https://github.com/linkedin/Liger-Kernel
· scanned 2026-05-15 18:26 UTC (2 weeks, 6 days ago)
· 10 languages
267 findings (124 legacy + 143 scanner) 80th percentile · Python · medium (20-100K LoC) Scanner says 89 (lower by 10)
Last scanned 2 weeks, 6 days ago · v1 · 117 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 |
85.0 | 0.15 | 12.75 |
security_score |
86.0 | 0.25 | 21.50 |
testing_score |
100.0 | 0.20 | 20.00 |
documentation_score |
66.0 | 0.15 | 9.90 |
practices_score |
65.0 | 0.15 | 9.75 |
code_quality |
50.0 | 0.10 | 5.00 |
| Overall | 1.00 | 78.9 |
All 2151 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 |
|---|---|---|---|
get_default_dependencies |
software | healthy | setup.py:10 |
get_optional_dependencies |
software | healthy | setup.py:32 |
is_xpu_available |
software | healthy | setup.py:53 |
is_ascend_available |
software | healthy | setup.py:74 |
get_platform |
software | healthy | setup.py:88 |
clear_gpu_cache |
software | healthy | test/conftest.py:8 |
is_transformers_v5_or_later |
software | healthy | test/utils.py:44 |
get_mllama_rope_config |
software | healthy | test/utils.py:49 |
get_qwen3_vl_rope_config |
software | healthy | test/utils.py:72 |
set_seed |
software | healthy | test/utils.py:93 |
require_deterministic |
software | healthy | test/utils.py:124 |
wrapper |
software | healthy | test/utils.py:126 |
get_logprobs |
software | healthy | test/utils.py:138 |
get_topk |
software | healthy | test/utils.py:143 |
assert_verbose_allclose |
software | healthy | test/utils.py:148 |
simple_collate_fn |
software | healthy | test/utils.py:227 |
multimodal_collate_fn |
software | healthy | test/utils.py:245 |
load_tokenizer_config |
software | healthy | test/utils.py:268 |
load_image_processing_config |
software | healthy | test/utils.py:278 |
load_processor_config |
software | healthy | test/utils.py:285 |
train_bpe_tokenizer |
software | healthy | test/utils.py:292 |
supports_bfloat16 |
software | healthy | test/utils.py:317 |
is_torchvision_available |
software | healthy | test/utils.py:328 |
revert_liger_kernel_to_granite |
software | healthy | test/utils.py:335 |
revert_liger_kernel_to_llama |
software | healthy | test/utils.py:347 |
revert_liger_kernel_to_smollm3 |
software | healthy | test/utils.py:359 |
revert_liger_kernel_to_mllama |
software | healthy | test/utils.py:371 |
revert_liger_kernel_to_llama4 |
software | healthy | test/utils.py:394 |
revert_liger_kernel_to_ministral |
software | healthy | test/utils.py:417 |
revert_liger_kernel_to_mistral |
software | healthy | test/utils.py:429 |
revert_liger_kernel_to_mixtral |
software | healthy | test/utils.py:441 |
revert_liger_kernel_to_gemma |
software | healthy | test/utils.py:453 |
revert_liger_kernel_to_gemma2 |
software | healthy | test/utils.py:465 |
revert_liger_kernel_to_gemma3_text |
software | healthy | test/utils.py:477 |
revert_liger_kernel_to_gemma4_text |
software | healthy | test/utils.py:491 |
revert_liger_kernel_to_gemma3 |
software | healthy | test/utils.py:508 |
revert_liger_kernel_to_Paligemma |
software | healthy | test/utils.py:522 |
revert_liger_kernel_to_qwen2 |
software | healthy | test/utils.py:540 |
revert_liger_kernel_to_qwen3 |
software | healthy | test/utils.py:553 |
revert_liger_kernel_to_qwen3_moe |
software | healthy | test/utils.py:565 |
revert_liger_kernel_to_gpt_oss |
software | healthy | test/utils.py:577 |
revert_liger_kernel_to_qwen2_vl |
software | healthy | test/utils.py:589 |
revert_liger_kernel_to_qwen2_5_vl |
software | healthy | test/utils.py:600 |
revert_liger_kernel_to_qwen3_vl |
software | healthy | test/utils.py:611 |
revert_liger_kernel_to_qwen3_vl_moe |
software | healthy | test/utils.py:622 |
revert_liger_kernel_to_phi3 |
software | healthy | test/utils.py:633 |
revert_liger_kernel_to_pixtral |
software | healthy | test/utils.py:645 |
revert_liger_kernel_to_olmo2 |
software | healthy | test/utils.py:656 |
revert_liger_kernel_to_olmo3 |
software | healthy | test/utils.py:668 |
revert_liger_kernel_to_glm4 |
software | healthy | test/utils.py:680 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
mkdocs.yml |
software | healthy | mkdocs.yml |
README.md |
software | healthy | README.md |
.pre-commit-config.yaml |
software | healthy | .pre-commit-config.yaml |
setup.py |
software | healthy | setup.py |
pyproject.toml |
software | healthy | pyproject.toml |
Makefile |
software | healthy | Makefile |
__init__.py |
software | healthy | test/__init__.py |
conftest.py |
software | healthy | test/conftest.py |
utils.py |
software | healthy | test/utils.py |
__init__.py |
software | healthy | test/convergence/__init__.py |
__init__.py |
software | healthy | test/convergence/bf16/__init__.py |
test_mini_models_with_logits.py |
software | healthy | test/convergence/bf16/test_mini_models_with_logits.py |
test_mini_models_multimodal.py |
software | healthy | test/convergence/bf16/test_mini_models_multimodal.py |
test_mini_models.py |
software | healthy | test/convergence/bf16/test_mini_models.py |
__init__.py |
software | healthy | test/convergence/fp32/__init__.py |
test_mini_models_with_logits.py |
software | healthy | test/convergence/fp32/test_mini_models_with_logits.py |
test_mini_models_multimodal.py |
software | healthy | test/convergence/fp32/test_mini_models_multimodal.py |
test_mini_models.py |
software | healthy | test/convergence/fp32/test_mini_models.py |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Qwen/Qwen2-VL-7B-Instruct/token… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Qwen/Qwen2.5-VL-7B-Instruct/tok… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Qwen/Qwen3-VL-4B-Instruct/token… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/OpenGVLab/InternVL3-1B-hf/token… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/HuggingFaceTB/SmolVLM2-256M-Vid… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/meta-llama/Llama-3.2-11B-Vision… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/meta-llama/Llama-4-Scout-17B-16… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Google/Gemma3/gemma-3-4b-it/tok… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Google/Paligemma/paligemma-3b-p… |
tokenizer_config.json |
software | healthy | test/resources/fake_configs/Llava/llava-1.5-7b-hf/tokenizer… |
preprocessor_config.json |
software | healthy | test/resources/fake_configs/Llava/llava-1.5-7b-hf/preproces… |
processor_config.json |
software | healthy | test/resources/fake_configs/Llava/llava-1.5-7b-hf/processor… |
generate_tokenized_dataset.py |
software | healthy | test/resources/scripts/generate_tokenized_dataset.py |
dataset_info.json |
software | healthy | test/resources/tiny_shakespeare_tokenized/dataset_info.json |
state.json |
software | healthy | test/resources/tiny_shakespeare_tokenized/state.json |
test_sparsemax.py |
software | healthy | test/transformers/test_sparsemax.py |
test_embedding.py |
software | healthy | test/transformers/test_embedding.py |
test_geglu.py |
software | healthy | test/transformers/test_geglu.py |
test_rms_norm.py |
software | healthy | test/transformers/test_rms_norm.py |
test_poly_norm.py |
software | healthy | test/transformers/test_poly_norm.py |
test_monkey_patch.py |
software | healthy | test/transformers/test_monkey_patch.py |
test_tiled_mlp.py |
software | healthy | test/transformers/test_tiled_mlp.py |
test_group_norm.py |
software | healthy | test/transformers/test_group_norm.py |
test_tvd.py |
software | healthy | test/transformers/test_tvd.py |
test_qwen2vl_mrope.py |
software | healthy | test/transformers/test_qwen2vl_mrope.py |
test_swiglu.py |
software | healthy | test/transformers/test_swiglu.py |
test_grpo_loss.py |
software | healthy | test/transformers/test_grpo_loss.py |
test_transformers.py |
software | healthy | test/transformers/test_transformers.py |
test_fused_neighborhood_attention.py |
software | healthy | test/transformers/test_fused_neighborhood_attention.py |
test_mm_int8int2.py |
software | healthy | test/transformers/test_mm_int8int2.py |
test_layer_norm.py |
software | healthy | test/transformers/test_layer_norm.py |
test_dyt.py |
software | healthy | test/transformers/test_dyt.py |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
MiniModelConfig |
software | healthy | test/utils.py:220 |
HFAlignmentLoss |
software | healthy | test/utils.py:882 |
HFDistillationLoss |
software | healthy | test/utils.py:1093 |
BaseRMSNorm |
software | healthy | test/transformers/test_rms_norm.py:36 |
LlamaRMSNorm |
software | healthy | test/transformers/test_rms_norm.py:57 |
GemmaRMSNorm |
software | healthy | test/transformers/test_rms_norm.py:82 |
NaivePolyNorm |
software | healthy | test/transformers/test_poly_norm.py:33 |
TorchTVDLoss |
software | healthy | test/transformers/test_tvd.py:10 |
_FalconH1MLPRef |
software | healthy | test/transformers/test_swiglu.py:498 |
_FakeConfig |
software | healthy | test/transformers/test_swiglu.py:551 |
TorchNeighborhoodAttention |
software | healthy | test/transformers/test_fused_neighborhood_attention.py:19 |
TorchNeighborhoodAttentionLayer |
software | healthy | test/transformers/test_fused_neighborhood_attention.py:184 |
TorchDyT |
software | healthy | test/transformers/test_dyt.py:24 |
TorchLMHeadCE |
software | healthy | test/transformers/test_fused_linear_cross_entropy.py:22 |
LigerLMHeadCE |
software | healthy | test/transformers/test_fused_linear_cross_entropy.py:70 |
CrossEntropyWithZLoss |
software | healthy | test/transformers/test_cross_entropy.py:22 |
TorchLMHeadJSD |
software | healthy | test/transformers/test_fused_linear_jsd.py:18 |
LigerLMHeadJSD |
software | healthy | test/transformers/test_fused_linear_jsd.py:52 |
TorchMultiTokenAttention |
software | healthy | test/transformers/test_multi_token_attention.py:22 |
TorchSparseMultiTokenAttention |
software | healthy | test/transformers/test_multi_token_attention.py:180 |
BaseAddRMSNorm |
software | healthy | test/transformers/test_fused_add_rms_norm.py:33 |
LlamaAddRMSNorm |
software | healthy | test/transformers/test_fused_add_rms_norm.py:51 |
GemmaAddRMSNorm |
software | healthy | test/transformers/test_fused_add_rms_norm.py:73 |
NPUKLDivLoss |
software | healthy | test/transformers/test_jsd.py:21 |
JSD |
software | healthy | test/transformers/test_jsd.py:46 |
TorchReLUSquared |
software | healthy | test/transformers/test_relu_squared.py:18 |
MiniMHCLM |
software | healthy | test/transformers/test_mhc.py:412 |
LigerLMHeadSimPO |
software | healthy | test/chunked_loss/test_simpo_loss.py:18 |
TorchLMHeadGRPO |
software | healthy | test/chunked_loss/test_grpo_loss.py:34 |
LigerLMHeadGRPO |
software | healthy | test/chunked_loss/test_grpo_loss.py:293 |
HFKTOLoss |
software | healthy | test/chunked_loss/test_kto_loss.py:19 |
TorchLMHeadKTO |
software | healthy | test/chunked_loss/test_kto_loss.py:89 |
LigerLMHeadKTO |
software | healthy | test/chunked_loss/test_kto_loss.py:124 |
HFCosineLoss |
software | healthy | test/chunked_loss/test_cosine_loss.py:17 |
TorchCosineLoss |
software | healthy | test/chunked_loss/test_cosine_loss.py:49 |
LigerCosineLoss |
software | healthy | test/chunked_loss/test_cosine_loss.py:94 |
HFDPOLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:19 |
HFAPOZeroLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:67 |
HFAPODownLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:117 |
HFSPPPOHARDLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:168 |
HFNCAPAIRLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:203 |
HFHingeLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:238 |
HFBCOPAIRLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:271 |
HFRobustLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:303 |
HFEXOPAIRLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:340 |
HFDiscoPOPLoss |
software | healthy | test/chunked_loss/test_dpo_loss.py:382 |
TorchLMHeadDPO |
software | healthy | test/chunked_loss/test_dpo_loss.py:420 |
TorchLMHeadAPOZero |
software | healthy | test/chunked_loss/test_dpo_loss.py:455 |
TorchLMHeadAPODown |
software | healthy | test/chunked_loss/test_dpo_loss.py:490 |
TorchLMHeadSPPOHARD |
software | healthy | test/chunked_loss/test_dpo_loss.py:525 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
test |
software | healthy | test |
convergence |
software | healthy | test/convergence |
bf16 |
software | healthy | test/convergence/bf16 |
fp32 |
software | healthy | test/convergence/fp32 |
resources |
software | healthy | test/resources |
fake_configs |
software | healthy | test/resources/fake_configs |
Qwen |
software | healthy | test/resources/fake_configs/Qwen |
Qwen2-VL-7B-Instruct |
software | healthy | test/resources/fake_configs/Qwen/Qwen2-VL-7B-Instruct |
Qwen2.5-VL-7B-Instruct |
software | healthy | test/resources/fake_configs/Qwen/Qwen2.5-VL-7B-Instruct |
Qwen3-VL-4B-Instruct |
software | healthy | test/resources/fake_configs/Qwen/Qwen3-VL-4B-Instruct |
OpenGVLab |
software | healthy | test/resources/fake_configs/OpenGVLab |
InternVL3-1B-hf |
software | healthy | test/resources/fake_configs/OpenGVLab/InternVL3-1B-hf |
HuggingFaceTB |
software | healthy | test/resources/fake_configs/HuggingFaceTB |
SmolVLM2-256M-Video-Instruct |
software | healthy | test/resources/fake_configs/HuggingFaceTB/SmolVLM2-256M-Vid… |
meta-llama |
software | healthy | test/resources/fake_configs/meta-llama |
Llama-3.2-11B-Vision-Instruct |
software | healthy | test/resources/fake_configs/meta-llama/Llama-3.2-11B-Vision… |
Llama-4-Scout-17B-16E-Instruct |
software | healthy | test/resources/fake_configs/meta-llama/Llama-4-Scout-17B-16… |
Google |
software | healthy | test/resources/fake_configs/Google |
Gemma3 |
software | healthy | test/resources/fake_configs/Google/Gemma3 |
gemma-3-4b-it |
software | healthy | test/resources/fake_configs/Google/Gemma3/gemma-3-4b-it |
Paligemma |
software | healthy | test/resources/fake_configs/Google/Paligemma |
paligemma-3b-pt-224 |
software | healthy | test/resources/fake_configs/Google/Paligemma/paligemma-3b-p… |
Llava |
software | healthy | test/resources/fake_configs/Llava |
llava-1.5-7b-hf |
software | healthy | test/resources/fake_configs/Llava/llava-1.5-7b-hf |
scripts |
software | healthy | test/resources/scripts |
tiny_shakespeare_tokenized |
software | healthy | test/resources/tiny_shakespeare_tokenized |
transformers |
software | healthy | test/transformers |
chunked_loss |
software | healthy | test/chunked_loss |
triton |
software | healthy | test/triton |
docs |
software | healthy | docs |
benchmark |
software | healthy | benchmark |
scripts |
software | healthy | benchmark/scripts |
dev |
software | healthy | dev |
modal |
software | healthy | dev/modal |
examples |
software | healthy | examples |
medusa |
software | healthy | examples/medusa |
scripts |
software | healthy | examples/medusa/scripts |
lightning |
software | healthy | examples/lightning |
huggingface |
software | healthy | examples/huggingface |
config |
software | healthy | examples/huggingface/config |
alignment |
software | healthy | examples/alignment |
.github |
software | healthy | .github |
ISSUE_TEMPLATE |
software | healthy | .github/ISSUE_TEMPLATE |
workflows |
software | healthy | .github/workflows |
src |
software | healthy | src |
liger_kernel |
software | healthy | src/liger_kernel |
ops |
software | healthy | src/liger_kernel/ops |
experimental |
software | healthy | src/liger_kernel/ops/experimental |
backends |
software | healthy | src/liger_kernel/ops/backends |
_ascend |
software | healthy | src/liger_kernel/ops/backends/_ascend |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
build |
cicd | healthy | .github/workflows/publish-release.yml |
build |
cicd | healthy | .github/workflows/publish-nightly.yml |
checkstyle |
cicd | healthy | .github/workflows/checkstyle.yml |
deploy |
cicd | healthy | .github/workflows/docs.yml |
handle-failure |
cicd | healthy | .github/workflows/merge-queue-failure.yml |
tests |
cicd | healthy | .github/workflows/intel-ci.yml |
tests |
cicd | healthy | .github/workflows/amd-ci.yml |
benchmarks |
cicd | healthy | .github/workflows/benchmark.yml |
nvi-correctness-tests |
cicd | healthy | .github/workflows/nvi-ci.yml |
nvi-convergence-tests |
cicd | healthy | .github/workflows/nvi-ci.yml |
nvi-correctness-tests-with-transformers-4-52-0 |
cicd | healthy | .github/workflows/nvi-ci.yml |
nvi-convergence-tests-with-transformers-4-52-0 |
cicd | healthy | .github/workflows/nvi-ci.yml |
| Label | Layer | Status | Path |
|---|---|---|---|
gha::publish-release |
cicd | healthy | .github/workflows/publish-release.yml |
gha::publish-nightly |
cicd | healthy | .github/workflows/publish-nightly.yml |
gha::checkstyle |
cicd | healthy | .github/workflows/checkstyle.yml |
gha::docs |
cicd | healthy | .github/workflows/docs.yml |
gha::merge-queue-failure |
cicd | healthy | .github/workflows/merge-queue-failure.yml |
gha::intel-ci |
cicd | healthy | .github/workflows/intel-ci.yml |
gha::amd-ci |
cicd | healthy | .github/workflows/amd-ci.yml |
gha::benchmark |
cicd | healthy | .github/workflows/benchmark.yml |
gha::nvi-ci |
cicd | healthy | .github/workflows/nvi-ci.yml |
| Label | Layer | Status | Path |
|---|---|---|---|
MODAL_TOKEN_SECRET |
cicd | healthy | — |
PYPI_PASSWORD |
cicd | healthy | — |
PYPI_USERNAME |
cicd | healthy | — |
MODAL_TOKEN_ID |
cicd | healthy | — |
PYPI_NIGHTLY_PASSWORD |
cicd | healthy | — |
| Label | Layer | Status | Path |
|---|---|---|---|
repobility-clone-1mhwebrt |
software | healthy | /tmp/repobility-clone-1mhwebrt |
| Label | Layer | Status | Path |
|---|---|---|---|
gpu (detected) |
hardware | healthy | setup.py |
This page is publicly accessible at:
https://repobility.com/scan/fedca033-f0ab-4858-8e75-ec394950c9b6/
To check status programmatically (no auth required):
curl -s https://repobility.com/api/v1/public/scan/fedca033-f0ab-4858-8e75-ec394950c9b6/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.