lightning-thunder
PyTorch compiler that accelerates training and inference. Get built-in optimizations for performance, memory, parallelism, and easily write your own.
File Explorer
- docker-build.yml
- gpu-coverage.yml
- gpu-tests.yml
- notebook-runs.yml
- bug_report.md
- documentation.md
- feature_request.md
- program_coverage.md
- auto-cc.yml
- ci-benchmark.yml
- ci-checks.yml
- ci-jit-coverage.yml
- ci-quickstart.yml
- ci-testing.yml
- docs-build.yml
- label-conflicts.yml
- labeler.yml
- release-nightly.yml
- release-pypi.yml
- stale-branches.yaml
- CODEOWNERS
- dependabot.yml
- labeling-config.yml
- lightning-probot.yml
- load-jit-coverage-report.py
- load-quickstart-report.py
- PULL_REQUEST_TEMPLATE.md
- run-benchmark-as-lit-jobs.py
- run-jit-coverage-as-lit-studios.py
- run-quickstart-as-lit-jobs.py
- all-tests.yaml
- notebooks.yaml
- transformer-engine.yaml
- Dockerfile
- Dockerfile
- Dockerfile
- README.md
- how_it_works.png
- icon.svg
- lightning_thunder_lightmode_nobyline.png
- LightningThunderDarkModewByline.png
- LightningThunderLightModewByline.png
- logo-large.svg
- logo-small.svg
- logo.png
- logo.svg
- normalized_training_throughput_zero2.png
- pretrain_perf.png
- training_throughput_single.png
- copybutton.js
- theme_variables.jinja
- contrib_thunder.rst
- extending.rst
- inside_thunder.rst
- faq.rst
- inspecting_traces.rst
- mlp_mnist.rst
- overview.rst
- sharp_edges.rst
- examine.rst
- hello_world.rst
- installation.rst
- additional_executors.rst
- benchmarking.rst
- ddp.rst
- whats_next.rst
- index.rst
- index.rst
- baseutils.rst
- codeutils.rst
- devices.rst
- dtypes.rst
- index.rst
- langctxs.rst
- prims.rst
- proxies.rst
- pytree.rst
- rematerialization.rst
- symbol.rst
- trace.rst
- transforms.rst
- index.rst
- index.rst
- index.rst
- apexex.rst
- cudnnex.rst
- index.rst
- nvfuserex.rst
- passes.rst
- pythonex.rst
- torch_compile.rst
- torchex.rst
- triton_crossentropy.rst
- utils.rst
- index.rst
- index.rst
- index.rst
- index.rst
- index.rst
- thunder.rst
- conf.py
- index.rst
- .build_docs.sh
- .readthedocs.yaml
- make.bat
- Makefile
- all.txt
- jit_coverage_hf.py
- requirements.txt
- hf_bert.py
- hf_llm.py
- mlp.py
- requirements.txt
- vit_hf.py
- vit_tv.py
- extend.ipynb
- fsdp_tutorial.ipynb
- thunder-add-vjp-rule.md
- .ignore.ci
- adding_custom_operator.ipynb
- adding_custom_operator_backward.ipynb
- extend_thunder_with_cuda_python.ipynb
- hello_world_thunderfx.ipynb
- liger_kernel.ipynb
- thunder_trace_intro.ipynb
- writing_a_trace_transform_cpu_offloading.ipynb
- zero_to_thunder.ipynb
- base.txt
- coverage.txt
- devel.txt
- docs.txt
- notebooks.txt
- test.txt
- bisect_nvfuser.py
- build_from_source.sh
- remove-torch-lines.sh
- sanity-check.sh
- validate_build.py
- __init__.py
- benchmark_hf.py
- benchmark_inference.py
- benchmark_litgpt.py
- benchmark_peft.py
- conftest.py
- distributed.py
- einsum.py
- layers_for_inference_benchmark.py
- targets.py
- test_benchmark_litgpt.py
- utils.py
- __init__.py
- langctx.py
- utils.py
- __init__.py
- baseutils.py
- codeutils.py
- compile_data.py
- devices.py
- dtypes.py
- interpreter.py
- jit_ext.py
- langctxs.py
- module.py
- options.py
- patterns.py
- prims.py
- profile.py
- proxies.py
- pytree.py
- recipe.py
- rematerialization.py
- symbol.py
- trace.py
- trace_interpreter.py
- transform_common.py
- transforms.py
- update_aliases.py
- utils.py
- vjp_utils.py
- __init__.py
- benchmark.py
- check_trace.py
- debug_transform.py
- nvtx_profile_transform.py
- profile_transform.py
- utils.py
- __init__.py
- column_wise.py
- common.py
- optimize_comm.py
- row_wise.py
- __init__.py
- ddp.py
- ddp_v2.py
- fsdp.py
- fsdp_v2.py
- __init__.py
- bucketing.py
- checkpoint.py
- prims.py
- utils.py
- __init__.py
- benchmark_utils.py
- compiler.py
- compiler_graph_benchmark.py
- report.py
- repro_script_template.py
- splitter.py
- utils.py
- __init__.py
- memory_calculation.py
- __init__.py
- apex_entropyex_impl.py
- apex_fused_rms_norm_impl.py
- apexex.py
- cudnn_layernormex.py
- cudnn_sdpa.py
- cudnnex.py
- custom_op_ex.py
- data_dependent_partition.py
- fa3ex.py
- nvfuserex.py
- nvfuserex_impl.py
- passes.py
- pythonex.py
- sdpaex.py
- torch_autograd.py
- torch_compile.py
- torchex.py
- transformer_engineex.py
- transformer_engineex_impl.py
- triton_crossentropy.py
- triton_crossentropy_impl.py
- triton_utils.py
- utils.py
- __init__.py
- __init__.py
- langctx.py
- __init__.py
- distributed.py
- fp8.py
- profile.py
- quantization.py
- reduce_overhead.py
- __init__.py
- base.py
- hf_transformers.py
- __init__.py
- test_coverage_hf_diffusers.py
- test_coverage_trace.py
- __init__.py
- helper.py
- modules.py
- test_checkpoint.py
- test_ddp.py
- test_dtensor.py
- test_fsdp.py
- test_moe.py
- test_ops.py
- test_tensor_parallel.py
- __init__.py
- bf16.py
- conftest.py
- framework.py
- hf_bart_self_attn.py
- litgpt_model.py
- llama2_model.py
- llama4_moe.py
- make_tensor.py
- module_example.py
- nanogpt_model.py
- opinfos.py
- README.md
- test_apex_cross_entropy_executor.py
- test_apex_fused_norms.py
- test_auto_register_torchops.py
- test_autocast.py
- test_check_trace.py
- test_core.py
- test_cudnn_executor.py
- test_dynamo.py
- test_einops.py
- test_elementwise.py
- test_examine.py
- test_examine_memory.py
- test_extend.py
- test_fa3_executor.py
- test_grad.py
- test_inplace_copy.py
- test_interpreter.py
- test_jit_general.py
- test_networks.py
- test_numpy_langctx.py
- test_nvfuser.py
- test_nvfuser_remat.py
- test_ops.py
- test_patterns.py
- test_pythonex.py
- test_randomness.py
- test_recipes.py
- test_reductions.py
- test_sdpaex_executor.py
- test_shape_ops.py
- test_torch_compile_executor.py
- test_torch_library_custom_op.py
- test_torch_library_custom_op_with_lists.py
- test_transformer_engine_executor.py
- test_transforms.py
- test_triton_ce.py
- test_update_aliases.py
- utils.py
- __init__.py
- dtensor_codeutils.py
- dtensor_proxy.py
- dtensor_torch_and_prims.py
- dtensor_utils.py
- __init__.py
- custom_op.py
- default_torch_ops.py
- langctx.py
- __init__.py
- autocast.py
- autodiff.py
- constant_folding.py
- cudagraph.py
- extraction_only_prologue_transform.py
- materialization.py
- prune_prologue_checks.py
- qlora.py
- quantization.py
- te_inference.py
- utils.py
- __about__.py
- __init__.py
- common.py
- py.typed
- .codecov.yml
- .git-blame-ignore-revs
- .gitignore
- .pre-commit-config.yaml
- LICENSE
- Makefile
- MANIFEST.in
- pyproject.toml
- README.md
- requirements.txt
- setup.py
# Use via CDN
jsDelivrjsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.
Link
Example
Command Glossary
Commands referenced in this DOCs, explained below.
git clone
View Details ▼
git clone
Clone an existing repository.
git clone {{remote_repository_location}} {{path/to/directory}}
Clone an existing repository into a new directory (the default directory is the repository name):
git clone --recursive {{remote_repository_location}}
Clone an existing repository and its submodules:
git clone {{[-n|--no-checkout]}} {{remote_repository_location}}
Clone only the `.git` directory of an existing repository:
pip install
View Details ▼
pip install
Install Python packages.
pip install {{package1 package2 ...}}
Install one or more packages:
pip install {{package1 package2 ...}} {{[-U|--upgrade]}}
Upgrade all specified packages to the latest version, installing any that are not already present:
pip install {{package}}=={{version}}
Install a specific version of a package:
// repository documentation
Was this content helpful?
(0 ratings)