liger-kernel-nightly 0.5.10.dev20250618170631__tar.gz → 0.5.10.dev20250624183504__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.
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/PKG-INFO +1 -1
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/dev/modal/benchmarks.py +1 -1
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/pyproject.toml +1 -1
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/monkey_patch.py +4 -1
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/rms_norm.py +1 -3
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel_nightly.egg-info/PKG-INFO +1 -1
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/ISSUE_TEMPLATE/bug_report.yaml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/ISSUE_TEMPLATE/feature_request.yaml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/pull_request_template.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/amd-ci.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/benchmark.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/docs.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/intel-ci.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/nvi-ci.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/publish-nightly.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.github/workflows/publish-release.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.gitignore +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/.idea/workspace.xml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/LICENSE +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/Makefile +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/NOTICE +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/benchmarks_visualizer.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/data/all_benchmark_data.csv +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_cpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_distill_jsd_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_dpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_dyt.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_embedding.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_fused_linear_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_fused_linear_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_fused_neighborhood_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_geglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_group_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_kl_div.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_kto_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_layer_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_multi_token_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_orpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_qwen2vl_mrope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_rms_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_rope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_simpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_softmax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_sparse_multi_token_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_sparsemax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_swiglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/benchmark_tvd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/benchmark/scripts/utils.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/dev/fmt-requirements.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/dev/modal/tests.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/dev/modal/tests_bwd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/Examples.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/Getting-Started.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/High-Level-APIs.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/Low-Level-APIs.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/acknowledgement.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/contributing.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/banner.GIF +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/compose.gif +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/e2e-memory.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/e2e-tps.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/logo-banner.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/patch.gif +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/images/post-training.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/index.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/docs/license.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/alignment/accelerate_config.yaml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/alignment/run_orpo.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/callback.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/config/fsdp_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/gemma_7b_mem.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/gemma_7b_tp.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/llama_mem_alloc.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/llama_tps.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/qwen_mem_alloc.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/img/qwen_tps.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/launch_on_modal.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/requirements.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/run_benchmarks.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/run_gemma.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/run_llama.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/run_qwen.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/run_qwen2_vl.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/training.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/huggingface/training_multimodal.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/lightning/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/lightning/requirements.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/lightning/training.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/callback.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Memory_Stage1_num_head_3.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Memory_Stage1_num_head_5.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Memory_Stage2_num_head_3.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Memory_Stage2_num_head_5.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Throughput_Stage1_num_head_3.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Throughput_Stage1_num_head_5.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Throughput_Stage2_num_head_3.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/docs/images/Throughput_Stage2_num_head_5.png +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/fsdp/acc-fsdp.conf +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/medusa_util.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/requirements.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/scripts/llama3_8b_medusa.sh +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/examples/medusa/train.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/licenses/LICENSE-Apache-2.0 +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/licenses/LICENSE-MIT-AutoAWQ +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/licenses/LICENSE-MIT-Efficient-Cross-Entropy +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/licenses/LICENSE-MIT-llmc +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/licenses/LICENSE-MIT-triton +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/mkdocs.yml +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/setup.cfg +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/setup.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/README.md +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/cpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/dpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/functional.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/fused_linear_distillation.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/fused_linear_ppo.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/fused_linear_preference.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/fused_linear_unpaired_preference.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/grpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/jsd_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/kto_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/orpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/chunked_loss/simpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/env_report.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/dyt.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/experimental/embedding.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/experimental/mm_int8int2.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/fused_linear_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/fused_linear_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/fused_neighborhood_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/geglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/group_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/grpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/kl_div.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/layer_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/multi_token_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/qwen2vl_mrope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/rms_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/rope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/softmax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/sparsemax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/swiglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/tvd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/ops/utils.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/auto_model.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/dyt.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/experimental/embedding.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/fsdp.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/functional.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/fused_linear_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/fused_linear_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/fused_neighborhood_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/geglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/group_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/grpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/kl_div.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/layer_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/gemma.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/gemma2.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/gemma3.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/glm4.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/llama.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/llava.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/loss_utils.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/mistral.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/mixtral.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/mllama.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/olmo2.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/paligemma.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/phi3.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/qwen2.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/qwen2_5_vl.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/qwen2_vl.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/qwen3.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/model/qwen3_moe.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/multi_token_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/qwen2vl_mrope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/rope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/softmax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/sparsemax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/swiglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/trainer/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/trainer/orpo_trainer.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/trainer_integration.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/transformers/tvd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/triton/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/triton/monkey_patch.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel/utils.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel_nightly.egg-info/SOURCES.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel_nightly.egg-info/dependency_links.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel_nightly.egg-info/requires.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/src/liger_kernel_nightly.egg-info/top_level.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_cpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_dpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_grpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_jsd_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_kto_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_orpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/chunked_loss/test_simpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/conftest.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/bf16/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/bf16/test_mini_models.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/bf16/test_mini_models_multimodal.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/bf16/test_mini_models_with_logits.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/fp32/__init__.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/fp32/test_mini_models.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/fp32/test_mini_models_multimodal.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/convergence/fp32/test_mini_models_with_logits.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Google/Gemma3/gemma-3-4b-it/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Google/Paligemma/paligemma-3b-pt-224/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/preprocessor_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/processor_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Llava/llava-1.5-7b-hf/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Qwen/Qwen2-VL-7B-Instruct/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/Qwen/Qwen2.5-VL-7B-Instruct/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/fake_configs/meta-llama/Llama-3.2-11B-Vision-Instruct/tokenizer_config.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/scripts/generate_tokenized_dataset.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/tiny_shakespeare.txt +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/tiny_shakespeare_tokenized/data-00000-of-00001.arrow +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/tiny_shakespeare_tokenized/dataset_info.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/resources/tiny_shakespeare_tokenized/state.json +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_auto_model.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_dyt.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_embedding.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_flex_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_fused_linear_cross_entropy.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_fused_linear_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_fused_neighborhood_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_geglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_group_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_grpo_loss.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_jsd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_kl_div.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_layer_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_mm_int8int2.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_monkey_patch.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_multi_token_attention.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_qwen2vl_mrope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_rms_norm.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_rope.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_softmax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_sparsemax.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_swiglu.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_trainer_integration.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_transformers.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/transformers/test_tvd.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/triton/test_triton_monkey_patch.py +0 -0
- {liger_kernel_nightly-0.5.10.dev20250618170631 → liger_kernel_nightly-0.5.10.dev20250624183504}/test/utils.py +0 -0
@@ -14,7 +14,7 @@ app = modal.App("liger_benchmarks", image=image)
|
|
14
14
|
repo = image.add_local_dir(ROOT_PATH, remote_path=REMOTE_ROOT_PATH)
|
15
15
|
|
16
16
|
|
17
|
-
@app.function(gpu="
|
17
|
+
@app.function(gpu=["100"], image=repo, timeout=60 * 90)
|
18
18
|
def liger_benchmarks():
|
19
19
|
import os
|
20
20
|
import subprocess
|
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
|
|
4
4
|
|
5
5
|
[project]
|
6
6
|
name = "liger_kernel_nightly"
|
7
|
-
version = "0.5.10.
|
7
|
+
version = "0.5.10.dev20250624183504"
|
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" }
|
@@ -54,7 +54,7 @@ def _bind_method_to_module(module, method_name: str, new_method: Callable):
|
|
54
54
|
module.__dict__[method_name] = new_method.__get__(module, module.__class__)
|
55
55
|
|
56
56
|
|
57
|
-
def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", in_place=True):
|
57
|
+
def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", in_place=True, row_mode=None):
|
58
58
|
# Check if the module is a PEFT ModulesToSaveWrapper
|
59
59
|
# If it is, we need to patch the modules_to_save.default and original_modules
|
60
60
|
if PEFT_AVAILABLE and isinstance(module, peft.utils.other.ModulesToSaveWrapper):
|
@@ -64,12 +64,14 @@ def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", i
|
|
64
64
|
getattr(module, "variance_epsilon", None) or getattr(module, "eps", None) or eps
|
65
65
|
)
|
66
66
|
module.modules_to_save.default.in_place = in_place
|
67
|
+
module.modules_to_save.default.row_mode = row_mode
|
67
68
|
module.original_module.offset = offset
|
68
69
|
module.original_module.casting_mode = casting_mode
|
69
70
|
module.original_module.variance_epsilon = (
|
70
71
|
getattr(module, "variance_epsilon", None) or getattr(module, "eps", None) or eps
|
71
72
|
)
|
72
73
|
module.original_module.in_place = in_place
|
74
|
+
module.original_module.row_mode = row_mode
|
73
75
|
_bind_method_to_module(module.modules_to_save.default, "forward", LigerRMSNorm.forward)
|
74
76
|
_bind_method_to_module(module.modules_to_save.default, "extra_repr", LigerRMSNorm.extra_repr)
|
75
77
|
_bind_method_to_module(module.original_module, "forward", LigerRMSNorm.forward)
|
@@ -81,6 +83,7 @@ def _patch_rms_norm_module(module, offset=0.0, eps=1e-6, casting_mode="llama", i
|
|
81
83
|
module.casting_mode = casting_mode
|
82
84
|
module.variance_epsilon = getattr(module, "variance_epsilon", None) or getattr(module, "eps", None) or eps
|
83
85
|
module.in_place = in_place
|
86
|
+
module.row_mode = row_mode
|
84
87
|
_bind_method_to_module(module, "forward", LigerRMSNorm.forward)
|
85
88
|
_bind_method_to_module(module, "extra_repr", LigerRMSNorm.extra_repr)
|
86
89
|
module.__class__.__name__ = LigerRMSNorm.__name__
|
@@ -41,9 +41,7 @@ class LigerRMSNorm(nn.Module):
|
|
41
41
|
)
|
42
42
|
|
43
43
|
def extra_repr(self):
|
44
|
-
return (
|
45
|
-
f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}, offset={self.offset}, in_place={self.in_place}"
|
46
|
-
)
|
44
|
+
return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}, offset={self.offset}, in_place={self.in_place}, row_mode={self.row_mode}"
|
47
45
|
|
48
46
|
|
49
47
|
class LigerRMSNormForGemma(LigerRMSNorm):
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|