https://github.com/pytorch/torchtitan
· scanned 2026-05-15 18:26 UTC (2 weeks, 6 days ago)
· 10 languages
561 findings (401 legacy + 160 scanner) 82nd percentile · Python · medium (20-100K LoC) Scanner says 90 (lower by 11)
Last scanned 2 weeks, 6 days ago · v1 · 398 findings from 1 source. Findings combine the legacy security pipeline AND the multi-layer engine (atlas, wiring, flows, ranked) AND verified AI agent contributions.
| Component | Sub-score | Weight | Contribution |
|---|---|---|---|
structure_score |
60.0 | 0.15 | 9.00 |
security_score |
92.7 | 0.25 | 23.18 |
testing_score |
85.0 | 0.20 | 17.00 |
documentation_score |
86.0 | 0.15 | 12.90 |
practices_score |
80.0 | 0.15 | 12.00 |
code_quality |
50.0 | 0.10 | 5.00 |
| Overall | 1.00 | 79.1 |
All 3181 nodes from the latest scan, grouped by kind. Each node is a unit the engine identified (file, function, endpoint, table…). Most users won't need this view — it's primarily for debugging the engine's graph extraction or for AI agents that want to enumerate the project structure.
| Label | Layer | Status | Path |
|---|---|---|---|
_hash_model_impl |
software | healthy | tests/utils.py:14 |
hash_named_tensor |
software | healthy | tests/utils.py:30 |
hash_model |
software | healthy | tests/utils.py:64 |
hash_gradient |
software | healthy | tests/utils.py:94 |
__repr__ |
software | healthy | tests/integration_tests/__init__.py:29 |
build_h100_tests_list |
software | healthy | tests/integration_tests/h100.py:15 |
build_flux_test_list |
software | healthy | tests/integration_tests/flux.py:16 |
run_single_test |
software | healthy | tests/integration_tests/flux.py:61 |
run_tests |
software | healthy | tests/integration_tests/flux.py:114 |
main |
software | healthy | tests/integration_tests/flux.py:134 |
build_features_test_list |
software | healthy | tests/integration_tests/features.py:25 |
build_model_tests_list |
software | healthy | tests/integration_tests/models.py:11 |
__init__ |
software | healthy | tests/integration_tests/run_tests.py:41 |
acquire |
software | healthy | tests/integration_tests/run_tests.py:46 |
release |
software | healthy | tests/integration_tests/run_tests.py:54 |
_run_cmd |
software | healthy | tests/integration_tests/run_tests.py:60 |
_emit_block |
software | healthy | tests/integration_tests/run_tests.py:90 |
run_single_test |
software | healthy | tests/integration_tests/run_tests.py:105 |
_filter_tests |
software | healthy | tests/integration_tests/run_tests.py:187 |
run_tests |
software | healthy | tests/integration_tests/run_tests.py:217 |
_runner |
software | healthy | tests/integration_tests/run_tests.py:248 |
main |
software | healthy | tests/integration_tests/run_tests.py:305 |
__init__ |
software | healthy | tests/unit_tests/test_set_determinism.py:25 |
get_mesh |
software | healthy | tests/unit_tests/test_set_determinism.py:50 |
get_optional_mesh |
software | healthy | tests/unit_tests/test_set_determinism.py:75 |
get_all_meshes |
software | healthy | tests/unit_tests/test_set_determinism.py:83 |
__getitem__ |
software | healthy | tests/unit_tests/test_set_determinism.py:87 |
get_coordinate |
software | healthy | tests/unit_tests/test_set_determinism.py:91 |
setUp |
software | healthy | tests/unit_tests/test_set_determinism.py:99 |
tearDown |
software | healthy | tests/unit_tests/test_set_determinism.py:103 |
test_seed_uniqueness_2d_mesh |
software | healthy | tests/unit_tests/test_set_determinism.py:113 |
test_seed_uniqueness_3d_mesh |
software | healthy | tests/unit_tests/test_set_determinism.py:177 |
test_set_determinism_single_gpu |
software | healthy | tests/unit_tests/test_set_determinism.py:257 |
test_detect_anomaly_enabled |
software | healthy | tests/unit_tests/test_set_determinism.py:281 |
test_detect_anomaly_disabled_by_default |
software | healthy | tests/unit_tests/test_set_determinism.py:301 |
_build_dataloader |
software | healthy | tests/unit_tests/test_mm_dataset_checkpointing.py:32 |
test_cc12m_resumption |
software | healthy | tests/unit_tests/test_mm_dataset_checkpointing.py:53 |
test_ignore_index_equal_per_token_contribution |
software | healthy | tests/unit_tests/test_loss.py:20 |
test_ignore_index_gradient_accumulation_consistency |
software | healthy | tests/unit_tests/test_loss.py:96 |
test_accumulate_matches_cat |
software | healthy | tests/unit_tests/test_loss.py:148 |
test_accumulate_with_dtype_conversion |
software | healthy | tests/unit_tests/test_loss.py:168 |
test_too_many_adds_raises |
software | healthy | tests/unit_tests/test_loss.py:186 |
__init__ |
software | healthy | tests/unit_tests/test_loss.py:198 |
forward |
software | healthy | tests/unit_tests/test_loss.py:206 |
_make_model_and_loss |
software | healthy | tests/unit_tests/test_loss.py:213 |
test_numerical_equivalence |
software | healthy | tests/unit_tests/test_loss.py:221 |
test_different_chunk_counts |
software | healthy | tests/unit_tests/test_loss.py:283 |
test_config_build |
software | healthy | tests/unit_tests/test_embedding.py:20 |
test_config_build_without_fields_raises |
software | healthy | tests/unit_tests/test_embedding.py:28 |
test_init_states |
software | healthy | tests/unit_tests/test_embedding.py:33 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
OverrideDefinitions |
software | healthy | tests/integration_tests/__init__.py:16 |
GPUPool |
software | healthy | tests/integration_tests/run_tests.py:34 |
FakeParallelDims |
software | healthy | tests/unit_tests/test_set_determinism.py:16 |
TestSetDeterminismWithFakeMesh |
software | healthy | tests/unit_tests/test_set_determinism.py:96 |
TestMMDatasetCheckpointing |
software | healthy | tests/unit_tests/test_mm_dataset_checkpointing.py:29 |
TestLoss |
software | healthy | tests/unit_tests/test_loss.py:19 |
TestGradAccumulator |
software | healthy | tests/unit_tests/test_loss.py:147 |
_FakeDecoder |
software | healthy | tests/unit_tests/test_loss.py:195 |
TestChunkedCELoss |
software | healthy | tests/unit_tests/test_loss.py:212 |
TestEmbedding |
software | healthy | tests/unit_tests/test_embedding.py:17 |
PresetConfig |
software | healthy | tests/unit_tests/test_embedding.py:82 |
TestFluxDataLoader |
software | healthy | tests/unit_tests/test_dataset_flux.py:15 |
TestProfilerConfig |
software | healthy | tests/unit_tests/test_profiler.py:13 |
TestProfilerInit |
software | healthy | tests/unit_tests/test_profiler.py:50 |
TestProfilerDisabledPaths |
software | healthy | tests/unit_tests/test_profiler.py:61 |
TestProfilerEnabledPaths |
software | healthy | tests/unit_tests/test_profiler.py:143 |
TestConfigurable |
software | healthy | tests/unit_tests/test_configurable.py:13 |
OldStyleComponent |
software | healthy | tests/unit_tests/test_configurable.py:14 |
Config |
software | healthy | tests/unit_tests/test_configurable.py:127 |
NoKwargsComponent |
software | healthy | tests/unit_tests/test_configurable.py:25 |
MyComponent |
software | healthy | tests/unit_tests/test_configurable.py:38 |
BadSlots |
software | healthy | tests/unit_tests/test_configurable.py:55 |
BadKwOnly |
software | healthy | tests/unit_tests/test_configurable.py:64 |
BadBoth |
software | healthy | tests/unit_tests/test_configurable.py:73 |
Inner |
software | healthy | tests/unit_tests/test_configurable.py:116 |
Outer |
software | healthy | tests/unit_tests/test_configurable.py:125 |
SimpleModel |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py:19 |
TestParamGroupConfig |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py:52 |
TestOptimizersContainerWithParamGroups |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py:240 |
TestOptimizersInBackwardWithParamGroups |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py:276 |
TestDCPWithParamGroups |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py:306 |
TestLinear |
software | healthy | tests/unit_tests/test_linear.py:17 |
TestTPKVHeadsValidation |
software | healthy | tests/unit_tests/test_tp_kv_heads_validation.py:110 |
TestLRScheduler |
software | healthy | tests/unit_tests/test_lr_scheduler.py:17 |
FakeOptimizersContainer |
software | healthy | tests/unit_tests/test_checkpoint.py:22 |
FakeLRSchedulersContainer |
software | healthy | tests/unit_tests/test_checkpoint.py:39 |
FakeDataLoader |
software | healthy | tests/unit_tests/test_checkpoint.py:52 |
DummyFuture |
software | healthy | tests/unit_tests/test_checkpoint.py:65 |
DummyAsyncResult |
software | healthy | tests/unit_tests/test_checkpoint.py:70 |
DummyTrainerConfig |
software | healthy | tests/unit_tests/test_checkpoint.py:86 |
TestCheckpointManager |
software | healthy | tests/unit_tests/test_checkpoint.py:103 |
FakeModelWithFreqsCis |
software | healthy | tests/unit_tests/test_checkpoint.py:566 |
DummyDataset |
software | healthy | tests/unit_tests/test_dataloader.py:16 |
DummyTokenizer |
software | healthy | tests/unit_tests/test_dataloader.py:24 |
TestParallelAwareDataloader |
software | healthy | tests/unit_tests/test_dataloader.py:51 |
TestModuleInitStates |
software | healthy | tests/unit_tests/test_module.py:17 |
SimpleModule |
software | healthy | tests/unit_tests/test_module.py:23 |
Child |
software | healthy | tests/unit_tests/test_module.py:35 |
Parent |
software | healthy | tests/unit_tests/test_module.py:257 |
BufferModule |
software | healthy | tests/unit_tests/test_module.py:53 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
README.md |
software | healthy | README.md |
CONTRIBUTING.md |
software | healthy | CONTRIBUTING.md |
run_train.sh |
software | healthy | run_train.sh |
.pre-commit-config.yaml |
software | healthy | .pre-commit-config.yaml |
pyproject.toml |
software | healthy | pyproject.toml |
CODE_OF_CONDUCT.md |
software | healthy | CODE_OF_CONDUCT.md |
__init__.py |
software | healthy | tests/__init__.py |
README.md |
software | healthy | tests/README.md |
utils.py |
software | healthy | tests/utils.py |
__init__.py |
software | healthy | tests/integration_tests/__init__.py |
h100.py |
software | healthy | tests/integration_tests/h100.py |
flux.py |
software | healthy | tests/integration_tests/flux.py |
features.py |
software | healthy | tests/integration_tests/features.py |
models.py |
software | healthy | tests/integration_tests/models.py |
run_tests.py |
software | healthy | tests/integration_tests/run_tests.py |
test_set_determinism.py |
software | healthy | tests/unit_tests/test_set_determinism.py |
test_mm_dataset_checkpointing.py |
software | healthy | tests/unit_tests/test_mm_dataset_checkpointing.py |
test_loss.py |
software | healthy | tests/unit_tests/test_loss.py |
test_embedding.py |
software | healthy | tests/unit_tests/test_embedding.py |
test_dataset_flux.py |
software | healthy | tests/unit_tests/test_dataset_flux.py |
__init__.py |
software | healthy | tests/unit_tests/__init__.py |
test_profiler.py |
software | healthy | tests/unit_tests/test_profiler.py |
test_configurable.py |
software | healthy | tests/unit_tests/test_configurable.py |
test_optimizer_param_groups.py |
software | healthy | tests/unit_tests/test_optimizer_param_groups.py |
test_linear.py |
software | healthy | tests/unit_tests/test_linear.py |
test_tp_kv_heads_validation.py |
software | healthy | tests/unit_tests/test_tp_kv_heads_validation.py |
test_lr_scheduler.py |
software | healthy | tests/unit_tests/test_lr_scheduler.py |
test_checkpoint.py |
software | healthy | tests/unit_tests/test_checkpoint.py |
test_lora.py |
software | healthy | tests/unit_tests/test_lora.py |
test_quantization.py |
software | healthy | tests/unit_tests/test_quantization.py |
test_dataloader.py |
software | healthy | tests/unit_tests/test_dataloader.py |
test_module.py |
software | healthy | tests/unit_tests/test_module.py |
test_activation_checkpoint.py |
software | healthy | tests/unit_tests/test_activation_checkpoint.py |
test_config_manager.py |
software | healthy | tests/unit_tests/test_config_manager.py |
test_dataset_checkpointing.py |
software | healthy | tests/unit_tests/test_dataset_checkpointing.py |
test_train_spec.py |
software | healthy | tests/unit_tests/test_train_spec.py |
test_download_hf_assets.py |
software | healthy | tests/unit_tests/test_download_hf_assets.py |
test_compile_moe.py |
software | healthy | tests/unit_tests/test_compile_moe.py |
test_expert_parallel.py |
software | healthy | tests/unit_tests/test_expert_parallel.py |
test_fsdp_moe_sharding.py |
software | healthy | tests/unit_tests/test_fsdp_moe_sharding.py |
test_tokenizer.py |
software | healthy | tests/unit_tests/test_tokenizer.py |
test_rope.py |
software | healthy | tests/unit_tests/test_rope.py |
test_parallel_dims.py |
software | healthy | tests/unit_tests/test_parallel_dims.py |
test_weight_tying.py |
software | healthy | tests/unit_tests/test_weight_tying.py |
test_chat_dataset.py |
software | healthy | tests/unit_tests/test_chat_dataset.py |
test_rmsnorm.py |
software | healthy | tests/unit_tests/test_rmsnorm.py |
__init__.py |
software | healthy | tests/unit_tests/observability/__init__.py |
test_structured_logging.py |
software | healthy | tests/unit_tests/observability/test_structured_logging.py |
tokenizer_config.json |
software | healthy | tests/assets/tokenizer/tokenizer_config.json |
tokenizer.json |
software | healthy | tests/assets/tokenizer/tokenizer.json |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
tests |
software | healthy | tests |
integration_tests |
software | healthy | tests/integration_tests |
unit_tests |
software | healthy | tests/unit_tests |
observability |
software | healthy | tests/unit_tests/observability |
assets |
software | healthy | tests/assets |
tokenizer |
software | healthy | tests/assets/tokenizer |
flux_test_encoders |
software | healthy | tests/assets/flux_test_encoders |
clip-vit-large-patch14 |
software | healthy | tests/assets/flux_test_encoders/clip-vit-large-patch14 |
t5-v1_1-xxl |
software | healthy | tests/assets/flux_test_encoders/t5-v1_1-xxl |
sft_test |
software | healthy | tests/assets/sft_test |
c4_test |
software | healthy | tests/assets/c4_test |
cc12m_test |
software | healthy | tests/assets/cc12m_test |
benchmarks |
software | healthy | benchmarks |
docs |
software | healthy | docs |
scripts |
software | healthy | scripts |
generate |
software | healthy | scripts/generate |
ci |
software | healthy | scripts/ci |
checkpoint_conversion |
software | healthy | scripts/checkpoint_conversion |
.github |
software | healthy | .github |
ISSUE_TEMPLATE |
software | healthy | .github/ISSUE_TEMPLATE |
scripts |
software | healthy | .github/scripts |
workflows |
software | healthy | .github/workflows |
torchtitan |
software | healthy | torchtitan |
models |
software | healthy | torchtitan/models |
llama4 |
software | healthy | torchtitan/models/llama4 |
common |
software | healthy | torchtitan/models/common |
gpt_oss |
software | healthy | torchtitan/models/gpt_oss |
flux |
software | healthy | torchtitan/models/flux |
inference |
software | healthy | torchtitan/models/flux/inference |
model |
software | healthy | torchtitan/models/flux/model |
llama3 |
software | healthy | torchtitan/models/llama3 |
qwen3 |
software | healthy | torchtitan/models/qwen3 |
qwen3_vl |
software | healthy | torchtitan/models/qwen3_vl |
deepseek_v3 |
software | healthy | torchtitan/models/deepseek_v3 |
config |
software | healthy | torchtitan/config |
ops |
software | healthy | torchtitan/ops |
components |
software | healthy | torchtitan/components |
quantization |
software | healthy | torchtitan/components/quantization |
observability |
software | healthy | torchtitan/observability |
structured_logger |
software | healthy | torchtitan/observability/structured_logger |
distributed |
software | healthy | torchtitan/distributed |
deepep |
software | healthy | torchtitan/distributed/deepep |
tools |
software | healthy | torchtitan/tools |
hf_datasets |
software | healthy | torchtitan/hf_datasets |
multimodal |
software | healthy | torchtitan/hf_datasets/multimodal |
utils |
software | healthy | torchtitan/hf_datasets/multimodal/utils |
protocols |
software | healthy | torchtitan/protocols |
experiments |
software | healthy | torchtitan/experiments |
forge |
software | healthy | torchtitan/experiments/forge |
transformers_modeling_backend |
software | healthy | torchtitan/experiments/transformers_modeling_backend |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_graph_trainer.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_graph_trainer_h100.… |
claude-code |
cicd | healthy | .github/workflows/claude-code.yml |
docker-build |
cicd | healthy | .github/workflows/docker-builds.yml |
set-matrix |
cicd | healthy | .github/workflows/integration_test_8gpu_features.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_features.yaml |
check_todos |
cicd | healthy | .github/workflows/todo_debt_tracker.yaml |
set-matrix |
cicd | healthy | .github/workflows/integration_test_8gpu_torchft.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_torchft.yaml |
build-test |
cicd | healthy | .github/workflows/unit_test_cpu_torchft.yaml |
set-matrix |
cicd | healthy | .github/workflows/integration_test_8gpu_models.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_models.yaml |
release-build |
cicd | healthy | .github/workflows/release.yml |
pypi-publish |
cicd | healthy | .github/workflows/release.yml |
set-matrix |
cicd | healthy | .github/workflows/integration_test_8gpu_transformers_modeli… |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_transformers_modeli… |
set-matrix |
cicd | healthy | .github/workflows/integration_test_4gpu_rl.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_4gpu_rl.yaml |
set |
cicd | healthy | .github/workflows/set-matrix.yaml |
build-test |
cicd | healthy | .github/workflows/unit_test_cpu.yaml |
lint |
cicd | healthy | .github/workflows/lint.yaml |
set-matrix |
cicd | healthy | .github/workflows/integration_test_8gpu_h100.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_h100.yaml |
build-test |
cicd | healthy | .github/workflows/integration_test_8gpu_rl_h100.yaml |
generate-matrix |
cicd | healthy | .github/workflows/build_whl_and_publish.yaml |
build |
cicd | healthy | .github/workflows/build_whl_and_publish.yaml |
| Label | Layer | Status | Path |
|---|---|---|---|
gha::integration_test_8gpu_graph_trainer |
cicd | healthy | .github/workflows/integration_test_8gpu_graph_trainer.yaml |
gha::integration_test_8gpu_graph_trainer_h100 |
cicd | healthy | .github/workflows/integration_test_8gpu_graph_trainer_h100.… |
gha::claude-code |
cicd | healthy | .github/workflows/claude-code.yml |
gha::docker-builds |
cicd | healthy | .github/workflows/docker-builds.yml |
gha::integration_test_8gpu_features |
cicd | healthy | .github/workflows/integration_test_8gpu_features.yaml |
gha::todo_debt_tracker |
cicd | healthy | .github/workflows/todo_debt_tracker.yaml |
gha::integration_test_8gpu_torchft |
cicd | healthy | .github/workflows/integration_test_8gpu_torchft.yaml |
gha::unit_test_cpu_torchft |
cicd | healthy | .github/workflows/unit_test_cpu_torchft.yaml |
gha::integration_test_8gpu_models |
cicd | healthy | .github/workflows/integration_test_8gpu_models.yaml |
gha::release |
cicd | healthy | .github/workflows/release.yml |
gha::integration_test_8gpu_transformers_modeling_backend |
cicd | healthy | .github/workflows/integration_test_8gpu_transformers_modeli… |
gha::integration_test_4gpu_rl |
cicd | healthy | .github/workflows/integration_test_4gpu_rl.yaml |
gha::set-matrix |
cicd | healthy | .github/workflows/set-matrix.yaml |
gha::unit_test_cpu |
cicd | healthy | .github/workflows/unit_test_cpu.yaml |
gha::lint |
cicd | healthy | .github/workflows/lint.yaml |
gha::integration_test_8gpu_h100 |
cicd | healthy | .github/workflows/integration_test_8gpu_h100.yaml |
gha::integration_test_8gpu_rl_h100 |
cicd | healthy | .github/workflows/integration_test_8gpu_rl_h100.yaml |
gha::build_whl_and_publish |
cicd | healthy | .github/workflows/build_whl_and_publish.yaml |
| Label | Layer | Status | Path |
|---|---|---|---|
repobility-clone-7xrml0a6 |
software | healthy | /tmp/repobility-clone-7xrml0a6 |
| Label | Layer | Status | Path |
|---|---|---|---|
127.0.0.1 |
network | healthy | .pre-commit-config.yaml |
| Label | Layer | Status | Path |
|---|---|---|---|
loadbalancer |
network | healthy | torchtitan/config/configs.py |
| Label | Layer | Status | Path |
|---|---|---|---|
gpu (detected) |
hardware | healthy | run_train.sh |
| Label | Layer | Status | Path |
|---|---|---|---|
vps::aws |
hardware | healthy | .github/workflows/integration_test_8gpu_graph_trainer_h100.… |
| Label | Layer | Status | Path |
|---|---|---|---|
GITHUB_TOKEN |
cicd | healthy | — |
This page is publicly accessible at:
https://repobility.com/scan/9eb441b0-4d1c-45b8-9825-0c639f48bd8b/
To check status programmatically (no auth required):
curl -s https://repobility.com/api/v1/public/scan/9eb441b0-4d1c-45b8-9825-0c639f48bd8b/Important — please don't re-submit the same URL repeatedly. The submission endpoint is idempotent: re-submitting the same git URL returns this same scan_token, not a new one. To re-scan this repo, sign up free and use the dashboard.