__.py +0 -0
  161. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/__init__.py +0 -0
  162. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/calculate_wt_maxunit2d_kernel.cu +124 -0
  163. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/calculate_wt_maxunit2d_op.cpp +24 -0
  164. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/include/__init__.py +0 -0
  165. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/include/utils.h +14 -0
  166. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/setup.py +26 -0
  167. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/cuda_version/wt_maxunit2d_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  168. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/original_version.py +10 -0
  169. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/pytorch_version.py +45 -0
  170. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/WtMaxunit2D/refactored_version.py +56 -0
  171. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/__init__.py +0 -0
  172. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/__init__.py +0 -0
  173. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/calc_wt_add_eq_kernel.cu +109 -0
  174. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/calc_wt_add_eq_op.cpp +106 -0
  175. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/include/__init__.py +0 -0
  176. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/include/utils.h +28 -0
  177. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/setup.py +29 -0
  178. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/cuda_version/wt_add_equal_ops.cpython-310-x86_64-linux-gnu.so +0 -0
  179. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/original_version.py +31 -0
  180. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/pytorch_version.py +53 -0
  181. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_add_equal/refactored_version.py +54 -0
  182. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_mul/__init__.py +0 -0
  183. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_mul/cuda_v1.py +173 -0
  184. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_mul/original_version.py +21 -0
  185. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_mul/pytorch_version.py +29 -0
  186. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/Wt_mul/refactored_version.py +31 -0
  187. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/cuda_utils/__init__.py +0 -0
  188. dl_backtrace-0.1.7/dl_backtrace/pytorch_backtrace/dlbacktrace/utils/default_v2.py +384 -0
  189. dl_backtrace-0.1.7/dl_backtrace/version.py +34 -0
  190. dl_backtrace-0.1.7/dl_backtrace.egg-info/PKG-INFO +544 -0
  191. dl_backtrace-0.1.7/dl_backtrace.egg-info/SOURCES.txt +235 -0
  192. dl_backtrace-0.1.7/dl_backtrace.egg-info/dependency_links.txt +1 -0
  193. dl_backtrace-0.1.7/dl_backtrace.egg-info/requires.txt +17 -0
  194. dl_backtrace-0.1.7/dl_backtrace.egg-info/top_level.txt +4 -0
  195. dl_backtrace-0.1.7/docs/BUILD_DOCS.md +259 -0
  196. dl_backtrace-0.1.7/docs/assets/css/custom.css +145 -0
  197. dl_backtrace-0.1.7/docs/assets/images/AryaXAI-light.jpg +0 -0
  198. dl_backtrace-0.1.7/docs/assets/images/AryaXAI.png +0 -0
  199. dl_backtrace-0.1.7/docs/assets/images/arya-fav.png +0 -0
  200. dl_backtrace-0.1.7/docs/assets/images/arya-logo-mini.png +0 -0
  201. dl_backtrace-0.1.7/docs/assets/images/dlb_logo.png +0 -0
  202. dl_backtrace-0.1.7/docs/assets/images/lexsi_logo.svg +12 -0
  203. dl_backtrace-0.1.7/docs/developer/architecture.md +65 -0
  204. dl_backtrace-0.1.7/docs/developer/contributing.md +463 -0
  205. dl_backtrace-0.1.7/docs/developer/cuda-development.md +400 -0
  206. dl_backtrace-0.1.7/docs/developer/execution-engine.md +631 -0
  207. dl_backtrace-0.1.7/docs/developer/testing.md +93 -0
  208. dl_backtrace-0.1.7/docs/examples/colab-notebooks.md +262 -0
  209. dl_backtrace-0.1.7/docs/examples/use-cases.md +63 -0
  210. dl_backtrace-0.1.7/docs/guide/best-practices.md +132 -0
  211. dl_backtrace-0.1.7/docs/guide/introduction.md +424 -0
  212. dl_backtrace-0.1.7/docs/guide/pytorch/auto-sampler.md +550 -0
  213. dl_backtrace-0.1.7/docs/guide/pytorch/dlbacktrace.md +207 -0
  214. dl_backtrace-0.1.7/docs/guide/pytorch/execution-engines.md +302 -0
  215. dl_backtrace-0.1.7/docs/guide/pytorch/moe-models.md +555 -0
  216. dl_backtrace-0.1.7/docs/guide/pytorch/operations.md +273 -0
  217. dl_backtrace-0.1.7/docs/guide/pytorch/overview.md +92 -0
  218. dl_backtrace-0.1.7/docs/guide/pytorch/pipeline.md +751 -0
  219. dl_backtrace-0.1.7/docs/guide/pytorch/temperature-scaling.md +565 -0
  220. dl_backtrace-0.1.7/docs/guide/pytorch/tracing.md +300 -0
  221. dl_backtrace-0.1.7/docs/guide/relevance/overview.md +309 -0
  222. dl_backtrace-0.1.7/docs/guide/relevance/parameters.md +57 -0
  223. dl_backtrace-0.1.7/docs/home/features.md +409 -0
  224. dl_backtrace-0.1.7/docs/home/installation.md +229 -0
  225. dl_backtrace-0.1.7/docs/home/quickstart.md +704 -0
  226. dl_backtrace-0.1.7/docs/home/whats-new.md +256 -0
  227. dl_backtrace-0.1.7/docs/index.md +161 -0
  228. dl_backtrace-0.1.7/docs/javascripts/mathjax.js +19 -0
  229. dl_backtrace-0.1.7/docs/support/changelog.md +328 -0
  230. dl_backtrace-0.1.7/docs/support/faq.md +313 -0
  231. dl_backtrace-0.1.7/docs/support/known-issues.md +72 -0
  232. dl_backtrace-0.1.7/docs/support/license.md +221 -0
  233. dl_backtrace-0.1.7/docs-requirements.txt +8 -0
  234. dl_backtrace-0.1.7/mkdocs.yml +113 -0
  235. dl_backtrace-0.1.7/pyproject.toml +55 -0
  236. dl_backtrace-0.1.7/requirements.txt +19 -0
  237. dl_backtrace-0.1.7/setup.cfg +4 -0
@@ -0,0 +1,61 @@
1
+ #!/usr/bin/env python3
2
+ """Create the next patch release on GitHub.
3
+
4
+ Tags follow semantic versioning without a leading "v" (e.g. 0.1.1). Dev
5
+ pre-releases (".devN") are ignored when computing the next patch.
6
+ """
7
+
8
+ from subprocess import PIPE, CalledProcessError, run
9
+ from typing import List
10
+
11
+ INITIAL_VERSION = "0.1.1"
12
+
13
+
14
+ def _run(cmd: List[str]):
15
+ return run(cmd, check=False, stdout=PIPE, stderr=PIPE, text=True)
16
+
17
+
18
+ def get_last_version() -> str:
19
+ """Return the most recent non-dev tag, if any."""
20
+ tag_result = _run(["git", "tag", "--list", "*.*.*", "--sort=-v:refname"])
21
+ if tag_result.returncode == 0:
22
+ for tag in tag_result.stdout.splitlines():
23
+ tag = tag.strip()
24
+ if tag and ".dev" not in tag:
25
+ return tag
26
+
27
+ release_result = _run(["gh", "release", "list", "--limit", "20"])
28
+ if release_result.returncode == 0:
29
+ for line in release_result.stdout.splitlines():
30
+ fields = line.split("\t")
31
+ tag = fields[2] if len(fields) > 2 else ""
32
+ if tag and ".dev" not in tag:
33
+ return tag
34
+
35
+ return ""
36
+
37
+
38
+ def bump_patch_number(version_number: str) -> str:
39
+ major, minor, patch = version_number.split(".")
40
+ return f"{major}.{minor}.{int(patch) + 1}"
41
+
42
+
43
+ def create_new_patch_release():
44
+ last_version_number = get_last_version()
45
+ new_version_number = (
46
+ bump_patch_number(last_version_number) if last_version_number else INITIAL_VERSION
47
+ )
48
+
49
+ print(f"Creating release {new_version_number}")
50
+ try:
51
+ run(
52
+ ["gh", "release", "create", new_version_number, "--generate-notes", "--latest"],
53
+ check=True,
54
+ )
55
+ except CalledProcessError as err:
56
+ print(err.stderr)
57
+ raise
58
+
59
+
60
+ if __name__ == "__main__":
61
+ create_new_patch_release()
@@ -0,0 +1,94 @@
1
+ name: Publish to PyPI
2
+
3
+ on:
4
+ workflow_dispatch:
5
+ push:
6
+ tags:
7
+ - '*.*.*'
8
+ - '!*.*.dev*'
9
+
10
+ jobs:
11
+ pypi:
12
+ runs-on: ubuntu-latest
13
+ permissions:
14
+ contents: write
15
+ steps:
16
+ - name: Checkout
17
+ uses: actions/checkout@v3
18
+ with:
19
+ fetch-depth: 0
20
+
21
+ - name: Fetch tags
22
+ run: git fetch --prune --tags
23
+
24
+ # (optional but recommended) Pin Python version
25
+ - name: Set up Python
26
+ uses: actions/setup-python@v5
27
+ with:
28
+ python-version: '3.11'
29
+
30
+ # 🔧 NEW: make sure packaging / build stack are up to date
31
+ - name: Install build tooling
32
+ run: |
33
+ python -m pip install --upgrade pip
34
+ python -m pip install --upgrade setuptools wheel
35
+ python -m pip install --upgrade build twine packaging
36
+
37
+ # (optional) clean old build artefacts, just in case
38
+ - name: Clean build artefacts
39
+ run: rm -rf dist build ./*.egg-info
40
+
41
+ - name: Get Release Notes
42
+ id: release_notes
43
+ run: |
44
+ TAG_NAME=$(gh release view --json tagName | jq -r .tagName)
45
+ RELEASE_NOTES=$(gh release view --json url | jq -r .url)
46
+ RELEASE_NOTES="${RELEASE_NOTES//$'\n'/\\n}"
47
+ echo "TAG_NAME=${TAG_NAME}" >> $GITHUB_OUTPUT
48
+ echo "RELEASE_NOTES=${RELEASE_NOTES}" >> $GITHUB_OUTPUT
49
+ env:
50
+ GH_TOKEN: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
51
+
52
+ - name: Build package
53
+ run: python -m build
54
+
55
+ - name: Publish package to PyPI
56
+ run: python -m twine upload dist/*
57
+ env:
58
+ TWINE_USERNAME: __token__
59
+ TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
60
+
61
+ - name: Send release notes to Slack
62
+ id: slack
63
+ uses: slackapi/slack-github-action@v1.24.0
64
+ with:
65
+ channel-id: 'aryaxai-dl-backtrace-release'
66
+ payload: |
67
+ {
68
+ "text": "AryaXAI SDK release",
69
+ "blocks": [
70
+ {
71
+ "type": "section",
72
+ "text": {
73
+ "type": "mrkdwn",
74
+ "text": "*Version:* ${{ steps.release_notes.outputs.TAG_NAME }}"
75
+ }
76
+ },
77
+ {
78
+ "type": "section",
79
+ "text": {
80
+ "type": "mrkdwn",
81
+ "text": "*Release Notes:* ${{ steps.release_notes.outputs.RELEASE_NOTES }}"
82
+ }
83
+ },
84
+ {
85
+ "type": "section",
86
+ "text": {
87
+ "type": "mrkdwn",
88
+ "text": "*PyPI URL:* https://pypi.org/project/dl-backtrace/${{ steps.release_notes.outputs.TAG_NAME }}"
89
+ }
90
+ }
91
+ ]
92
+ }
93
+ env:
94
+ SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
@@ -0,0 +1,30 @@
1
+ name: Create release
2
+
3
+ on:
4
+ workflow_dispatch:
5
+ push:
6
+ branches:
7
+ - dlb_v2
8
+
9
+ jobs:
10
+ github:
11
+ runs-on: ubuntu-latest
12
+ permissions:
13
+ contents: write
14
+ steps:
15
+ - name: Checkout
16
+ uses: actions/checkout@v4
17
+ with:
18
+ fetch-depth: 0
19
+
20
+ - name: Set up Python
21
+ uses: actions/setup-python@v5
22
+ with:
23
+ python-version: '3.11'
24
+
25
+ - name: Ensure tags are available
26
+ run: git fetch --prune --tags
27
+ - name: Create new release
28
+ run: .github/scripts/release.py
29
+ env:
30
+ GH_TOKEN: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
@@ -0,0 +1,30 @@
1
+ .vscode
2
+
3
+ # Compiled python modules.
4
+ *.pyc
5
+ __pycache__
6
+
7
+ # Setuptools distribution folder.
8
+ /dist/
9
+ build
10
+
11
+ # Python egg metadata, regenerated from source files by setuptools.
12
+ *.egg-info
13
+ .eggs
14
+
15
+ # Documentation files
16
+ _build
17
+ _static
18
+ _templates
19
+
20
+ # Venv
21
+ *venv
22
+
23
+ # package version file
24
+ version.py
25
+ benchmarks/debug_experiments/*
26
+ dl_backtrace_logs/*
27
+
28
+ site
29
+
30
+ .github
@@ -0,0 +1,91 @@
1
+ # Lexsi Labs Source Available License (LSAL) – Version 1.0
2
+
3
+ ## Preamble
4
+
5
+ This Source Available License governs use of the software known as **Deep Learning Backtrace (DLB or DLBactrace)** (the “Licensed Work”), developed and owned by **Lexsi Labs (Lithasa Technologies Pvt. Ltd.)** (“Licensor”).
6
+
7
+ This is **not** an open-source license as defined by the [Open Source Initiative (OSI)](https://opensource.org/). However, it allows broad access to the source code for **evaluation, learning, and research purposes**.
8
+
9
+ ---
10
+
11
+ ## 1. Grant of Rights
12
+
13
+ The Licensor grants you (the “Licensee”) a **non-exclusive, worldwide, non-transferable, royalty-free license** to:
14
+
15
+ * Use, copy, modify, and create derivative works of the Licensed Work for **personal, academic, or research purposes** only.
16
+ * Redistribute unmodified or modified versions of the Licensed Work for **non-commercial purposes** only, provided this License text is **included in full and clearly displayed**.
17
+
18
+ ---
19
+
20
+ ## 2. Restrictions
21
+
22
+ You may **not**, without a separate **commercial license** from Lexsi Labs:
23
+
24
+ * Use the Licensed Work, or any derivative work, for **any commercial purpose**, including but not limited to:
25
+
26
+ * Offering it as part of a commercial product or service (including SaaS, hosted, or API offerings);
27
+ * Embedding it in proprietary software;
28
+ * Using it to generate revenue, provide paid consulting, or deliver commercial services.
29
+
30
+ * **Re-license**, **rebrand**, or **redistribute** the Licensed Work under different terms.
31
+
32
+ * Use any **Lexsi Labs**, **DLB**, or related **trademarks, logos, or branding**, except to identify unmodified, licensed copies.
33
+
34
+ ---
35
+
36
+ ## 3. Ownership
37
+
38
+ All **rights, title, and interest** in and to the Licensed Work remain with **Lithasa Technologies Pvt. Ltd.**
39
+ Nothing in this License transfers ownership or any implied patent rights to the Licensee.
40
+
41
+ ---
42
+
43
+ ## 4. Contributions
44
+
45
+ If you submit modifications, pull requests, or patches (“Contributions”) to Lexsi Labs:
46
+
47
+ * You grant Lexsi Labs a **perpetual, worldwide, royalty-free right** to use, modify, distribute, and license your Contributions under any terms, including commercial ones.
48
+ * You represent that you have the **right to make such contributions**.
49
+
50
+ ---
51
+
52
+ ## 5. Warranty Disclaimer
53
+
54
+ The Licensed Work is provided **“as is”**, without warranty of any kind, express or implied, including but not limited to the warranties of **merchantability**, **fitness for a particular purpose**, and **non-infringement**.
55
+
56
+ ---
57
+
58
+ ## 6. Liability Limitation
59
+
60
+ In no event shall **Lexsi Labs** or its contributors be liable for any damages arising from the use of the Licensed Work, whether in contract, tort, or otherwise, even if advised of the possibility of such damages.
61
+
62
+ ---
63
+
64
+ ## 7. Termination
65
+
66
+ This License terminates automatically if you breach any of its terms.
67
+ Upon termination, you must immediately cease use and destroy all copies of the Licensed Work in your possession.
68
+
69
+ ---
70
+
71
+ ## 8. Governing Law
72
+
73
+ This License shall be governed by and construed in accordance with the **laws of India**, without regard to its conflict of law principles.
74
+
75
+ ---
76
+
77
+ ## 9. Contact for Commercial Licensing
78
+
79
+ For **commercial use, partnership, or redistribution rights**, contact:
80
+ 📧 **[support@lexsi.ai](mailto:support@lexsi.ai)**
81
+ 🌐 **[https://lexsi.ai](https://lexsi.ai)**
82
+
83
+ ---
84
+
85
+ ## 10. Notice
86
+
87
+ **Deep Learning Backtrace (DLB)** © 2025 **Lithasa Technologies Pvt. Ltd.**
88
+ Licensed under the **Lexsi Labs Source Available License (LSAL) v1.0**.
89
+ **Not for commercial use without explicit permission.**
90
+
91
+ ---
@@ -0,0 +1,8 @@
1
+ include requirements.txt
2
+ include LICENSE.md
3
+ recursive-include dl_backtrace *.pickle
4
+ recursive-exclude * __pycache__
5
+ recursive-exclude * *.py[co]
6
+ recursive-include docs *.md
7
+ recursive-include tests *.py
8
+ include README.md