liger-kernel-nightly 0.5.9.dev20250505192105__tar.gz → 0.5.9.dev20250508211521__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 (246) hide show
  1. liger_kernel_nightly-0.5.9.dev20250508211521/.idea/workspace.xml +79 -0
  2. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/PKG-INFO +31 -19
  3. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/README.md +30 -18
  4. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/pyproject.toml +1 -1
  5. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/dpo_loss.py +1 -1
  6. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel_nightly.egg-info/PKG-INFO +31 -19
  7. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel_nightly.egg-info/SOURCES.txt +1 -0
  8. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_dpo_loss.py +2 -0
  9. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/ISSUE_TEMPLATE/bug_report.yaml +0 -0
  10. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/ISSUE_TEMPLATE/feature_request.yaml +0 -0
  11. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/pull_request_template.md +0 -0
  12. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/amd-ci.yml +0 -0
  13. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/docs.yml +0 -0
  14. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/intel-ci.yml +0 -0
  15. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/nvi-ci.yml +0 -0
  16. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/publish-nightly.yml +0 -0
  17. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.github/workflows/publish-release.yml +0 -0
  18. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/.gitignore +0 -0
  19. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/LICENSE +0 -0
  20. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/Makefile +0 -0
  21. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/NOTICE +0 -0
  22. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/README.md +0 -0
  23. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/__init__.py +0 -0
  24. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/benchmarks_visualizer.py +0 -0
  25. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/data/all_benchmark_data.csv +0 -0
  26. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/__init__.py +0 -0
  27. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_cpo_loss.py +0 -0
  28. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_cross_entropy.py +0 -0
  29. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_distill_jsd_loss.py +0 -0
  30. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_dpo_loss.py +0 -0
  31. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_dyt.py +0 -0
  32. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_embedding.py +0 -0
  33. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_fused_linear_cross_entropy.py +0 -0
  34. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_fused_linear_jsd.py +0 -0
  35. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_geglu.py +0 -0
  36. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_group_norm.py +0 -0
  37. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_jsd.py +0 -0
  38. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_kl_div.py +0 -0
  39. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_kto_loss.py +0 -0
  40. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_layer_norm.py +0 -0
  41. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_orpo_loss.py +0 -0
  42. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_qwen2vl_mrope.py +0 -0
  43. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_rms_norm.py +0 -0
  44. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_rope.py +0 -0
  45. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_simpo_loss.py +0 -0
  46. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_swiglu.py +0 -0
  47. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/benchmark_tvd.py +0 -0
  48. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/benchmark/scripts/utils.py +0 -0
  49. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/dev/fmt-requirements.txt +0 -0
  50. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/dev/modal/tests.py +0 -0
  51. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/dev/modal/tests_bwd.py +0 -0
  52. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/Examples.md +0 -0
  53. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/Getting-Started.md +0 -0
  54. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/High-Level-APIs.md +0 -0
  55. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/Low-Level-APIs.md +0 -0
  56. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/acknowledgement.md +0 -0
  57. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/contributing.md +0 -0
  58. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/banner.GIF +0 -0
  59. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/compose.gif +0 -0
  60. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/e2e-memory.png +0 -0
  61. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/e2e-tps.png +0 -0
  62. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/logo-banner.png +0 -0
  63. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/patch.gif +0 -0
  64. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/images/post-training.png +0 -0
  65. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/index.md +0 -0
  66. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/docs/license.md +0 -0
  67. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/alignment/accelerate_config.yaml +0 -0
  68. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/alignment/run_orpo.py +0 -0
  69. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/README.md +0 -0
  70. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/callback.py +0 -0
  71. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/config/fsdp_config.json +0 -0
  72. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/gemma_7b_mem.png +0 -0
  73. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/gemma_7b_tp.png +0 -0
  74. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/llama_mem_alloc.png +0 -0
  75. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/llama_tps.png +0 -0
  76. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/qwen_mem_alloc.png +0 -0
  77. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/img/qwen_tps.png +0 -0
  78. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/launch_on_modal.py +0 -0
  79. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/requirements.txt +0 -0
  80. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/run_benchmarks.sh +0 -0
  81. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/run_gemma.sh +0 -0
  82. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/run_llama.sh +0 -0
  83. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/run_qwen.sh +0 -0
  84. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/run_qwen2_vl.sh +0 -0
  85. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/training.py +0 -0
  86. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/huggingface/training_multimodal.py +0 -0
  87. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/lightning/README.md +0 -0
  88. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/lightning/requirements.txt +0 -0
  89. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/lightning/training.py +0 -0
  90. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/README.md +0 -0
  91. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/callback.py +0 -0
  92. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Memory_Stage1_num_head_3.png +0 -0
  93. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Memory_Stage1_num_head_5.png +0 -0
  94. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Memory_Stage2_num_head_3.png +0 -0
  95. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Memory_Stage2_num_head_5.png +0 -0
  96. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Throughput_Stage1_num_head_3.png +0 -0
  97. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Throughput_Stage1_num_head_5.png +0 -0
  98. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Throughput_Stage2_num_head_3.png +0 -0
  99. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/docs/images/Throughput_Stage2_num_head_5.png +0 -0
  100. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/fsdp/acc-fsdp.conf +0 -0
  101. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/medusa_util.py +0 -0
  102. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/requirements.txt +0 -0
  103. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/scripts/llama3_8b_medusa.sh +0 -0
  104. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/examples/medusa/train.py +0 -0
  105. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/licenses/LICENSE-Apache-2.0 +0 -0
  106. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/licenses/LICENSE-MIT-AutoAWQ +0 -0
  107. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/licenses/LICENSE-MIT-Efficient-Cross-Entropy +0 -0
  108. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/licenses/LICENSE-MIT-llmc +0 -0
  109. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/licenses/LICENSE-MIT-triton +0 -0
  110. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/mkdocs.yml +0 -0
  111. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/setup.cfg +0 -0
  112. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/setup.py +0 -0
  113. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/__init__.py +0 -0
  114. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/README.md +0 -0
  115. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/__init__.py +0 -0
  116. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/cpo_loss.py +0 -0
  117. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/functional.py +0 -0
  118. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/fused_linear_distillation.py +0 -0
  119. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/fused_linear_ppo.py +0 -0
  120. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/fused_linear_preference.py +0 -0
  121. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/fused_linear_unpaired_preference.py +0 -0
  122. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/grpo_loss.py +0 -0
  123. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/jsd_loss.py +0 -0
  124. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/kto_loss.py +0 -0
  125. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/orpo_loss.py +0 -0
  126. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/chunked_loss/simpo_loss.py +0 -0
  127. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/env_report.py +0 -0
  128. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/__init__.py +0 -0
  129. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/cross_entropy.py +0 -0
  130. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/dyt.py +0 -0
  131. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/experimental/embedding.py +0 -0
  132. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/experimental/mm_int8int2.py +0 -0
  133. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/fused_linear_cross_entropy.py +0 -0
  134. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/fused_linear_jsd.py +0 -0
  135. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/geglu.py +0 -0
  136. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/group_norm.py +0 -0
  137. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/jsd.py +0 -0
  138. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/kl_div.py +0 -0
  139. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/layer_norm.py +0 -0
  140. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/qwen2vl_mrope.py +0 -0
  141. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/rms_norm.py +0 -0
  142. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/rope.py +0 -0
  143. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/swiglu.py +0 -0
  144. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/tvd.py +0 -0
  145. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/ops/utils.py +0 -0
  146. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/__init__.py +0 -0
  147. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/auto_model.py +0 -0
  148. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/cross_entropy.py +0 -0
  149. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/dyt.py +0 -0
  150. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/experimental/embedding.py +0 -0
  151. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/functional.py +0 -0
  152. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/fused_linear_cross_entropy.py +0 -0
  153. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/fused_linear_jsd.py +0 -0
  154. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/geglu.py +0 -0
  155. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/gema3_rms.py +0 -0
  156. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/group_norm.py +0 -0
  157. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/jsd.py +0 -0
  158. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/kl_div.py +0 -0
  159. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/layer_norm.py +0 -0
  160. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/__init__.py +0 -0
  161. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/gemma.py +0 -0
  162. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/gemma2.py +0 -0
  163. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/gemma3.py +0 -0
  164. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/glm4.py +0 -0
  165. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/llama.py +0 -0
  166. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/llava.py +0 -0
  167. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/loss_utils.py +0 -0
  168. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/mistral.py +0 -0
  169. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/mixtral.py +0 -0
  170. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/mllama.py +0 -0
  171. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/olmo2.py +0 -0
  172. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/paligemma.py +0 -0
  173. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/phi3.py +0 -0
  174. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/qwen2.py +0 -0
  175. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/qwen2_5_vl.py +0 -0
  176. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/qwen2_vl.py +0 -0
  177. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/model/qwen3.py +0 -0
  178. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/monkey_patch.py +0 -0
  179. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/qwen2vl_mrope.py +0 -0
  180. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/rms_norm.py +0 -0
  181. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/rope.py +0 -0
  182. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/swiglu.py +0 -0
  183. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/trainer/__init__.py +0 -0
  184. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/trainer/orpo_trainer.py +0 -0
  185. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/trainer_integration.py +0 -0
  186. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/transformers/tvd.py +0 -0
  187. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/triton/__init__.py +0 -0
  188. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/triton/monkey_patch.py +0 -0
  189. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel/utils.py +0 -0
  190. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel_nightly.egg-info/dependency_links.txt +0 -0
  191. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel_nightly.egg-info/requires.txt +0 -0
  192. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/src/liger_kernel_nightly.egg-info/top_level.txt +0 -0
  193. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/__init__.py +0 -0
  194. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/__init__.py +0 -0
  195. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_cpo_loss.py +0 -0
  196. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_grpo_loss.py +0 -0
  197. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_jsd_loss.py +0 -0
  198. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_kto_loss.py +0 -0
  199. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_orpo_loss.py +0 -0
  200. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/chunked_loss/test_simpo_loss.py +0 -0
  201. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/conftest.py +0 -0
  202. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/__init__.py +0 -0
  203. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/bf16/__init__.py +0 -0
  204. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/bf16/test_mini_models.py +0 -0
  205. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/bf16/test_mini_models_multimodal.py +0 -0
  206. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/bf16/test_mini_models_with_logits.py +0 -0
  207. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/fp32/__init__.py +0 -0
  208. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/fp32/test_mini_models.py +0 -0
  209. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/fp32/test_mini_models_multimodal.py +0 -0
  210. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/convergence/fp32/test_mini_models_with_logits.py +0 -0
  211. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Google/Gemma3/gemma-3-4b-it/tokenizer_config.json +0 -0
  212. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Google/Paligemma/paligemma-3b-pt-224/tokenizer_config.json +0 -0
  213. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/preprocessor_config.json +0 -0
  214. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/processor_config.json +0 -0
  215. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/tokenizer_config.json +0 -0
  216. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Qwen/Qwen2-VL-7B-Instruct/tokenizer_config.json +0 -0
  217. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/Qwen/Qwen2.5-VL-7B-Instruct/tokenizer_config.json +0 -0
  218. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/fake_configs/meta-llama/Llama-3.2-11B-Vision-Instruct/tokenizer_config.json +0 -0
  219. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/scripts/generate_tokenized_dataset.py +0 -0
  220. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/tiny_shakespeare.txt +0 -0
  221. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/tiny_shakespeare_tokenized/data-00000-of-00001.arrow +0 -0
  222. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/tiny_shakespeare_tokenized/dataset_info.json +0 -0
  223. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/resources/tiny_shakespeare_tokenized/state.json +0 -0
  224. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_auto_model.py +0 -0
  225. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_cross_entropy.py +0 -0
  226. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_dyt.py +0 -0
  227. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_embedding.py +0 -0
  228. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_flex_attention.py +0 -0
  229. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_fused_linear_cross_entropy.py +0 -0
  230. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_fused_linear_jsd.py +0 -0
  231. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_geglu.py +0 -0
  232. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_group_norm.py +0 -0
  233. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_jsd.py +0 -0
  234. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_kl_div.py +0 -0
  235. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_layer_norm.py +0 -0
  236. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_mm_int8int2.py +0 -0
  237. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_monkey_patch.py +0 -0
  238. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_qwen2vl_mrope.py +0 -0
  239. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_rms_norm.py +0 -0
  240. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_rope.py +0 -0
  241. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_swiglu.py +0 -0
  242. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_trainer_integration.py +0 -0
  243. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_transformers.py +0 -0
  244. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/transformers/test_tvd.py +0 -0
  245. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/triton/test_triton_monkey_patch.py +0 -0
  246. {liger_kernel_nightly-0.5.9.dev20250505192105 → liger_kernel_nightly-0.5.9.dev20250508211521}/test/utils.py +0 -0
@@ -0,0 +1,79 @@
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <project version="4">
3
+ <component name="AutoImportSettings">
4
+ <option name="autoReloadType" value="SELECTIVE" />
5
+ </component>
6
+ <component name="ChangeListManager">
7
+ <list default="true" id="d7400753-faa8-4997-a53e-65fd3a6e6146" name="Changes" comment="Reference Unsloth in header" />
8
+ <option name="SHOW_DIALOG" value="false" />
9
+ <option name="HIGHLIGHT_CONFLICTS" value="true" />
10
+ <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
11
+ <option name="LAST_RESOLUTION" value="IGNORE" />
12
+ </component>
13
+ <component name="Git.Settings">
14
+ <option name="RECENT_BRANCH_BY_REPOSITORY">
15
+ <map>
16
+ <entry key="$PROJECT_DIR$" value="main" />
17
+ </map>
18
+ </option>
19
+ <option name="RECENT_GIT_ROOT_PATH" value="$PROJECT_DIR$" />
20
+ </component>
21
+ <component name="GitHubPullRequestSearchHistory"><![CDATA[{
22
+ "lastFilter": {
23
+ "state": "OPEN",
24
+ "assignee": "momochen"
25
+ }
26
+ }]]></component>
27
+ <component name="GithubPullRequestsUISettings"><![CDATA[{
28
+ "selectedUrlAndAccountId": {
29
+ "url": "https://github.com/momochen/Liger-Kernel",
30
+ "accountId": "639f3e12-86db-4b12-a409-51cc017415fb"
31
+ }
32
+ }]]></component>
33
+ <component name="ProjectColorInfo"><![CDATA[{
34
+ "associatedIndex": 5
35
+ }]]></component>
36
+ <component name="ProjectId" id="2lfyDxCjSnvFrbllYmf9VBSCcMx" />
37
+ <component name="ProjectViewState">
38
+ <option name="hideEmptyMiddlePackages" value="true" />
39
+ <option name="showLibraryContents" value="true" />
40
+ </component>
41
+ <component name="PropertiesComponent"><![CDATA[{
42
+ "keyToString": {
43
+ "RunOnceActivity.ShowReadmeOnStart": "true",
44
+ "git-widget-placeholder": "ref__unsloth",
45
+ "last_opened_file_path": "/Users/ychen/workspace/github/Liger-Kernel"
46
+ }
47
+ }]]></component>
48
+ <component name="SharedIndexes">
49
+ <attachedChunks>
50
+ <set>
51
+ <option value="bundled-python-sdk-975db3bf15a3-31b6be0877a2-com.jetbrains.pycharm.community.sharedIndexes.bundled-PC-241.18034.82" />
52
+ </set>
53
+ </attachedChunks>
54
+ </component>
55
+ <component name="SpellCheckerSettings" RuntimeDictionaries="0" Folders="0" CustomDictionaries="0" DefaultDictionary="application-level" UseSingleDictionary="true" transferred="true" />
56
+ <component name="TaskManager">
57
+ <task active="true" id="Default" summary="Default task">
58
+ <changelist id="d7400753-faa8-4997-a53e-65fd3a6e6146" name="Changes" comment="" />
59
+ <created>1725585310555</created>
60
+ <option name="number" value="Default" />
61
+ <option name="presentableId" value="Default" />
62
+ <updated>1725585310555</updated>
63
+ </task>
64
+ <task id="LOCAL-00001" summary="Reference Unsloth in header">
65
+ <option name="closed" value="true" />
66
+ <created>1725585434299</created>
67
+ <option name="number" value="00001" />
68
+ <option name="presentableId" value="LOCAL-00001" />
69
+ <option name="project" value="LOCAL" />
70
+ <updated>1725585434299</updated>
71
+ </task>
72
+ <option name="localTasksCounter" value="2" />
73
+ <servers />
74
+ </component>
75
+ <component name="VcsManagerConfiguration">
76
+ <MESSAGE value="Reference Unsloth in header" />
77
+ <option name="LAST_COMMIT_MESSAGE" value="Reference Unsloth in header" />
78
+ </component>
79
+ </project>
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: liger_kernel_nightly
3
- Version: 0.5.9.dev20250505192105
3
+ Version: 0.5.9.dev20250508211521
4
4
  Summary: Efficient Triton kernels for LLM Training
5
5
  License: BSD 2-CLAUSE LICENSE
6
6
  Copyright 2024 LinkedIn Corporation
@@ -56,7 +56,6 @@ Requires-Dist: mkdocs-material; extra == "dev"
56
56
  <th style="padding: 10px;" colspan="2">Stable</th>
57
57
  <th style="padding: 10px;" colspan="2">Nightly</th>
58
58
  <th style="padding: 10px;">Discord</th>
59
- <th style="padding: 10px;">Build</th>
60
59
  </tr>
61
60
  <tr>
62
61
  <td style="padding: 10px;">
@@ -84,23 +83,6 @@ Requires-Dist: mkdocs-material; extra == "dev"
84
83
  <img src="https://dcbadge.vercel.app/api/server/gpumode?style=flat" alt="Join Our Discord">
85
84
  </a>
86
85
  </td>
87
- <td style="padding: 10px;">
88
- <div style="display: block;">
89
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
90
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
91
- </a>
92
- </div>
93
- <div style="display: block;">
94
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
95
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
96
- </a>
97
- </div>
98
- <div style="display: block;">
99
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
100
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
101
- </a>
102
- </div>
103
- </td>
104
86
  </tr>
105
87
  </table>
106
88
 
@@ -387,6 +369,36 @@ loss.backward()
387
369
  - [Axolotl](https://axolotl.ai/): Integrating Liger Kernel into Axolotl.
388
370
  - [Llama-Factory](https://github.com/hiyouga/LLaMA-Factory): Integrating Liger Kernel into Llama-Factory.
389
371
 
372
+
373
+ ## CI status
374
+
375
+ <table style="width: 100%; text-align: center; border-collapse: collapse;">
376
+ <tr>
377
+ <th style="padding: 10px;">Build</th>
378
+ </tr>
379
+ <tr>
380
+ <td style="padding: 10px;">
381
+ <div style="display: block;">
382
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
383
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
384
+ </a>
385
+ </div>
386
+ <div style="display: block;">
387
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
388
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
389
+ </a>
390
+ </div>
391
+ <div style="display: block;">
392
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
393
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
394
+ </a>
395
+ </div>
396
+ </td>
397
+ </tr>
398
+ </table>
399
+
400
+
401
+
390
402
  ## Contact
391
403
 
392
404
  - For issues, create a Github ticket in this repository
@@ -8,7 +8,6 @@
8
8
  <th style="padding: 10px;" colspan="2">Stable</th>
9
9
  <th style="padding: 10px;" colspan="2">Nightly</th>
10
10
  <th style="padding: 10px;">Discord</th>
11
- <th style="padding: 10px;">Build</th>
12
11
  </tr>
13
12
  <tr>
14
13
  <td style="padding: 10px;">
@@ -36,23 +35,6 @@
36
35
  <img src="https://dcbadge.vercel.app/api/server/gpumode?style=flat" alt="Join Our Discord">
37
36
  </a>
38
37
  </td>
39
- <td style="padding: 10px;">
40
- <div style="display: block;">
41
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
42
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
43
- </a>
44
- </div>
45
- <div style="display: block;">
46
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
47
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
48
- </a>
49
- </div>
50
- <div style="display: block;">
51
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
52
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
53
- </a>
54
- </div>
55
- </td>
56
38
  </tr>
57
39
  </table>
58
40
 
@@ -339,6 +321,36 @@ loss.backward()
339
321
  - [Axolotl](https://axolotl.ai/): Integrating Liger Kernel into Axolotl.
340
322
  - [Llama-Factory](https://github.com/hiyouga/LLaMA-Factory): Integrating Liger Kernel into Llama-Factory.
341
323
 
324
+
325
+ ## CI status
326
+
327
+ <table style="width: 100%; text-align: center; border-collapse: collapse;">
328
+ <tr>
329
+ <th style="padding: 10px;">Build</th>
330
+ </tr>
331
+ <tr>
332
+ <td style="padding: 10px;">
333
+ <div style="display: block;">
334
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
335
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
336
+ </a>
337
+ </div>
338
+ <div style="display: block;">
339
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
340
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
341
+ </a>
342
+ </div>
343
+ <div style="display: block;">
344
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
345
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
346
+ </a>
347
+ </div>
348
+ </td>
349
+ </tr>
350
+ </table>
351
+
352
+
353
+
342
354
  ## Contact
343
355
 
344
356
  - For issues, create a Github ticket in this repository
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "liger_kernel_nightly"
7
- version = "0.5.9.dev20250505192105"
7
+ version = "0.5.9.dev20250508211521"
8
8
  description = "Efficient Triton kernels for LLM Training"
9
9
  urls = { "Homepage" = "https://github.com/linkedin/Liger-Kernel" }
10
10
  readme = { file = "README.md", content-type = "text/markdown" }
@@ -128,7 +128,7 @@ class LigerFusedLinearDPOLoss(torch.nn.Module):
128
128
  compute_nll_loss: bool = False,
129
129
  compiled: bool = True,
130
130
  use_ref_model: bool = True,
131
- average_log_prob: bool = True,
131
+ average_log_prob: bool = False,
132
132
  chunk_size: int = 1,
133
133
  ):
134
134
  """
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: liger_kernel_nightly
3
- Version: 0.5.9.dev20250505192105
3
+ Version: 0.5.9.dev20250508211521
4
4
  Summary: Efficient Triton kernels for LLM Training
5
5
  License: BSD 2-CLAUSE LICENSE
6
6
  Copyright 2024 LinkedIn Corporation
@@ -56,7 +56,6 @@ Requires-Dist: mkdocs-material; extra == "dev"
56
56
  <th style="padding: 10px;" colspan="2">Stable</th>
57
57
  <th style="padding: 10px;" colspan="2">Nightly</th>
58
58
  <th style="padding: 10px;">Discord</th>
59
- <th style="padding: 10px;">Build</th>
60
59
  </tr>
61
60
  <tr>
62
61
  <td style="padding: 10px;">
@@ -84,23 +83,6 @@ Requires-Dist: mkdocs-material; extra == "dev"
84
83
  <img src="https://dcbadge.vercel.app/api/server/gpumode?style=flat" alt="Join Our Discord">
85
84
  </a>
86
85
  </td>
87
- <td style="padding: 10px;">
88
- <div style="display: block;">
89
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
90
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
91
- </a>
92
- </div>
93
- <div style="display: block;">
94
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
95
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
96
- </a>
97
- </div>
98
- <div style="display: block;">
99
- <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
100
- <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
101
- </a>
102
- </div>
103
- </td>
104
86
  </tr>
105
87
  </table>
106
88
 
@@ -387,6 +369,36 @@ loss.backward()
387
369
  - [Axolotl](https://axolotl.ai/): Integrating Liger Kernel into Axolotl.
388
370
  - [Llama-Factory](https://github.com/hiyouga/LLaMA-Factory): Integrating Liger Kernel into Llama-Factory.
389
371
 
372
+
373
+ ## CI status
374
+
375
+ <table style="width: 100%; text-align: center; border-collapse: collapse;">
376
+ <tr>
377
+ <th style="padding: 10px;">Build</th>
378
+ </tr>
379
+ <tr>
380
+ <td style="padding: 10px;">
381
+ <div style="display: block;">
382
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml">
383
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/nvi-ci.yml/badge.svg?event=schedule" alt="Build">
384
+ </a>
385
+ </div>
386
+ <div style="display: block;">
387
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
388
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml/badge.svg?event=schedule" alt="Build">
389
+ </a>
390
+ </div>
391
+ <div style="display: block;">
392
+ <a href="https://github.com/linkedin/Liger-Kernel/actions/workflows/amd-ci.yml">
393
+ <img src="https://github.com/linkedin/Liger-Kernel/actions/workflows/intel-ci.yml/badge.svg?event=schedule" alt="Build">
394
+ </a>
395
+ </div>
396
+ </td>
397
+ </tr>
398
+ </table>
399
+
400
+
401
+
390
402
  ## Contact
391
403
 
392
404
  - For issues, create a Github ticket in this repository
@@ -15,6 +15,7 @@ setup.py
15
15
  .github/workflows/nvi-ci.yml
16
16
  .github/workflows/publish-nightly.yml
17
17
  .github/workflows/publish-release.yml
18
+ .idea/workspace.xml
18
19
  benchmark/README.md
19
20
  benchmark/__init__.py
20
21
  benchmark/benchmarks_visualizer.py
@@ -95,6 +95,7 @@ class TorchLMHeadDPO(torch.nn.Module):
95
95
  ref_x,
96
96
  self.ref_lin.weight,
97
97
  self.ref_lin.bias,
98
+ average_log_prob=True,
98
99
  )
99
100
 
100
101
 
@@ -118,6 +119,7 @@ class LigerLMHeadDPO(torch.nn.Module):
118
119
  beta=beta,
119
120
  use_ref_model=True,
120
121
  compute_nll_loss=compute_nll_loss,
122
+ average_log_prob=True,
121
123
  )
122
124
 
123
125
  def forward(self, x, ref_x, y):