A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks like TensorRT-LLM, TensorRT, vLLM, etc. to optimize inference speed.
This report summarizes scanner evidence for NVIDIA/Model-Optimizer, a Python project with 5,267 GitHub stars. SynthScan examined 464,726 lines of code across 2199 source files, recording 10398 pattern matches distributed across 24 syntactic categories. The legacy adjusted pattern signal is 28.2 (Elevated pattern signal); it is not an authorship estimate.
Use the findings to ask whether maintainers can explain intent, verify dependencies, simplify unnecessary scaffolding, and focus review effort. Risk Payload v1 adds ownership, supply-chain, over-engineering, and review-effort vectors when available. These are prompts for human verification, not probabilities, security findings, or measured maintenance costs.
Use each vector to guide manual verification of the underlying findings.
Can maintainers explain intent, assumptions, and trade-offs? 8302 finding(s).
Are imports declared and package/API names verified? 683 finding(s).
Are wrappers, guards, and repeated structures necessary? 6460 finding(s).
Which changes need focused human review? 10398 finding(s).
This chart tracks the legacy adjusted pattern score across scan runs. Changes can reflect source changes, scan coverage, or detector/scoring versions; they do not measure AI adoption or prove that a team refactored code. Compare runs only when their scoring versions are compatible. The dashed line tracks the raw pattern-hit count.
Groups findings by configured severity weight. Severity is not a confidence estimate, security rating, or proof that a package is missing. Verify imports against the project's manifests and inspect each finding in its local design context.
This horizontal bar chart shows where configured pattern scores accumulate across directories. Treat the bars as locations to inspect, not as evidence of authorship or code quality. Use the file-level findings to decide whether a directory needs a focused ownership, dependency, complexity, or review check.
The scanner identified 10398 distinct pattern matches across 24 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a detector match to inspect in context. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.
Reading the findings table: The Severity column indicates the configured finding weight (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — legacy comment-context matches receive a ×1.5 score weight. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier under the legacy scoring model.
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | experimental/dms/tests/test_chunked_prefill.py | 62 | def test_chunked_matches_unchunked(self, decoder_layers, seed): | CODE |
| LOW | experimental/dms/tests/test_dms_utils.py | 44 | def test_decisions_are_binary(self): | CODE |
| LOW | experimental/dms/tests/test_dms_utils.py | 55 | def test_probs_in_unit_interval(self): | CODE |
| LOW | experimental/dms/tests/test_dms_utils.py | 78 | def test_gradient_flows_through_decisions(self): | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 38 | def fake_flash_attn_with_kvcache( | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 111 | def _simple_code_for_dms_exact_attention(q, k, v, d, a, state, attn_scaling, window_size): | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 169 | def _simple_code_for_dms_fast_attention_inference(q, k, v, d, a, attn_scaling, window_size): | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 198 | def _generate_random_test_params(seed): | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 305 | def test_prefill_matches_reference(self, seed): | CODE |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 317 | def test_generate_after_prefill_matches_reference(self, seed): | CODE |
| LOW⚡ | experimental/dms/tests/test_paged_cache.py | 194 | def test_fast_update_matches_regular_update(self, seed): | CODE |
| LOW⚡ | experimental/dms/tests/test_paged_cache.py | 255 | def _run_paged_cache_update_test(seed, disable_eviction): | CODE |
| LOW | experimental/dms/tests/test_paged_cache.py | 410 | def test_update_with_eviction(self, seed): | CODE |
| LOW | experimental/dms/tests/test_paged_cache.py | 415 | def test_update_without_eviction(self, seed): | CODE |
| LOW⚡ | experimental/dms/models/qwen3/train.py | 105 | def _parse_data_blend_elements(blend_string: str) -> list[DataBlendElement]: | CODE |
| LOW⚡ | experimental/dms/dms/attention_prefill.py | 34 | def rewrite_cache_in_left_padding_style( | CODE |
| LOW | experimental/dms/dms/attention_prefill.py | 73 | def _rewrite_cache_in_left_padding_style_aux( | CODE |
| LOW⚡ | experimental/dms/dms/core.py | 37 | def setup_compile_limit_for_dms(compile_limit: int = 72): | CODE |
| LOW | experimental/dms/dms/core.py | 225 | def post_process_attention_output( | CODE |
| LOW | experimental/dms/dms/core.py | 287 | def dms_perform_chunked_prefill( | CODE |
| LOW | experimental/dms/dms/cache_paged.py | 220 | def _aux_update_many_handle_single_chunk( | CODE |
| LOW | experimental/dms/dms/cache_paged.py | 786 | def fast_update_ignore_eviction( | CODE |
| LOW | experimental/dms/dms/training/engine.py | 666 | def get_parameters_to_optimize(self): | CODE |
| LOW | experimental/dms/dms/training/engine.py | 673 | def gradient_checkpointing_enable(self, gradient_checkpointing_kwargs=None): | CODE |
| LOW⚡ | experimental/dms/dms/training/data.py | 63 | def _load_or_create_cached_dataset( | CODE |
| LOW | tools/launcher/core.py | 189 | def _explicit_slurm_fields_from_yaml(yaml_path: str | None, task_name: str) -> set[str] | None: | CODE |
| LOW | tools/launcher/core.py | 425 | def _inject_controlmaster_rsh(self, command: str) -> str: | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 48 | def test_missing_factory_raises(self, tmp_yaml): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 59 | def test_missing_slurm_config_raises(self, tmp_yaml): | STRING |
| LOW | tools/launcher/tests/test_core_extended.py | 67 | def test_environment_preserved(self, tmp_yaml): | STRING |
| LOW | tools/launcher/tests/test_core_extended.py | 88 | def test_dict_environment_interpolation(self): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 114 | def test_no_global_vars_no_error(self): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 124 | def test_slurm_repair_preserves_explicit_nullable_gpus(self, monkeypatch): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 146 | def test_slurm_repair_uses_raw_yaml_keys_for_default_equal_values(self, monkeypatch, tmp_yaml): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 191 | def test_valid_git_repo_from_nested_directory(self): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 197 | def test_nonexistent_directory(self): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 216 | def test_environment_list_merged_to_env(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 254 | def test_none_env_values_converted_to_empty_string(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 288 | def test_test_level_filters_pipeline(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 320 | def test_skipped_pipeline_not_run(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 342 | def test_detach_flag_passed_to_experiment(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 372 | def test_version_report_called(self, mock_docker, mock_exp, tmp_path, capsys): | CODE |
| LOW | tools/launcher/tests/test_core_extended.py | 450 | def test_reqs_script_wrapped_inline(self, mock_docker, mock_exp, tmp_path): | CODE |
| LOW⚡ | tools/launcher/tests/test_docker_launch.py | 37 | def test_echo_script_via_launch(self, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_docker_launch.py | 83 | def test_failing_script_via_launch(self, tmp_path): | STRING |
| LOW | tools/launcher/tests/test_core.py | 76 | def test_task_slots_collected(self): | CODE |
| LOW | tools/launcher/tests/test_core.py | 88 | def test_global_vars_interpolation_in_environment(self): | CODE |
| LOW | tools/launcher/tests/test_core.py | 99 | def test_global_vars_interpolation_in_args(self): | CODE |
| LOW | tools/launcher/tests/test_core.py | 110 | def test_global_vars_unresolved_passthrough(self): | CODE |
| LOW | tools/launcher/tests/test_core.py | 122 | def test_skip_and_allow_to_fail(self): | CODE |
| LOW | tools/launcher/tests/test_core.py | 141 | def test_create_task_from_yaml_uses_registry(self, tmp_yaml): | CODE |
| LOW | tools/launcher/tests/test_core.py | 161 | def test_task_configs_resolved_via_registry(self, tmp_yaml): | STRING |
| LOW | tools/launcher/tests/test_core.py | 237 | def test_runs_on_nonexistent_dir(self, capsys): | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 35 | def test_controlmaster_env_selects_controlmaster_tunnel( | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 77 | def test_connect_reuses_live_controlmaster(self, monkeypatch, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 99 | def test_connect_requires_reauth_when_controlmaster_missing(self, monkeypatch, tmp_path): | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 113 | def test_connect_requires_reauth_when_controlmaster_check_times_out( | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 139 | def test_installed_mode_skips_modelopt_source_mounts(self, mock_tunnel, mock_executor): | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 185 | def test_source_mode_adds_modelopt_source_mounts(self, mock_tunnel, mock_executor): | CODE |
| LOW | tools/launcher/tests/test_slurm_executor.py | 230 | def test_scratch_path_uses_experiment_title(self, mock_tunnel, mock_executor): | CODE |
| 4854 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | pyproject.toml | 161 | [tool.setuptools.exclude-package-data] | COMMENT |
| LOW | noxfile.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/conftest.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/test_chunked_prefill.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/test_dms_utils.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/utils.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/test_prefill_and_generate.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/tests/test_paged_cache.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/models/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | …perimental/dms/models/qwen3/configuration_qwen3_dms.py | 1 | # Adapted from https://github.com/huggingface/transformers/blob/47b0e478f324b54f177ea7998a0791870fdd0324/src/transformer | COMMENT |
| LOW | experimental/dms/models/qwen3/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/models/qwen3/train.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/models/qwen3/extract.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/models/qwen3/modeling_qwen3_dms.py | 1 | # Adapted from https://github.com/huggingface/transformers/blob/47b0e478f324b54f177ea7998a0791870fdd0324/src/transformer | COMMENT |
| LOW | experimental/dms/scripts/train.sh | 1 | #!/bin/bash | COMMENT |
| LOW | experimental/dms/scripts/evaluate.sh | 1 | #!/bin/bash | COMMENT |
| LOW | experimental/dms/scripts/train_small.sh | 1 | #!/bin/bash | COMMENT |
| LOW | experimental/dms/dms/attention_prefill.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/logging.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/attention.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/cache.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/core.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/cache_paged.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/training/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/training/engine.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | experimental/dms/dms/training/data.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/resource_monitor.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/slurm_config.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/core.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/launch.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/conftest.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_core_extended.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_docker_launch.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_core.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/__init__.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_slurm_executor.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_docker_execution.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_slurm_config.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_yaml_formats.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/tests/test_examples_resolve.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/common/query.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/common/check_regression.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/common/service_utils.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/smoke/hostname.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/smoke/nvidia_smi.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/megatron_bridge/import/import.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/megatron_lm/quantize/task.py | 1 | # SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | COMMENT |
| LOW | tools/launcher/common/megatron_lm/quantize/quantize.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/megatron_lm/train/sft.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/megatron_lm/train/sft.sh | 21 | util_install_extra_dep | COMMENT |
| LOW | tools/launcher/common/megatron_lm/export/export.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/megatron_lm/export/export.sh | 21 | util_install_extra_dep | COMMENT |
| LOW | tools/launcher/common/eagle3/dump_offline_data_vllm.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/eagle3/dump_offline_data_vllm.sh | 21 | COMMENT | |
| LOW | tools/launcher/common/eagle3/train_eagle_streaming.sh | 1 | #!/bin/bash | COMMENT |
| LOW | tools/launcher/common/eagle3/train_eagle_streaming.sh | 21 | # | COMMENT |
| LOW | tools/launcher/common/eagle3/train_eagle_streaming.sh | 41 | # | COMMENT |
| 1861 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | noxfile.py | 59 | # ─── CPU unit tests ─────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | noxfile.py | 98 | # ─── GPU sessions (run inside containers — no new venv) ────────────────────── | COMMENT |
| MEDIUM | noxfile.py | 165 | # ─── Code quality ───────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | noxfile.py | 178 | # ─── Docs ───────────────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | noxfile.py | 206 | # ─── Wheel build ────────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | experimental/dms/tests/test_paged_cache.py | 29 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | experimental/dms/tests/test_paged_cache.py | 31 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | experimental/dms/tests/test_paged_cache.py | 185 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | experimental/dms/tests/test_paged_cache.py | 187 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | experimental/dms/tests/test_paged_cache.py | 250 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | experimental/dms/tests/test_paged_cache.py | 252 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 62 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 64 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/models/qwen3/train.py | 100 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/models/qwen3/train.py | 102 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 144 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 146 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 202 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 204 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 256 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/models/qwen3/train.py | 258 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/attention_prefill.py | 29 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/attention_prefill.py | 31 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention_prefill.py | 142 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention_prefill.py | 144 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 45 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 47 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 98 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 100 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 176 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/attention.py | 178 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 31 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 33 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 154 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 156 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 412 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/cache.py | 414 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/core.py | 32 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/core.py | 34 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/core.py | 46 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/core.py | 48 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 76 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 78 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 114 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 116 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 256 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/core.py | 258 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 41 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 43 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 186 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 188 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 249 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 251 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 342 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 344 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 431 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 433 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 576 | # ============================================================================= | COMMENT |
| MEDIUM | experimental/dms/dms/training/engine.py | 578 | # ============================================================================= | COMMENT |
| MEDIUM⚡ | experimental/dms/dms/training/data.py | 53 | # ============================================================================= | COMMENT |
| 467 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tools/ci/check_model_hub_orgs.py | 35 | CODE | |
| LOW | tools/mcp/tests/test_bridge.py | 18 | CODE | |
| LOW | tools/mcp/modelopt_mcp/server.py | 33 | CODE | |
| LOW | tools/mcp/modelopt_mcp/bridge.py | 34 | CODE | |
| LOW | tools/mcp/modelopt_mcp/__init__.py | 51 | CODE | |
| LOW | tools/precommit/check_modelopt_recipes.py | 33 | CODE | |
| LOW | tools/precommit/normalize_yaml_ext.py | 23 | CODE | |
| LOW | tools/precommit/check_launcher_yaml.py | 36 | CODE | |
| LOW | …s/modelopt/skills/day0-release/scripts/gate_compare.py | 31 | CODE | |
| LOW | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 35 | CODE | |
| LOW | …ugins/modelopt/skills/day0-release/scripts/gate_run.py | 45 | CODE | |
| LOW | …ugins/modelopt/skills/day0-release/scripts/gate_ptq.py | 61 | CODE | |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 26 | CODE | |
| LOW | modelopt/onnx/__init__.py | 24 | CODE | |
| LOW | modelopt/onnx/__init__.py | 25 | CODE | |
| LOW | modelopt/onnx/autocast/__init__.py | 18 | CODE | |
| LOW | modelopt/onnx/autocast/__init__.py | 18 | CODE | |
| LOW | modelopt/onnx/autocast/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/quantization/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/quantization/__init__.py | 20 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 24 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 24 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 24 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 28 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 29 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 29 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 30 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 41 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 41 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 41 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 41 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 47 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 48 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 18 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 18 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 19 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 37 | CODE | |
| 604 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | modelopt/torch/sparsity/attention_sparsity/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/sparsity/weight_sparsity/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/nas/conversion.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/distill/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/peft/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/quantization/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/prune/plugins/mcore_minitron.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/speculative/mode.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | tests/unit/torch/opt/test_mode_registry.py | 0 | returns the value (str representation) of the mode. | STRING |
| HIGH | modelopt/torch/sparsity/attention_sparsity/mode.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/sparsity/weight_sparsity/mode.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/distill/mode.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/quantization/mode.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/prune/plugins/mcore_minitron.py | 0 | the mode that corresponds to the export mode of this mode. | STRING |
| HIGH | modelopt/torch/sparsity/weight_sparsity/mode.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/nas/conversion.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/distill/mode.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/quantization/mode.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/prune/plugins/mcore_minitron.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/speculative/mode.py | 0 | class to describe the layerwise knowledge-distillation mode. the properties of this mode can be inspected via the source | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | specifies the search algorithm to use for this mode (if any). | STRING |
| HIGH | modelopt/torch/opt/mode.py | 0 | specifies the search algorithm to use for this mode (if any). | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | specifies the search algorithm to use for this mode (if any). | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | the mode's entrypoint for updating the models state before saving. | STRING |
| HIGH | modelopt/torch/quantization/mode.py | 0 | the mode's entrypoint for updating the models state before saving. | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | the mode's entrypoint for updating the models state before saving. | STRING |
| HIGH | modelopt/torch/nas/autonas.py | 0 | the mode's entrypoint for updating the models state before new mode. | STRING |
| HIGH | modelopt/torch/peft/mode.py | 0 | the mode's entrypoint for updating the models state before new mode. | STRING |
| HIGH | modelopt/torch/quantization/mode.py | 0 | the mode's entrypoint for updating the models state before new mode. | STRING |
| HIGH | modelopt/torch/prune/fastnas.py | 0 | the mode's entrypoint for updating the models state before new mode. | STRING |
| HIGH | modelopt/torch/nas/hparams/concat.py | 0 | set the active value with a sanity check for choices and dynamic hparams. | STRING |
| HIGH | modelopt/torch/quantization/algorithms.py | 0 | set the active value with a sanity check for choices and dynamic hparams. | STRING |
| HIGH | modelopt/torch/opt/hparam.py | 0 | set the active value with a sanity check for choices and dynamic hparams. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq1_s.py | 0 | pack a floating-point weight into ggml-compatible iq2_xxs blocks. returned shapes are ``[*weight.shape[:-1], weight.shap | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xs.py | 0 | pack a floating-point weight into ggml-compatible iq2_xxs blocks. returned shapes are ``[*weight.shape[:-1], weight.shap | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xxs.py | 0 | pack a floating-point weight into ggml-compatible iq2_xxs blocks. returned shapes are ``[*weight.shape[:-1], weight.shap | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xs.py | 0 | predict one fp16 super-block scale for each flattened block. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xxs.py | 0 | predict one fp16 super-block scale for each flattened block. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq1_m.py | 0 | predict one fp16 super-block scale for each flattened block. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_s.py | 0 | predict one fp16 super-block scale for each flattened block. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xs.py | 0 | encode a moderate-size batch of flattened 256-value blocks, optionally error-weighted. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_xxs.py | 0 | encode a moderate-size batch of flattened 256-value blocks, optionally error-weighted. | STRING |
| HIGH | modelopt/torch/quantization/ggml/iq2_s.py | 0 | encode a moderate-size batch of flattened 256-value blocks, optionally error-weighted. | STRING |
| HIGH | modelopt/torch/quantization/qtensor/nf4_tensor.py | 0 | implements the int4 quantization on tensors for more efficient storage or computation. attributes: quantized_data (torch | STRING |
| HIGH | modelopt/torch/quantization/qtensor/nvfp4_tensor.py | 0 | implements the int4 quantization on tensors for more efficient storage or computation. attributes: quantized_data (torch | STRING |
| HIGH | modelopt/torch/quantization/qtensor/mxfp4_tensor.py | 0 | implements the int4 quantization on tensors for more efficient storage or computation. attributes: quantized_data (torch | STRING |
| HIGH | modelopt/torch/quantization/qtensor/fp8_tensor.py | 0 | implements the int4 quantization on tensors for more efficient storage or computation. attributes: quantized_data (torch | STRING |
| HIGH | modelopt/torch/quantization/qtensor/int4_tensor.py | 0 | implements the int4 quantization on tensors for more efficient storage or computation. attributes: quantized_data (torch | STRING |
| HIGH | tests/unit/torch/quantization/test_iq_formats.py | 0 | float64 weights are accepted, so a finite value too large for float32 must saturate. converting before sanitizing would | STRING |
| HIGH | tests/unit/torch/quantization/test_iq1_s.py | 0 | float64 weights are accepted, so a finite value too large for float32 must saturate. converting before sanitizing would | STRING |
| HIGH | tests/unit/torch/quantization/test_iq2_xs.py | 0 | float64 weights are accepted, so a finite value too large for float32 must saturate. converting before sanitizing would | STRING |
| HIGH | …ts/unit/torch/quantization/test_layerwise_calibrate.py | 0 | decoder layer that returns a tuple, mimicking huggingface decoder layers. | STRING |
| HIGH | …sts/gpu/torch/quantization/test_layerwise_calibrate.py | 0 | decoder layer that returns a tuple, mimicking huggingface decoder layers. | STRING |
| HIGH | …/gpu/torch/quantization/plugins/test_accelerate_gpu.py | 0 | decoder layer that returns a tuple, mimicking huggingface decoder layers. | STRING |
| 68 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | experimental/dms/tests/test_paged_cache.py | 255 | CODE | |
| LOW | experimental/dms/dms/core.py | 287 | CODE | |
| LOW | tools/launcher/core.py | 647 | CODE | |
| LOW | tools/launcher/core.py | 737 | CODE | |
| LOW | tools/launcher/core.py | 258 | CODE | |
| LOW | tools/launcher/common/query.py | 130 | CODE | |
| LOW | tools/mcp/modelopt_mcp/bridge.py | 598 | CODE | |
| LOW | tools/precommit/check_modelopt_recipes.py | 103 | CODE | |
| LOW | tools/precommit/check_modelopt_recipes.py | 184 | CODE | |
| LOW | tools/precommit/check_modelopt_recipes.py | 274 | CODE | |
| LOW | …ugins/modelopt/skills/day0-release/scripts/gate_ptq.py | 119 | CODE | |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 326 | CODE | |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 638 | CODE | |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 746 | CODE | |
| LOW | …lls/benchmark-model-kernels/scripts/benchmark_model.py | 131 | CODE | |
| LOW | …lls/benchmark-model-kernels/scripts/benchmark_model.py | 534 | CODE | |
| LOW | modelopt/onnx/trt_utils.py | 214 | CODE | |
| LOW | modelopt/onnx/trt_utils.py | 341 | CODE | |
| LOW | modelopt/onnx/trt_utils.py | 470 | CODE | |
| LOW | modelopt/onnx/utils.py | 319 | CODE | |
| LOW | modelopt/onnx/utils.py | 424 | CODE | |
| LOW | modelopt/onnx/utils.py | 876 | CODE | |
| LOW | modelopt/onnx/utils.py | 1184 | CODE | |
| LOW | modelopt/onnx/utils.py | 1348 | CODE | |
| LOW | modelopt/onnx/utils.py | 1369 | CODE | |
| LOW | modelopt/onnx/utils.py | 1452 | CODE | |
| LOW | modelopt/onnx/utils.py | 1597 | CODE | |
| LOW | modelopt/onnx/utils.py | 1864 | CODE | |
| LOW | modelopt/onnx/utils.py | 2107 | CODE | |
| LOW | modelopt/onnx/utils.py | 897 | CODE | |
| LOW | modelopt/onnx/utils.py | 2116 | CODE | |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 566 | CODE | |
| LOW | modelopt/onnx/autocast/utils.py | 84 | CODE | |
| LOW | modelopt/onnx/autocast/referencerunner.py | 160 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 313 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 649 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 698 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 859 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 950 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1304 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1363 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1510 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1557 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1632 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1739 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1778 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 433 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 469 | CODE | |
| LOW | modelopt/onnx/autocast/nodeclassifier.py | 303 | CODE | |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 831 | CODE | |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1054 | CODE | |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1112 | CODE | |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1281 | CODE | |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1445 | CODE | |
| LOW | modelopt/onnx/quantization/ort_calibration_per_node.py | 148 | CODE | |
| LOW | modelopt/onnx/quantization/ort_calibration_per_node.py | 311 | CODE | |
| LOW | modelopt/onnx/quantization/ort_calibration_per_node.py | 621 | CODE | |
| LOW | modelopt/onnx/quantization/ort_calibration_per_node.py | 798 | CODE | |
| LOW | modelopt/onnx/quantization/ort_utils.py | 157 | CODE | |
| LOW | modelopt/onnx/quantization/ort_utils.py | 233 | CODE | |
| 486 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | experimental/__init__.py | 35 | __all__ = [] | CODE |
| LOW | experimental/dms/models/qwen3/modeling_qwen3_dms.py | 491 | __all__ = [ | CODE |
| LOW | tools/launcher/slurm_config.py | 27 | __all__ = ["SlurmConfig", "slurm_factory"] | CODE |
| LOW | tools/launcher/__init__.py | 20 | __all__ = ["PACKAGE_DIR"] | CODE |
| LOW | tools/mcp/modelopt_mcp/__init__.py | 53 | __all__ = ["main"] | CODE |
| LOW⚡ | modelopt/onnx/autocast/graphsanitizer.py | 206 | def set_ir_version(self, max_ir_version: int | None) -> None: | CODE |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 493 | def _update_opset_version(self) -> None: | CODE |
| LOW | modelopt/onnx/quantization/ort_patches.py | 18 | __all__ = [] | CODE |
| LOW | modelopt/onnx/quantization/ort_calibration_per_node.py | 42 | __all__ = [] | CODE |
| LOW | modelopt/onnx/quantization/quantize.py | 86 | __all__ = ["quantize"] | CODE |
| LOW | modelopt/onnx/quantization/graph_indexing.py | 31 | __all__ = [ | CODE |
| LOW | modelopt/onnx/quantization/int4.py | 61 | __all__ = ["quantize"] | CODE |
| LOW | modelopt/onnx/quantization/graph_rewrites.py | 30 | __all__ = [ | CODE |
| LOW | modelopt/onnx/quantization/graph_selection.py | 37 | __all__ = [ | CODE |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 42 | __all__ = [ | CODE |
| LOW | modelopt/onnx/quantization/ort_calibration.py | 42 | __all__ = [] | CODE |
| LOW | modelopt/onnx/quantization/__main__.py | 31 | __all__ = ["main"] | CODE |
| LOW | modelopt/onnx/quantization/ort_quantization.py | 45 | __all__ = [] | CODE |
| LOW | modelopt/onnx/quantization/ort_session.py | 43 | __all__ = [] | CODE |
| LOW | modelopt/onnx/quantization/autotune/export_utils.py | 34 | __all__ = [ | CODE |
| LOW | modelopt/onnx/quantization/autotune/__init__.py | 56 | __all__ = ["MODE_PRESETS", "StoreWithExplicitFlag", "get_node_filter_list"] | CODE |
| LOW | modelopt/onnx/quantization/autotune/autotuner_base.py | 227 | def set_profile_region(self, region: Region | None, commit: bool = True) -> None: | CODE |
| LOW | modelopt/onnx/graph_surgery/__init__.py | 131 | __all__ = [ | CODE |
| LOW | modelopt/onnx/graph_surgery/utils/__init__.py | 47 | __all__ = [ | CODE |
| LOW | modelopt/onnx/export/__init__.py | 18 | __all__ = [ | CODE |
| LOW | modelopt/recipe/config.py | 35 | __all__ = [ | CODE |
| LOW | modelopt/recipe/presets.py | 42 | __all__ = [ | CODE |
| LOW | modelopt/recipe/loader.py | 44 | __all__ = ["load_config", "load_recipe"] | CODE |
| LOW | …pt/torch/sparsity/attention_sparsity/model_sparsify.py | 29 | __all__ = [ | CODE |
| LOW | modelopt/torch/sparsity/attention_sparsity/config.py | 776 | __all__ = [ | CODE |
| LOW | …delopt/torch/sparsity/attention_sparsity/conversion.py | 65 | def _set_attn_implementation(model: nn.Module, config: SparseAttentionConfig) -> None: | CODE |
| LOW | …/torch/sparsity/attention_sparsity/methods/registry.py | 43 | def set_calibration_mode(self, enabled: bool) -> None: | CODE |
| LOW | …/torch/sparsity/attention_sparsity/methods/__init__.py | 20 | __all__ = [ | CODE |
| LOW | …ch/sparsity/attention_sparsity/calibration/__init__.py | 22 | __all__ = [ | CODE |
| LOW | …arsity/attention_sparsity/calibration/ruler_dataset.py | 58 | logger = logging.getLogger(__name__) | CODE |
| LOW | …lopt/torch/sparsity/attention_sparsity/plugins/vllm.py | 55 | __all__ = [ | CODE |
| LOW | …ch/sparsity/attention_sparsity/plugins/vllm_runtime.py | 33 | __all__ = [ | CODE |
| LOW | …/torch/sparsity/attention_sparsity/plugins/__init__.py | 31 | __all__ = [ | CODE |
| LOW | …/attention_sparsity/plugins/sparse_attn_calibration.py | 44 | __all__ = [ | CODE |
| LOW | …elopt/torch/sparsity/weight_sparsity/sparsification.py | 29 | __all__ = ["export", "sparsify"] | CODE |
| LOW | modelopt/torch/sparsity/weight_sparsity/sparsegpt.py | 214 | def _setup_forward_hook(cls, mod: SparseModule) -> None: | CODE |
| LOW | modelopt/torch/sparsity/weight_sparsity/module.py | 25 | __all__ = ["SpDMRegistry", "SparseModule"] | CODE |
| LOW⚡ | modelopt/torch/kernels/sparsity/attention/__init__.py | 46 | def set_skip_softmax_context(active: bool) -> None: | CODE |
| LOW⚡ | modelopt/torch/kernels/sparsity/attention/__init__.py | 56 | __all__ = [ | CODE |
| LOW | …elopt/torch/kernels/quantization/attention/bmm2_qdq.py | 32 | __all__ = ["fake_quant_v_onwrite"] | CODE |
| LOW | modelopt/torch/kernels/quantization/gemm/fp4_kernel.py | 31 | __all__ = [ | CODE |
| LOW | …opt/torch/kernels/quantization/gemm/nvfp4_fp8_sweep.py | 39 | __all__ = [ | CODE |
| LOW | …t/torch/kernels/quantization/gemm/fp4_kernel_hopper.py | 29 | __all__ = ["fp4_fake_quant_block"] | CODE |
| LOW⚡ | …t/torch/kernels/quantization/gemm/gptq_fused_kernel.py | 35 | __all__ = ["gptq_fused_block_scalar"] | CODE |
| LOW | …opt/torch/kernels/common/attention/decode_attention.py | 37 | __all__ = ["attention_decode"] | CODE |
| LOW | modelopt/torch/kernels/common/attention/__init__.py | 47 | __all__ = [ | CODE |
| LOW | modelopt/torch/kernels/common/attention/triton_fa.py | 1465 | __all__ = ["LOG2E", "_apply_mask", "attention"] | CODE |
| LOW | …/torch/kernels/common/attention/hf_triton_attention.py | 302 | __all__ = [ | CODE |
| LOW | modelopt/torch/trace/analyzer.py | 42 | __all__ = ["analyze_symbols"] | CODE |
| LOW | modelopt/torch/trace/symbols.py | 26 | __all__ = ["SymInfo", "SymMap", "Symbol"] | CODE |
| LOW | modelopt/torch/trace/symbols.py | 507 | def set_symbol(self, mod: nn.Module, name: str, symbol: Symbol) -> None: | CODE |
| LOW | modelopt/torch/trace/tracer.py | 30 | __all__ = ["GraphCollection", "RobustTracer", "recursive_trace"] | CODE |
| LOW | modelopt/torch/trace/modules/concat.py | 43 | __all__ = ["ConcatNodeProcessor", "ConcatSymbol"] | CODE |
| LOW | modelopt/torch/trace/modules/nn.py | 24 | __all__ = ["SymDepth"] | CODE |
| LOW | modelopt/torch/trace/modules/nn.py | 89 | def set_skippable(self, idx: int, val: bool) -> None: | CODE |
| 321 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | experimental/dms/models/qwen3/modeling_qwen3_dms.py | 323 | # Create the masks | COMMENT |
| MEDIUM⚡ | tools/launcher/tests/test_docker_launch.py | 39 | # Create a trivial script | COMMENT |
| MEDIUM⚡ | tools/launcher/tests/test_docker_launch.py | 46 | # Create a YAML config | COMMENT |
| MEDIUM | tools/launcher/common/megatron_lm/quantize/quantize.sh | 82 | # This function handles the exit status (fails the CI). | COMMENT |
| MEDIUM | tools/launcher/common/eagle3/train_eagle.sh | 38 | # This function handles the exit status (fails the CI). | COMMENT |
| MEDIUM⚡ | plugins/modelopt/skills/common/remote_exec.sh | 510 | # Create the remote workspace directory if it doesn't exist | COMMENT |
| MEDIUM | modelopt/onnx/logging_config.py | 22 | # Create a parent logger for all ONNX components | COMMENT |
| MEDIUM | modelopt/onnx/utils.py | 908 | # Create a temporary model with just this graph for graphsurgeon | COMMENT |
| MEDIUM | modelopt/onnx/autocast/logging_config.py | 27 | # Create a logger for all AutoCast components as a child of modelopt.onnx | COMMENT |
| MEDIUM | modelopt/onnx/autocast/__main__.py | 33 | """This module provides the command line interface (CLI) entry point for AutoCast.""" | STRING |
| MEDIUM | modelopt/onnx/quantization/qdq_utils.py | 1389 | # Create a new scale tensor | COMMENT |
| MEDIUM | modelopt/onnx/quantization/qdq_utils.py | 1505 | # Create a new scale tensor | COMMENT |
| MEDIUM | modelopt/onnx/quantization/ort_calibration_per_node.py | 120 | """This function is an auxiliary function of collect_data to solve the OOM issue in the MinMax Calibrator. | STRING |
| MEDIUM | modelopt/onnx/quantization/ort_calibration_per_node.py | 773 | # Create a new onnx model | COMMENT |
| MEDIUM | modelopt/onnx/quantization/ort_calibration_per_node.py | 882 | # Create a new onnx model | COMMENT |
| MEDIUM | modelopt/onnx/quantization/fp8.py | 270 | # Create a temp file for intermediate model | COMMENT |
| MEDIUM | modelopt/onnx/quantization/quantize.py | 16 | """This module is used to convert an ONNX model (w/o QDQ nodes) and calibration data into an ONNX model with QDQ nodes. | STRING |
| MEDIUM | modelopt/onnx/quantization/graph_indexing.py | 241 | # Create a dictionary to store tensor producer nodes | COMMENT |
| MEDIUM | modelopt/onnx/quantization/graph_indexing.py | 284 | # Create a dictionary to store tensor consumer nodes | COMMENT |
| MEDIUM | modelopt/onnx/quantization/int4.py | 549 | # Create a temp file for augmented model | COMMENT |
| MEDIUM | modelopt/onnx/quantization/int4.py | 1109 | # Create a temp file for augmented model | COMMENT |
| MEDIUM | modelopt/onnx/quantization/ort_calibration.py | 419 | """This function is an auxiliary function of collect_data to solve the OOM issue in the MinMax Calibrator. | STRING |
| MEDIUM | modelopt/onnx/quantization/int8.py | 272 | # Create a temp file for intermediate model | COMMENT |
| MEDIUM | modelopt/onnx/export/nvfp4_exporter.py | 362 | # Create a Cast node to convert the input to FP16/BF16 | COMMENT |
| MEDIUM | modelopt/onnx/export/int4_exporter.py | 182 | # Create a new scale tensor | COMMENT |
| MEDIUM | …arsity/attention_sparsity/calibration/ruler_dataset.py | 972 | # Create a simple QA pair | COMMENT |
| MEDIUM | modelopt/torch/utils/speech_dataset_utils.py | 128 | # Define the DataLoader (batches will be created automatically by DataLoader) | COMMENT |
| MEDIUM | modelopt/torch/utils/plugins/megatron_generate.py | 346 | # Create a static inference context if KV-cache is enabled. | COMMENT |
| MEDIUM | modelopt/torch/models/falcon/modeling_ptq.py | 39 | # Create a QuantFalconLinear class on the fly | COMMENT |
| MEDIUM | …delopt/torch/quantization/plugins/pytorch_geometric.py | 25 | >>> # Create a model with PyG layers | STRING |
| MEDIUM | modelopt/torch/quantization/plugins/attention.py | 333 | # Create a new class method from bytecode | COMMENT |
| MEDIUM | modelopt/torch/quantization/plugins/attention.py | 348 | # Create a new subclass on the fly | COMMENT |
| MEDIUM | modelopt/torch/quantization/backends/gemm_registry.py | 16 | """This module provides a registry for low-precision GEMM (General Matrix Multiplication) implementations. | STRING |
| MEDIUM | modelopt/torch/quantization/backends/nvfp4_gemm.py | 16 | """This module provides a GEMM function for nvfp4 quantization.""" | STRING |
| MEDIUM | …opt/torch/quantization/backends/fp8_per_tensor_gemm.py | 16 | """This module provides a GEMM function for fp8 per tensor quantization.""" | STRING |
| MEDIUM | modelopt/torch/quantization/utils/core_utils.py | 1050 | # Create a new QFSDPParam or FSDPParam based on weight type | COMMENT |
| MEDIUM | modelopt/torch/quantization/utils/core_utils.py | 792 | """This function is directly copied from the latest version of torch FSDP.""" | STRING |
| MEDIUM⚡ | modelopt/torch/quantization/qtensor/base_qtensor.py | 166 | # Create a QTensorWrapper with the correct metadata during unsharding | COMMENT |
| MEDIUM | modelopt/torch/quantization/qtensor/mxfp4_tensor.py | 97 | # Create a new tensor with alternating values | COMMENT |
| MEDIUM | modelopt/torch/quantization/qtensor/mxfp4_tensor.py | 122 | # Create a tensor with the E2M1 values | COMMENT |
| MEDIUM | modelopt/torch/_deploy/utils/torch_onnx.py | 169 | # Create a dictionary with all necessary attributes | COMMENT |
| MEDIUM | modelopt/torch/_deploy/utils/torch_onnx.py | 191 | # Create a new instance without calling __init__ and set the attributes | COMMENT |
| MEDIUM | modelopt/torch/_deploy/_runtime/common.py | 23 | """This function is supposed to use as a decorator to measure the execution time of another function. | STRING |
| MEDIUM | modelopt/torch/puzzletron/entrypoint.py | 16 | """This module provides the main compression function for a model using MIP-based NAS search algorithm.""" | STRING |
| MEDIUM | modelopt/torch/puzzletron/tools/logger.py | 123 | # Define a custom function to redirect warnings to logger | COMMENT |
| MEDIUM | …ch/puzzletron/replacement_library/replacement_utils.py | 15 | """This module provides helper functions for parsing, sorting, and analyzing layer replacement | STRING |
| MEDIUM | modelopt/torch/opt/conversion.py | 593 | model = ... # Create the model-like object | STRING |
| MEDIUM | modelopt/torch/opt/plugins/megatron.py | 42 | # Create a copy to avoid modifying the original metadata dict | COMMENT |
| MEDIUM | modelopt/torch/speculative/utils.py | 292 | # Initialize tracking variables | COMMENT |
| MEDIUM | modelopt/torch/speculative/utils.py | 523 | # Import required modules | COMMENT |
| MEDIUM | modelopt/torch/speculative/plugins/hf_medusa.py | 71 | # Create a list of Medusa heads | COMMENT |
| MEDIUM | modelopt/torch/export/convert_hf_config.py | 270 | # Create a hashable key from the layer config | COMMENT |
| MEDIUM | modelopt/torch/export/trtllm/layer_utils.py | 1217 | # Create a dummy module so that is_mlp may catch it. | COMMENT |
| MEDIUM | modelopt/torch/export/trtllm/model_config_export.py | 547 | # Create a NFS workspace under the export folder which is also assumed to be NFS. | COMMENT |
| MEDIUM⚡ | tests/unit/onnx/test_onnx_utils.py | 486 | # Create the ONNX initializers | COMMENT |
| MEDIUM⚡ | tests/unit/onnx/test_onnx_utils.py | 496 | # Create the ONNX graph with the nodes and initializers | COMMENT |
| MEDIUM⚡ | tests/unit/onnx/test_onnx_utils.py | 501 | # Create the ONNX model | COMMENT |
| MEDIUM | tests/unit/onnx/test_onnx_utils.py | 470 | # Create the ONNX graph with the nodes | COMMENT |
| MEDIUM | tests/unit/onnx/autocast/test_referencerunner.py | 34 | # Create a model with 2 inputs and 2 outputs | COMMENT |
| MEDIUM | tests/unit/onnx/autocast/test_referencerunner.py | 236 | # Create a single numpy array and save as NPZ (incorrect format) | COMMENT |
| 70 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | tools/mcp/modelopt_mcp/bridge.py | 243 | except ImportError: | CODE |
| HIGH | tools/mcp/modelopt_mcp/bridge.py | 256 | except ImportError: | CODE |
| HIGH | tools/mcp/modelopt_mcp/bridge.py | 645 | except (yaml.YAMLError, OSError): | CODE |
| HIGH | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 283 | except Exception: | CODE |
| HIGH | modelopt/onnx/quantization/int4.py | 119 | except ImportError: | CODE |
| HIGH | modelopt/onnx/quantization/int4.py | 487 | except FileNotFoundError: | CODE |
| HIGH | modelopt/onnx/quantization/int4.py | 494 | except FileNotFoundError: | CODE |
| HIGH | modelopt/onnx/quantization/autotune/autotuner_base.py | 927 | except (StopIteration, ValueError): | CODE |
| HIGH | modelopt/recipe/loader.py | 271 | except (TypeError, KeyError, ValueError): | CODE |
| HIGH | modelopt/torch/__init__.py | 58 | except ImportError: | CODE |
| HIGH | …delopt/torch/sparsity/attention_sparsity/conversion.py | 167 | except (ImportError, Exception): | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 213 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 221 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 232 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 240 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 259 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 344 | except ImportError: | CODE |
| HIGH | …sity/attention_sparsity/methods/triton_skip_softmax.py | 271 | except ImportError: | CODE |
| HIGH | …rch/sparsity/attention_sparsity/plugins/huggingface.py | 118 | except ImportError: | CODE |
| HIGH | …rch/sparsity/attention_sparsity/plugins/huggingface.py | 127 | except ImportError: | CODE |
| HIGH | modelopt/torch/trace/tracer.py | 98 | except self._expected_tracing_errors: | CODE |
| HIGH | modelopt/torch/nas/plugins/mbridge.py | 155 | except ImportError: | CODE |
| HIGH | modelopt/torch/utils/mlflow.py | 252 | except (OSError, IndexError): | CODE |
| HIGH | modelopt/torch/utils/mlflow.py | 907 | except Exception: | CODE |
| HIGH | modelopt/torch/utils/dataset_utils.py | 608 | except ImportError: | CODE |
| HIGH | modelopt/torch/utils/dataset_utils.py | 614 | except ImportError: | CODE |
| HIGH | modelopt/torch/utils/vlm_dataset_utils.py | 210 | except Exception: | CODE |
| HIGH | modelopt/torch/quantization/model_calib.py | 381 | except ImportError: | CODE |
| HIGH | modelopt/torch/quantization/plugins/huggingface.py | 1199 | except ImportError: | CODE |
| HIGH | modelopt/torch/quantization/plugins/huggingface.py | 1390 | except ImportError: | CODE |
| HIGH | …elopt/torch/quantization/plugins/transformer_engine.py | 50 | except ImportError: | CODE |
| HIGH | modelopt/torch/quantization/plugins/diffusion/ltx2.py | 58 | except Exception: | CODE |
| HIGH | …lopt/torch/quantization/plugins/diffusion/diffusers.py | 322 | except ImportError: | CODE |
| HIGH | modelopt/torch/quantization/qtensor/nvfp4_tensor.py | 313 | except ImportError: | CODE |
| HIGH | …etron/tools/validate_puzzle_with_multi_replacements.py | 255 | except Exception: | CODE |
| HIGH | modelopt/torch/puzzletron/tools/validate_model.py | 236 | except Exception: | CODE |
| HIGH | modelopt/torch/puzzletron/plugins/mbridge/base.py | 49 | except ImportError: | CODE |
| HIGH | modelopt/torch/puzzletron/utils/vllm_adapter.py | 193 | except AttributeError: | CODE |
| HIGH | modelopt/torch/puzzletron/utils/vllm_adapter.py | 152 | except AttributeError: | CODE |
| HIGH | …ron/anymodel/models/gpt_oss/gpt_oss_pruned_to_mxfp4.py | 372 | except PermissionError: | CODE |
| HIGH | modelopt/torch/export/hf_export_handlers.py | 121 | except ImportError: | CODE |
| HIGH | modelopt/torch/export/unified_export_megatron.py | 405 | except OSError: | CODE |
| HIGH | modelopt/torch/export/unified_export_megatron.py | 426 | except (OSError, ValueError, ImportError): | CODE |
| HIGH | modelopt/torch/export/unified_export_megatron.py | 417 | except (OSError, TypeError, ValueError, ImportError): | CODE |
| HIGH | modelopt/torch/export/diffusers_utils.py | 366 | except (TypeError, ValueError): | CODE |
| HIGH | modelopt/torch/export/unified_export_hf.py | 1009 | except ImportError: | CODE |
| HIGH | modelopt/torch/export/unified_export_hf.py | 1485 | except (ImportError, AttributeError): | CODE |
| HIGH | …onnx/quantization/test_quantize_onnx_torch_int4_awq.py | 60 | except ImportError: | CODE |
| HIGH | tests/_test_utils/deploy_utils.py | 46 | except ImportError: | CODE |
| HIGH | tests/_test_utils/deploy_utils.py | 53 | except ImportError: | CODE |
| HIGH | tests/_test_utils/deploy_utils.py | 60 | except ImportError: | CODE |
| HIGH | tests/_test_utils/torch/distributed/utils.py | 88 | except ImportError: | CODE |
| HIGH | …ing/distributed_generate/server_generate_vlm_sglang.py | 236 | except Exception: | CODE |
| HIGH | …ing/distributed_generate/server_generate_vlm_sglang.py | 250 | except Exception: | CODE |
| HIGH | …ing/distributed_generate/server_generate_vlm_sglang.py | 202 | except Exception: | CODE |
| HIGH | examples/diffusers/fastgen/fastgen_data/__init__.py | 89 | except Exception: # pragma: no cover - never block import on a version probe | CODE |
| HIGH | examples/diffusers/fastgen/dmd2/recipe.py | 289 | except Exception: | CODE |
| HIGH | examples/diffusers/fastgen/dmd2/recipe.py | 1325 | except Exception: | CODE |
| HIGH | examples/diffusers/quantization/models_utils.py | 283 | except ValueError: | CODE |
| HIGH | …nchmark/kl_divergence_metrics/compute_kl_divergence.py | 83 | except Exception: | CODE |
| 11 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tools/resource_monitor.py | 148 | except Exception: | CODE |
| LOW | tools/resource_monitor.py | 160 | except Exception: | CODE |
| LOW | tools/resource_monitor.py | 214 | except Exception: | CODE |
| LOW | tools/launcher/common/query.py | 79 | except Exception as e: | CODE |
| LOW | tools/mcp/modelopt_mcp/bridge.py | 2245 | except Exception as e: | CODE |
| LOW | tools/precommit/check_modelopt_recipes.py | 145 | except Exception: | CODE |
| LOW | tools/precommit/check_modelopt_recipes.py | 212 | except Exception as exc: | CODE |
| LOW | tools/precommit/check_launcher_yaml.py | 110 | except Exception as exc: | CODE |
| LOW | tools/precommit/check_launcher_yaml.py | 119 | except Exception as exc: | CODE |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 264 | except Exception: | CODE |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 283 | except Exception: | CODE |
| LOW | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 293 | except Exception: | CODE |
| MEDIUM | …nchmark-model-kernels/scripts/benchmark_via_builtin.py | 252 | def _gpu_description() -> str: | CODE |
| LOW | …lls/benchmark-model-kernels/scripts/benchmark_model.py | 152 | except Exception: | CODE |
| LOW | …lls/benchmark-model-kernels/scripts/benchmark_model.py | 159 | except Exception as exc: | CODE |
| LOW | modelopt/onnx/trt_utils.py | 57 | except Exception as e: | CODE |
| LOW | modelopt/onnx/trt_utils.py | 85 | except Exception as e: | CODE |
| LOW | modelopt/onnx/trt_utils.py | 107 | except Exception as e: | CODE |
| LOW | modelopt/onnx/trt_utils.py | 375 | except Exception as e: | CODE |
| LOW | modelopt/onnx/logging_config.py | 55 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 462 | except Exception: | CODE |
| LOW | modelopt/onnx/utils.py | 505 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 625 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 724 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 920 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 1101 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 1283 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 2036 | except Exception as e: | CODE |
| LOW | modelopt/onnx/utils.py | 2041 | except Exception as e: | CODE |
| LOW | modelopt/onnx/autocast/logging_config.py | 85 | except Exception as e: | CODE |
| LOW⚡ | modelopt/onnx/autocast/graphsanitizer.py | 197 | except Exception as e: | CODE |
| LOW⚡ | modelopt/onnx/autocast/graphsanitizer.py | 222 | except Exception as e: | CODE |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 256 | except Exception as e: | CODE |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 399 | except Exception as e: | CODE |
| LOW | modelopt/onnx/autocast/referencerunner.py | 214 | except Exception: | CODE |
| MEDIUM | modelopt/onnx/autocast/precisionconverter.py | 457 | def _can_propagate_type(from_type, to_type): | CODE |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 465 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 816 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 919 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 959 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1039 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/ort_utils.py | 261 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/ort_utils.py | 359 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/ort_utils.py | 400 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/ort_utils.py | 408 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/ort_utils.py | 516 | except Exception as exc: | CODE |
| LOW | modelopt/onnx/quantization/quantize.py | 233 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/quantize.py | 266 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/int4.py | 72 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/extensions.py | 32 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 435 | except Exception: | CODE |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 481 | except Exception: | CODE |
| LOW | modelopt/onnx/quantization/ort_calibration.py | 393 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/calib_utils.py | 174 | except Exception as e: | CODE |
| LOW⚡ | modelopt/onnx/quantization/ort_session.py | 61 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/autotune/benchmark.py | 143 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/autotune/benchmark.py | 235 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/autotune/benchmark.py | 303 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/autotune/benchmark.py | 422 | except Exception as e: | CODE |
| LOW | modelopt/onnx/quantization/autotune/benchmark.py | 716 | except Exception as e: | CODE |
| 246 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tools/launcher/common/megatron_lm/quantize/quantize.sh | 49 | # Step 1: quantize | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 28 | # Step 1: Build input conversations (model-agnostic) | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 42 | # Step 2: Streaming EAGLE3 training (node0 serve TP=4 / node1 train), 4 GPU/node | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 84 | # Step 3: Benchmark speculative decoding (VLLM backend, Kimi served at TP=4) | COMMENT |
| LOW | …r/examples/moonshotai/Kimi-K2.5/hf_offline_dflash.yaml | 19 | # Step 1: Data synthesis via vLLM server | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 24 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 49 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 67 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 89 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 34 | # Step 1: Online DFlash training | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 64 | # Step 2: vLLM smoke test (uses exported checkpoint from training) | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 80 | # Step 3: HF AR evaluation | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 25 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 49 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 67 | # Step 3: Train EAGLE3 draft head (offline, single task) and export checkpoint | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 92 | # Step 4: PTQ — quantize the EAGLE3 model using offline hidden states | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 108 | # Step 5: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_online_dflash2.yaml | 39 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_online_dflash2.yaml | 55 | # Step 2: Online DFlash2 training (the script exports the drafter at the end). | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 21 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 44 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 61 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 83 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 35 | # Step 1: Build input conversations | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 48 | # Step 2: Streaming DFlash2 training — node 0 vllm serve (TP=4), node 1 trainer. | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 108 | # Step 3: vLLM smoke test (uses the exported checkpoint from training). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 23 | # Step 1: Build input conversations | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 36 | # Step 2: Streaming EAGLE3 training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 68 | # Step 3: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 26 | # Step 1: Build input conversations | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 39 | # Step 2: Streaming DFlash training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 85 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3-8B/hf_online_lilicorr.yaml | 31 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/hf_online_lilicorr.yaml | 47 | # Step 2: Online LiLiCorr training (the script exports the drafter at the end). | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3-8B/megatron_lm_ptq.yaml | 66 | # Step 3: TRT-LLM eval MMLU on all exported checkpoints | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_dspark_ptq_nvfp4.yaml | 27 | # Step 1: Quantize. w4a16_nvfp4 keeps activations in bf16; use --qformat nvfp4 | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_dspark_ptq_nvfp4.yaml | 50 | # Step 2: Acceptance length on MT-Bench. Compare against the same run with | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_domino.yaml | 30 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_domino.yaml | 46 | # Step 2: Online Domino training (the script exports the drafter at the end). | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 28 | # Step 1: Build input conversations | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 41 | # Step 2: Streaming DFlash training — node 0 vllm serve, node 1 trainer. | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 82 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 24 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 48 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 66 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 84 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 36 | # Step 1: Build input conversations | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 49 | # Step 2: Streaming EAGLE3 training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 88 | # Step 3: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 40 | # Step 1: Build input conversations | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 53 | # Step 2: Streaming DFlash training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 107 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 22 | # Step 1: Online DFlash training | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 55 | # Step 2: vLLM smoke test with exported DFlash draft | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 70 | # Step 3: HF AR evaluation | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3.5-4B/specdec_bench.yaml | 34 | # Step 1: qualitative split — quality / acceptance-rate numbers. | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3.5-4B/specdec_bench.yaml | 62 | # Step 2: throughput_32k split — long-context throughput. | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 28 | # Step 1: qualitative split — quality / acceptance-rate numbers with MTP draft=3. | COMMENT |
| LOW⚡ | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 72 | # Step 2: throughput_32k split — long-context throughput with MTP draft=3. | COMMENT |
| LOW⚡ | …ples/MiniMax/MiniMax-M2.7-DFlash/hf_online_dflash.yaml | 29 | # Step 1: Online DFlash training. | COMMENT |
| 133 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | modelopt/onnx/utils.py | 1305 | Get the target type from a Cast node. Args: cast_node: The Cast node to extract type from. Returns: | STRING |
| HIGH | modelopt/onnx/autocast/utils.py | 66 | Get a single consumer node and raise exception if there are multiple consumers. Args: model: The ONNX model | STRING |
| HIGH | modelopt/onnx/quantization/qdq_utils.py | 555 | Get scale and zero point tensors for a node. Args: node: ONNX node to get scale and zero point for | STRING |
| HIGH | modelopt/onnx/quantization/qdq_utils.py | 644 | Convert a weight tensor to INT8/FP8 format based on scale and zero point. Args: weight_array: The weight te | STRING |
| HIGH | modelopt/onnx/quantization/qdq_utils.py | 722 | Convert FP32/FP16 weights of the given ONNX model to INT8/FP8 weights. This function converts a model with QDQ (Qua | STRING |
| HIGH | modelopt/onnx/quantization/qdq_utils.py | 834 | Remove DQ nodes from the input and Q from the output of quantized custom ops for TensorRT compatibility. TensorRT r | STRING |
| HIGH | modelopt/onnx/quantization/qdq_utils.py | 983 | Remove Q nodes from the inputs of a quantized ONNX model. This supports generating quantized models with low-precis | STRING |
| HIGH | modelopt/onnx/quantization/ort_utils.py | 54 | Run a 'trtexec' command via subprocess. Args: args: Arguments to pass to trtexec (without the 'trtexec' com | STRING |
| HIGH | modelopt/onnx/quantization/ort_utils.py | 76 | Check if the `trtexec` CLI tool is available in PATH and is >= min_version. Args: min_version (str): Minimu | STRING |
| HIGH | modelopt/onnx/quantization/autotune/region_pattern.py | 130 | Check if this pattern matches another pattern or region. This method provides three distinct behaviors dependin | STRING |
| HIGH | modelopt/onnx/quantization/autotune/region_pattern.py | 201 | Internal function: Match this pattern against a region. Args: region: The region to match against | STRING |
| HIGH | modelopt/onnx/quantization/autotune/region_pattern.py | 408 | Create signature for a single node including its parameters. Includes operation type and key attributes that af | STRING |
| HIGH | modelopt/onnx/quantization/autotune/common.py | 756 | Load pattern cache from a YAML file. Reads a previously saved pattern cache file and reconstructs all pattern | STRING |
| HIGH | modelopt/onnx/quantization/autotune/utils.py | 54 | Validate that a file path exists. Args: path: Path string to validate (can be None) description: De | STRING |
| HIGH | modelopt/onnx/quantization/autotune/workflows.py | 42 | Benchmark ONNX model inference latency using TensorRT Python API. Args: model_path: Path to ONNX model file | STRING |
| HIGH | modelopt/onnx/quantization/autotune/autotuner_base.py | 445 | Compute Q/DQ insertion points for the best schemes (assuming best=True). Args: best: If True, use t | STRING |
| HIGH | modelopt/onnx/quantization/autotune/autotuner_base.py | 571 | Export ONNX model with Q/DQ nodes inserted according to tested schemes. This method creates a modified version | STRING |
| HIGH | modelopt/onnx/graph_surgery/__init__.py | 91 | Run a graph surgery by name. This is the unified entry point for all graph surgeries. It dispatches to the appr | STRING |
| HIGH | …/torch/sparsity/attention_sparsity/methods/registry.py | 145 | Extract numeric parts for proper version sorting. Args: version_str: Version string (e.g., "v1", "v2", "v10 | STRING |
| HIGH | …/torch/sparsity/attention_sparsity/methods/registry.py | 196 | Get sparse attention method by name and optional version. Args: name: Method name to retrieve versi | STRING |
| HIGH | …arsity/attention_sparsity/calibration/ruler_dataset.py | 111 | Load Paul Graham essays from local files. Reads essay .txt files from data_dir/essays. Files must be downloaded | STRING |
| HIGH | …arsity/attention_sparsity/calibration/ruler_dataset.py | 506 | Generate target lengths as descending powers of 2. Args: max_seqlen: Maximum sequence length num_le | STRING |
| HIGH | …h/sparsity/attention_sparsity/calibration/calibrate.py | 45 | Extract tokenizer name/path from model config. Args: model: Model to extract tokenizer from Returns: | STRING |
| HIGH | …h/sparsity/attention_sparsity/calibration/calibrate.py | 66 | Extract and validate calibration config from sparse_cfg. Args: config: Sparse attention configuration dict | STRING |
| HIGH | …ization/linear_attention/fla_chunk_gated_delta_rule.py | 508 | Args: q (torch.Tensor): queries of shape `[B, T, H, K]`. k (torch.Tensor): | STRING |
| HIGH | …/torch/kernels/quantization/conv/implicit_gemm_cuda.py | 98 | Conv3D via implicit GEMM with BF16 WMMA tensor cores. Args: x: Input tensor [N, Cin, D, H, W] w: We | STRING |
| HIGH | modelopt/torch/kernels/quantization/gemm/fp8_kernel.py | 81 | Dequantizes the given weight tensor using the provided scale tensor. Reference: https://github.com/deepseek-ai/Deep | STRING |
| HIGH | modelopt/torch/kernels/common/attention/triton_fa.py | 1322 | Variable-length flash attention with GQA, autograd, optional sparsity, and paged KV. Args: q: [total_q_toke | STRING |
| HIGH | modelopt/torch/fastgen/factory.py | 58 | Return a trainable fake-score network initialized from the teacher. This is the unit-test / single-script path; fra | STRING |
| HIGH | modelopt/torch/utils/memory_monitor.py | 65 | Initialize a NVIDIA GPU memory monitor. This class monitors the memory usage of NVIDIA GPUs at specified interv | STRING |
| HIGH | modelopt/torch/utils/regex.py | 32 | Check if a name matches any of the given patterns. This utility function checks if a given name (e.g., module name, | STRING |
| HIGH | modelopt/torch/quantization/config.py | 248 | Find the last entry in a ``quant_cfg`` list whose ``quantizer_name`` key equals the query. This performs an **exact | STRING |
| HIGH | modelopt/torch/quantization/config.py | 1452 | Normalize a raw quant_cfg into a list of :class:`QuantizerCfgEntry` instances. Supports the following input forms: | STRING |
| HIGH | modelopt/torch/_deploy/utils/torch_onnx.py | 433 | Real quantizes the weights in the onnx model. Applies weight quantization to an ONNX model based on the quantizatio | STRING |
| HIGH | …lopt/torch/_deploy/_runtime/tensorrt/engine_builder.py | 43 | Run a 'trtexec' command via subprocess, logging the cmd and any failure output. The 'trtexec' binary is hardcoded a | STRING |
| HIGH | …etron/tools/validate_puzzle_with_multi_replacements.py | 67 | Validate puzzle solutions by applying layer replacements and evaluating model performance. Args: args: Conf | STRING |
| HIGH | modelopt/torch/puzzletron/tools/validate_model.py | 74 | Validate a language model on a dataset by calculating loss and optionally capturing activations. Args: args | STRING |
| HIGH | …/anymodel/model_descriptor/model_descriptor_factory.py | 41 | Resolve the model descriptor by loading the checkpoint config and mapping model_type. Args: pretrained: Pat | STRING |
| HIGH | modelopt/torch/opt/conversion.py | 558 | Load the modelopt state from a file. Args: modelopt_state_path: Target file location. **kwargs: add | STRING |
| HIGH | modelopt/torch/opt/mode.py | 138 | The mode's entrypoint for converting a model. The function signature of the convert entrypoint is described bel | STRING |
| HIGH | modelopt/torch/opt/mode.py | 166 | The mode's entrypoint for restoring a model. The function signature of the restore entrypoint is described belo | STRING |
| HIGH | …torch/prune/importance_hooks/compare_module_outputs.py | 206 | Compute RMSE and cosine similarity for a layer's outputs. Args: ref_data: List of reference tensors. | STRING |
| HIGH | modelopt/torch/export/model_utils.py | 95 | Check if a model is a Vision-Language Model (VLM) or multimodal model. This function detects various multimodal mod | STRING |
| HIGH | modelopt/torch/export/model_utils.py | 128 | Extract the language model lineage from a Vision-Language Model (VLM). This function handles the common patterns fo | STRING |
| HIGH | modelopt/torch/export/diffusers_utils.py | 610 | Extract the parent attention block path and QKV type for grouping. QKV projections should only be fused within the | STRING |
| HIGH | modelopt/torch/export/diffusers_utils.py | 1025 | Detect the diffusion model type for merge function dispatch. To add a new model type, add a detection clause here a | STRING |
| HIGH | modelopt/torch/export/unified_export_hf.py | 1044 | Exports the torch model to the packed checkpoint with original HF naming. The packed checkpoint will be consumed by | STRING |
| HIGH | modelopt/torch/export/unified_export_hf_streaming.py | 313 | Export a disk/CPU-offloaded model by streaming tensors layer-by-layer to shard files. The offloaded counterpart of | STRING |
| HIGH | examples/hf_ptq/cast_mxfp4_to_nvfp4.py | 112 | Walk the source MXFP4 checkpoint and build the per-layer amax map. Args: checkpoint_dir: Path to a Hugging | STRING |
| HIGH | …users/fastgen/preprocess/processors/caption_loaders.py | 123 | Factory method to get the appropriate caption loader. Args: format_name: One of 'sidecar', | STRING |
| HIGH | …es/diffusers/fastgen/preprocess/processors/registry.py | 73 | Get a processor instance by name. Args: name: Registered processor name Returns: | STRING |
| HIGH | …es/diffusers/fastgen/preprocess/processors/registry.py | 92 | Get a processor class by name (without instantiating). Args: name: Registered processor na | STRING |
| HIGH | examples/onnx_ptq/evaluation.py | 129 | Evaluate a model for the given dataset. Args: model: PyTorch model or DeviceModel to evaluate. tran | STRING |
| HIGH | examples/onnx_ptq/evaluation.py | 176 | Evaluate the accuracy of the model on the validation dataset. Args: model: Model to evaluate. val_l | STRING |
| HIGH | …amples/puzzletron/evaluation/hf_deployable_anymodel.py | 187 | Generate text based on the provided input prompts. This method processes input prompts through the loaded pipel | STRING |
| HIGH | …amples/puzzletron/evaluation/hf_deployable_anymodel.py | 569 | Perform inference using Ray with dictionary inputs and outputs. Args: inputs (dict[Any, Any]): Dict | STRING |
| HIGH | …acy_benchmark/perplexity_metrics/perplexity_metrics.py | 63 | Evaluate perplexity of a HuggingFace model on the WikiText-2 dataset. This function computes perplexity using | STRING |
| HIGH | examples/specdec_bench/specdec_bench/datasets/base.py | 44 | Prepare and save the dataset to the specified output directory. Downloads any external data, resolves all refer | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tools/launcher/common/megatron_lm/quantize/quantize.sh | 49 | # Step 1: quantize | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 28 | # Step 1: Build input conversations (model-agnostic) | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 42 | # Step 2: Streaming EAGLE3 training (node0 serve TP=4 / node1 train), 4 GPU/node | COMMENT |
| LOW | …examples/moonshotai/Kimi-K2.5/hf_streaming_eagle3.yaml | 84 | # Step 3: Benchmark speculative decoding (VLLM backend, Kimi served at TP=4) | COMMENT |
| LOW | …r/examples/moonshotai/Kimi-K2.5/hf_offline_dflash.yaml | 19 | # Step 1: Data synthesis via vLLM server | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 24 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 49 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 67 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 89 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 34 | # Step 1: Online DFlash training | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 64 | # Step 2: vLLM smoke test (uses exported checkpoint from training) | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 80 | # Step 3: HF AR evaluation | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 25 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 49 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 67 | # Step 3: Train EAGLE3 draft head (offline, single task) and export checkpoint | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 92 | # Step 4: PTQ — quantize the EAGLE3 model using offline hidden states | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 108 | # Step 5: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_online_dflash2.yaml | 39 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_online_dflash2.yaml | 55 | # Step 2: Online DFlash2 training (the script exports the drafter at the end). | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 21 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 44 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 61 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 83 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 35 | # Step 1: Build input conversations | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 48 | # Step 2: Streaming DFlash2 training — node 0 vllm serve (TP=4), node 1 trainer. | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 108 | # Step 3: vLLM smoke test (uses the exported checkpoint from training). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 23 | # Step 1: Build input conversations | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 36 | # Step 2: Streaming EAGLE3 training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 68 | # Step 3: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 26 | # Step 1: Build input conversations | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 39 | # Step 2: Streaming DFlash training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 85 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3-8B/hf_online_lilicorr.yaml | 31 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/hf_online_lilicorr.yaml | 47 | # Step 2: Online LiLiCorr training (the script exports the drafter at the end). | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3-8B/megatron_lm_ptq.yaml | 66 | # Step 3: TRT-LLM eval MMLU on all exported checkpoints | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_dspark_ptq_nvfp4.yaml | 27 | # Step 1: Quantize. w4a16_nvfp4 keeps activations in bf16; use --qformat nvfp4 | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_dspark_ptq_nvfp4.yaml | 50 | # Step 2: Acceptance length on MT-Bench. Compare against the same run with | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_domino.yaml | 30 | # Step 1: Build input conversations. example_data_config.yaml enables only the | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_domino.yaml | 46 | # Step 2: Online Domino training (the script exports the drafter at the end). | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 28 | # Step 1: Build input conversations | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 41 | # Step 2: Streaming DFlash training — node 0 vllm serve, node 1 trainer. | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 82 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 24 | # Step 1: Data synthesis via TRT-LLM server | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 48 | # Step 2: Dump hidden states from target model | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 66 | # Step 3: Train EAGLE3 draft head (offline, single task) | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 84 | # Step 4: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 36 | # Step 1: Build input conversations | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 49 | # Step 2: Streaming EAGLE3 training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 88 | # Step 3: Benchmark speculative decoding (VLLM backend) | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 40 | # Step 1: Build input conversations | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 53 | # Step 2: Streaming DFlash training — 2 serve replicas (TP=2) + 2 trainer nodes (2 GPU each). | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 107 | # Step 3: vLLM smoke test (DFlash, uses exported checkpoint from training) | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 22 | # Step 1: Online DFlash training | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 55 | # Step 2: vLLM smoke test with exported DFlash draft | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 70 | # Step 3: HF AR evaluation | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3.5-4B/specdec_bench.yaml | 34 | # Step 1: qualitative split — quality / acceptance-rate numbers. | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3.5-4B/specdec_bench.yaml | 62 | # Step 2: throughput_32k split — long-context throughput. | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 28 | # Step 1: qualitative split — quality / acceptance-rate numbers with MTP draft=3. | COMMENT |
| LOW⚡ | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 72 | # Step 2: throughput_32k split — long-context throughput with MTP draft=3. | COMMENT |
| LOW⚡ | …ples/MiniMax/MiniMax-M2.7-DFlash/hf_online_dflash.yaml | 29 | # Step 1: Online DFlash training. | COMMENT |
| 99 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | experimental/dms/configs/qwen3_8b.yaml | 7 | # Set resume_from_checkpoint to "auto" below, or pass an explicit path. | COMMENT |
| LOW | …xamples/MiniMax/MiniMax-M2.7-DFlash/specdec_bench.yaml | 15 | # Set draft_model to the trained, exported draft checkpoint (an exported-checkpoint-* | COMMENT |
| LOW | tools/precommit/check_modelopt_recipes.py | 284 | # Check if this file is inside a directory-format recipe. | COMMENT |
| LOW | tools/debugger/client.sh | 120 | # Check if server is still alive | COMMENT |
| LOW | tools/debugger/client.sh | 214 | # Check if there's a running command | COMMENT |
| LOW | plugins/modelopt/skills/common/remote_exec.sh | 305 | # Check if nvidia-container-cli is available (GPU support without pulling an image) | COMMENT |
| LOW | plugins/modelopt/skills/common/remote_exec.sh | 484 | # Check if it's a running container | COMMENT |
| LOW | plugins/modelopt/skills/deployment/scripts/deploy.sh | 230 | # Check if port is already in use | COMMENT |
| LOW | plugins/modelopt/skills/deployment/scripts/deploy.sh | 367 | # Check if process died | COMMENT |
| LOW | modelopt/onnx/trt_utils.py | 415 | # Check if there's a custom TensorRT op in the ONNX model. If so, make it ORT compatible by adding | COMMENT |
| LOW | modelopt/onnx/utils.py | 734 | # Set ir_version to 10, remove it once ORT supports ir_version 11 | COMMENT |
| LOW | modelopt/onnx/utils.py | 1049 | # Check if this node has subgraphs (GRAPH or GRAPHS attributes) | COMMENT |
| LOW | modelopt/onnx/utils.py | 1460 | # Check if the cast output is also a graph output | COMMENT |
| LOW | modelopt/onnx/autocast/logging_config.py | 44 | # Check if parent logger (modelopt.onnx) already has handlers configured | COMMENT |
| LOW | modelopt/onnx/autocast/convert.py | 270 | # Check if Q/DQ nodes require a higher opset | COMMENT |
| LOW⚡ | modelopt/onnx/autocast/graphsanitizer.py | 216 | # Set ir_version to 10, remove it once ORT supports ir_version 11 | COMMENT |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 578 | # Check if casting to FP64, change to FP32 | COMMENT |
| LOW | modelopt/onnx/autocast/graphsanitizer.py | 586 | # Check if the value attribute uses FP64 | COMMENT |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1351 | # Check if input comes from an initializer - don't remove cast in that case | COMMENT |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 1887 | # Check if we should skip this input for low precision conversion | COMMENT |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1131 | # Check if second input (weight) is from DequantizeLinear | COMMENT |
| LOW | modelopt/onnx/quantization/qdq_utils.py | 1349 | # Check if there's an optional Cast node between Reshape and Transpose/MatMul/Gemm | COMMENT |
| LOW | modelopt/onnx/quantization/quantize.py | 271 | # Check if data-dependent shape ops are present in the model | COMMENT |
| LOW | modelopt/onnx/quantization/quantize.py | 951 | # Check if intermediate files should be deleted | COMMENT |
| LOW | modelopt/onnx/quantization/quantize.py | 960 | # Check if the quantized model is valid | COMMENT |
| LOW | modelopt/onnx/quantization/graph_indexing.py | 118 | # Check if current non-wild node type does not match the expected path type | COMMENT |
| LOW | modelopt/onnx/quantization/graph_indexing.py | 152 | # Check if any child (forward path) or parent (backward path) can match the remaining path types | COMMENT |
| LOW | modelopt/onnx/quantization/partitioning.py | 211 | # Check if the node is already in some other partition | COMMENT |
| LOW | modelopt/onnx/quantization/int4.py | 694 | # Set ir_version to 10, remove it once ORT supports ir_version 11 | COMMENT |
| LOW | modelopt/onnx/quantization/int4.py | 1425 | # Set ir_version to 10, remove it once ORT supports ir_version 11 | COMMENT |
| LOW | modelopt/onnx/quantization/int4.py | 1521 | # Check if block_size is None and set default to 128 | COMMENT |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 116 | # Check if the first input (activation) comes from a CASK partition | COMMENT |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 198 | # Check if partition head is of default quantizable type | COMMENT |
| LOW | modelopt/onnx/quantization/qdq_graph.py | 862 | # Check if concat output is quantized | STRING |
| LOW | modelopt/onnx/quantization/ort_quantization.py | 134 | # Set ir_version to 10, remove it once ORT supports ir_version 11 | COMMENT |
| LOW | modelopt/onnx/quantization/autotune/region_search.py | 668 | # Check if convergence creates a reasonable-sized region | STRING |
| LOW | modelopt/onnx/quantization/autotune/common.py | 520 | # Check if this scheme is too similar to any already-filtered scheme | COMMENT |
| LOW | modelopt/onnx/graph_surgery/gqa_replacement.py | 144 | # Check if output is a graph output | COMMENT |
| LOW | modelopt/onnx/graph_surgery/gqa_replacement.py | 1383 | # Check if opset 21 is being used (requires IR version >= 10) | COMMENT |
| LOW | modelopt/onnx/graph_surgery/dq_transpose.py | 213 | # Check if DQ output is the weight input (input[1] for MatMul/Gemm) | COMMENT |
| LOW | modelopt/onnx/graph_surgery/dq_transpose.py | 251 | # Check if 2D | COMMENT |
| LOW | modelopt/onnx/graph_surgery/utils/whisper_utils.py | 111 | # Check if file already exists | COMMENT |
| LOW | modelopt/onnx/graph_surgery/utils/whisper_utils.py | 365 | # Check if file already exists | COMMENT |
| LOW | modelopt/onnx/graph_surgery/utils/graph_utils.py | 314 | # Check if ANY of the node's outputs are consumed | COMMENT |
| LOW | modelopt/onnx/export/int4_exporter.py | 73 | # Check if there's an optional Cast node between Reshape and Transpose/MatMul/Gemm | COMMENT |
| LOW | …h/sparsity/attention_sparsity/calibration/calibrate.py | 343 | # Check if any calibration succeeded | COMMENT |
| LOW | …rch/sparsity/attention_sparsity/plugins/huggingface.py | 78 | # Check if this is an attention module by name | COMMENT |
| LOW | modelopt/torch/distill/distillation_model.py | 85 | # Assign loss to specified modules. | COMMENT |
| LOW | modelopt/torch/distill/distillation.py | 78 | # Check if the model is a distillation model. | COMMENT |
| LOW⚡ | modelopt/torch/utils/graph.py | 45 | # Check if input and output degrees match. | COMMENT |
| LOW⚡ | modelopt/torch/utils/graph.py | 51 | # Check if the node operator matches. | COMMENT |
| LOW⚡ | modelopt/torch/utils/graph.py | 55 | # Check if the node target matches. | COMMENT |
| LOW | modelopt/torch/utils/graph.py | 66 | # Check if we have already matched these nodes in the current traversal. | COMMENT |
| LOW | modelopt/torch/utils/graph.py | 112 | # Check if the number of nodes match. | COMMENT |
| LOW | modelopt/torch/peft/convert.py | 65 | # Check if model is already in PEFT mode by looking for LoRA modules | COMMENT |
| LOW | modelopt/torch/quantization/algorithms.py | 2066 | # Check if we meet the constraint | COMMENT |
| LOW | modelopt/torch/quantization/mode.py | 354 | # Check if _calib_func is defined as a class attribute | COMMENT |
| LOW | modelopt/torch/quantization/nn/modules/quant_linear.py | 228 | # Check if real-quant GEMM is available | COMMENT |
| LOW | modelopt/torch/quantization/nn/modules/quant_linear.py | 297 | # Check if fn is a tensor_cast_fun and print warning if so | COMMENT |
| LOW | …lopt/torch/quantization/nn/modules/tensor_quantizer.py | 1229 | # Check if the input tensor is contiguous | COMMENT |
| 73 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | …perimental/dms/models/qwen3/configuration_qwen3_dms.py | 40 | CODE | |
| LOW | experimental/dms/models/qwen3/modeling_qwen3_dms.py | 274 | CODE | |
| LOW | experimental/dms/models/qwen3/modeling_qwen3_dms.py | 392 | CODE | |
| LOW | experimental/dms/dms/training/engine.py | 573 | CODE | |
| LOW | tools/launcher/slurm_config.py | 69 | CODE | |
| LOW | tools/launcher/launch.py | 125 | CODE | |
| LOW | tools/mcp/modelopt_mcp/server.py | 160 | CODE | |
| LOW | tools/mcp/modelopt_mcp/bridge.py | 1018 | CODE | |
| LOW | modelopt/onnx/autocast/convert.py | 65 | CODE | |
| LOW | modelopt/onnx/autocast/convert.py | 225 | CODE | |
| LOW | modelopt/onnx/autocast/precisionconverter.py | 86 | CODE | |
| LOW | modelopt/onnx/autocast/nodeclassifier.py | 403 | CODE | |
| LOW | modelopt/onnx/quantization/fp8.py | 139 | CODE | |
| LOW | modelopt/onnx/quantization/quantize.py | 320 | CODE | |
| LOW | modelopt/onnx/quantization/quantize.py | 466 | CODE | |
| LOW | modelopt/onnx/quantization/int4.py | 1437 | CODE | |
| LOW | modelopt/onnx/quantization/graph_selection.py | 474 | CODE | |
| LOW | modelopt/onnx/quantization/int8.py | 122 | CODE | |
| LOW | modelopt/onnx/quantization/ort_quantization.py | 265 | CODE | |
| LOW | modelopt/onnx/quantization/autotune/workflows.py | 163 | CODE | |
| LOW | modelopt/onnx/graph_surgery/gqa_replacement.py | 788 | CODE | |
| LOW | modelopt/onnx/graph_surgery/encoder_cross_kv.py | 319 | CODE | |
| LOW | modelopt/onnx/graph_surgery/utils/whisper_utils.py | 144 | CODE | |
| LOW | …arsity/attention_sparsity/calibration/ruler_dataset.py | 228 | CODE | |
| LOW | modelopt/torch/kernels/sparsity/attention/calibrate.py | 44 | CODE | |
| LOW | modelopt/torch/kernels/sparsity/attention/calibrate.py | 300 | CODE | |
| LOW | …els/quantization/linear_attention/fla_chunk_delta_h.py | 809 | CODE | |
| LOW | …els/quantization/linear_attention/fla_chunk_delta_h.py | 898 | CODE | |
| LOW | …ization/linear_attention/fla_chunk_gated_delta_rule.py | 68 | CODE | |
| LOW | …ization/linear_attention/fla_chunk_gated_delta_rule.py | 170 | CODE | |
| LOW | …ization/linear_attention/fla_chunk_gated_delta_rule.py | 489 | CODE | |
| LOW | …ization/linear_attention/fla_chunk_gated_delta_rule.py | 314 | CODE | |
| LOW | …opt/torch/kernels/common/attention/decode_attention.py | 263 | CODE | |
| LOW | modelopt/torch/kernels/common/attention/triton_fa.py | 275 | CODE | |
| LOW | modelopt/torch/kernels/common/attention/triton_fa.py | 1293 | CODE | |
| LOW | modelopt/torch/fastgen/methods/pdd.py | 644 | CODE | |
| LOW | modelopt/torch/fastgen/plugins/qwen_image_pdd.py | 102 | CODE | |
| LOW | modelopt/torch/utils/dataset_utils.py | 745 | CODE | |
| LOW | modelopt/torch/utils/dataset_utils.py | 1207 | CODE | |
| LOW | modelopt/torch/utils/vlm_dataset_utils.py | 379 | CODE | |
| LOW | modelopt/torch/utils/plugins/megatron_generate.py | 294 | CODE | |
| LOW | …delopt/torch/utils/plugins/megatron_preprocess_data.py | 510 | CODE | |
| LOW | modelopt/torch/quantization/algorithms.py | 392 | CODE | |
| LOW | modelopt/torch/quantization/model_quant.py | 583 | CODE | |
| LOW | modelopt/torch/quantization/export_onnx.py | 356 | CODE | |
| LOW | modelopt/torch/quantization/tensor_quant.py | 153 | CODE | |
| LOW | modelopt/torch/quantization/tensor_quant.py | 213 | CODE | |
| LOW | modelopt/torch/quantization/tensor_quant.py | 324 | CODE | |
| LOW | modelopt/torch/quantization/tensor_quant.py | 350 | CODE | |
| LOW | modelopt/torch/quantization/model_calib.py | 1027 | CODE | |
| LOW | modelopt/torch/quantization/calib/nvfp4_act_headroom.py | 75 | CODE | |
| LOW | modelopt/torch/quantization/plugins/vllm.py | 493 | CODE | |
| LOW | modelopt/torch/quantization/plugins/peft.py | 99 | CODE | |
| LOW | …lopt/torch/quantization/plugins/diffusion/diffusers.py | 226 | CODE | |
| LOW | …lopt/torch/quantization/plugins/diffusion/diffusers.py | 259 | CODE | |
| LOW | modelopt/torch/quantization/backends/nvfp4_gemm.py | 147 | CODE | |
| LOW | …opt/torch/quantization/backends/fp8_per_tensor_gemm.py | 170 | CODE | |
| LOW | modelopt/torch/quantization/utils/calib_utils.py | 166 | CODE | |
| LOW | modelopt/torch/_deploy/utils/torch_onnx.py | 482 | CODE | |
| LOW | …lopt/torch/_deploy/_runtime/tensorrt/engine_builder.py | 139 | CODE | |
| 73 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | experimental/dms/scripts/evaluate.sh | 18 | # Evaluate a trained DMS model using lm-eval-harness. | COMMENT |
| MEDIUM⚡ | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 53 | # actually running. Set them in the harness (planned for Phase 2 in | COMMENT |
| MEDIUM⚡ | …uncher/examples/Qwen/Qwen3.5-4B/specdec_bench_mtp.yaml | 63 | # When run standalone (without the harness), dump_env falls back to | COMMENT |
| MEDIUM⚡ | …/meta-llama/Llama-3.2-1B-Instruct/megatron_lm_qad.yaml | 12 | # args below, so flags pass via `args:` (robust) rather than space-containing env | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 300 | # file (config.yml lives under <harness>.<task>/artifacts/) can still | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 311 | # Depth-agnostic: repo-documented trees put <harness>.<task>/artifacts/ directly under the | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 338 | # Fall back to "<harness>.<task>[.<run_index>]": strip the run index only | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 339 | # when the trailing segment is numeric, and KEEP the harness, since two | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 366 | # like "<harness>.<task>": "h.task.1" + "h.task.2" are repeats of one task, whereas | COMMENT |
| MEDIUM | …modelopt/skills/day0-release/scripts/gate_verbosity.py | 371 | # - "<harness>.<task>" keys come from "<...>.<run_index>" -> "." in k | COMMENT |
| MEDIUM | …t/skills/evaluation/recipes/examples/example_eval.yaml | 9 | # Includes (default): ns_gpqa (nemo-skills harness, num_repeats=16, AA mcq-4choices). | COMMENT |
| MEDIUM | …t/skills/evaluation/recipes/examples/example_eval.yaml | 169 | # prompt (golden Nemotron-3-Ultra harness); repeat count kept at this recipe's | COMMENT |
| MEDIUM | plugins/modelopt/skills/evaluation/scripts/nel-next.sh | 39 | # a different harness for the baseline and the candidate run, folding a harness change into | COMMENT |
| MEDIUM | modelopt/onnx/quantization/fp8.py | 198 | # Either of m or n in matmul is 1, this matmul cannot utilize TensorCores. | COMMENT |
| MEDIUM | modelopt/onnx/quantization/int8.py | 172 | # Either of m or n in matmul is 1, this matmul cannot utilize TensorCores. | COMMENT |
| MEDIUM | modelopt/torch/trace/tracer.py | 34 | """A robust graph tracer for tracing dependencies between layers. | STRING |
| MEDIUM | modelopt/torch/utils/robust_json.py | 17 | """Provides a robust JSON encoder that can handle various types of objects. | STRING |
| LOW | modelopt/torch/utils/distributed.py | 471 | # NCCL has an issue with calling barrier. So we just use the gloo backebnd for group barriers. | COMMENT |
| LOW | modelopt/torch/utils/network.py | 423 | # simply return as args/kwargs in this case | COMMENT |
| LOW | modelopt/torch/utils/vlm_dataset_utils.py | 240 | # If it's already a PIL/numpy/torch image-like object, just return it and let the processor validate. | COMMENT |
| MEDIUM | …elopt/torch/quantization/plugins/transformer_engine.py | 89 | # signature — robust to TE versions that insert positional args between them | COMMENT |
| MEDIUM | …elopt/torch/quantization/plugins/transformer_engine.py | 262 | # signature — robust to TE versions that move the m_splits/non_tensor_args | COMMENT |
| MEDIUM | …pt/torch/prune/importance_hooks/base_hooks_analysis.py | 55 | >>> # Collect multiple batches for robust evaluation | STRING |
| MEDIUM | …/torch/kernels/quantization/conv/test_implicit_gemm.py | 79 | # Check mean diff is small (more robust than quantile for large tensors) | COMMENT |
| MEDIUM | …ch/prune/plugins/test_mcore_hybrid_minitron_pruning.py | 395 | # demonstrating that active_params-first ranking can elevate 4-layer above 3-layer models | COMMENT |
| MEDIUM | …elopt_recipes/general/speculative_decoding/dspark.yaml | 45 | # export + the offline acceptance-length harness instead. | COMMENT |
| MEDIUM | …lopt_recipes/general/speculative_decoding/dflash2.yaml | 48 | # via export + the offline acceptance-length harness instead. | COMMENT |
| MEDIUM | examples/hf_ptq/scripts/huggingface_example.sh | 269 | # Flags instructed by https://github.com/EleutherAI/lm-evaluation-harness/tree/main/lm_eval/tasks/llama3#paper | COMMENT |
| MEDIUM | …/diffusers/fastgen/preprocess/processors/qwen_image.py | 172 | # real prompt length; keeping the true mask makes the contract explicit and robust. | COMMENT |
| MEDIUM | examples/llm_eval/lm_eval_hf.py | 1 | # Adapted from https://github.com/EleutherAI/lm-evaluation-harness/tree/aa457edc3d64d81530159cd3a182932320c78f8c | COMMENT |
| MEDIUM | examples/llm_eval/run_lm_eval_vllm.sh | 19 | # Script to run lm-evaluation-harness against a running vLLM OpenAI-compatible server. | COMMENT |
| MEDIUM | examples/specdec_bench/specdec_bench/utils.py | 292 | # Each *_sha and modelopt_version prefers an env var set by the harness | COMMENT |
| MEDIUM | examples/specdec_bench/specdec_bench/utils.py | 295 | # harness. container_image and nmm_sandbox_sha are env-only — there is no | COMMENT |
| MEDIUM | examples/specdec_bench/specdec_bench/utils.py | 317 | # runs. The harness sets JIRA_TICKET only after verifying the checkpoint | COMMENT |
| MEDIUM | examples/torch_trt/torch_tensorrt_accuracy.py | 55 | # Reuse the ImageNet accuracy harness from the onnx_ptq example (sibling dir). | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | experimental/dms/configs/qwen3_8b.yaml | 3 | # Usage: | COMMENT |
| LOW | experimental/dms/scripts/train.sh | 20 | # Usage: | COMMENT |
| LOW | experimental/dms/scripts/evaluate.sh | 23 | # Usage: | COMMENT |
| LOW | tools/launcher/common/vllm/query.sh | 23 | # Usage: | COMMENT |
| LOW | tools/launcher/common/tensorrt_llm/query.sh | 23 | # Usage: | COMMENT |
| LOW | tools/launcher/common/hf/ptq.sh | 19 | # Usage: | COMMENT |
| LOW | …er/examples/moonshotai/Kimi-K2.5/hf_dflash_dryrun.yaml | 5 | # Usage: | COMMENT |
| LOW | …r/examples/moonshotai/Kimi-K2.5/hf_offline_dflash.yaml | 7 | # Usage: | COMMENT |
| LOW | …er/examples/moonshotai/Kimi-K2.5/hf_eagle3_dryrun.yaml | 31 | # Usage: | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-0.6B/hf_online_eagle3.yaml | 11 | # Usage: | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-0.6B/hf_online_dflash.yaml | 20 | # Usage: | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_dflash.yaml | 12 | # Usage: | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-0.6B/hf_offline_eagle3.yaml | 11 | # Usage: | COMMENT |
| LOW | tools/launcher/examples/Qwen/Qwen3-8B/hf_synth.yaml | 3 | # Usage: | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_online_eagle3.yaml | 11 | # Usage: | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_dflash.yaml | 25 | # Usage: | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/hf_offline_eagle3_ptq.yaml | 12 | # Usage: | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_online_dflash2.yaml | 22 | # Usage: | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_eagle3.yaml | 11 | # Usage: | COMMENT |
| LOW | …auncher/examples/Qwen/Qwen3-8B/eagle3_quick_check.yaml | 8 | # Usage: | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash2.yaml | 23 | # Usage: | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_eagle3_multi_node.yaml | 11 | # Usage: | COMMENT |
| LOW | …ples/Qwen/Qwen3-8B/hf_streaming_dflash_multi_node.yaml | 14 | # Usage: | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3-8B/hf_online_lilicorr.yaml | 22 | # Usage: | COMMENT |
| LOW | …/launcher/examples/Qwen/Qwen3-8B/hf_eagle3_dryrun.yaml | 28 | # Usage: | COMMENT |
| LOW | …s/launcher/examples/Qwen/Qwen3-8B/megatron_lm_ptq.yaml | 8 | # Usage: | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_dspark_ptq_nvfp4.yaml | 12 | # Usage: | COMMENT |
| LOW⚡ | …/launcher/examples/Qwen/Qwen3-8B/hf_online_domino.yaml | 21 | # Usage: | COMMENT |
| LOW | …uncher/examples/Qwen/Qwen3-8B/hf_streaming_dflash.yaml | 16 | # Usage: | COMMENT |
| LOW | …launcher/examples/Qwen/Qwen3-8B/hf_offline_eagle3.yaml | 11 | # Usage: | COMMENT |
| LOW | …cher/examples/Qwen/Qwen3-8B/megatron_lm_ptq_local.yaml | 3 | # Usage: | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_eagle3_multi_node.yaml | 24 | # Usage: | COMMENT |
| LOW | …Qwen/Qwen3-30B-A3B/hf_streaming_dflash_multi_node.yaml | 28 | # Usage: | COMMENT |
| LOW | …ncher/examples/Qwen/Qwen3-30B-A3B/megatron_lm_ptq.yaml | 6 | # Usage: | COMMENT |
| LOW⚡ | …auncher/examples/Qwen/Qwen3.5-4B/hf_online_dflash.yaml | 13 | # Usage: | COMMENT |
| LOW⚡ | …ples/MiniMax/MiniMax-M2.7-DFlash/hf_online_dflash.yaml | 20 | # Usage: | COMMENT |
| LOW⚡ | …les/MiniMax/MiniMax-M2.7-DFlash/hf_offline_dflash.yaml | 16 | # Usage: | COMMENT |
| LOW | …iniMax-M2.7-DFlash/hf_streaming_dflash_multi_node.yaml | 30 | # Usage: | COMMENT |
| LOW | …xamples/MiniMax/MiniMax-M2.7-DFlash/specdec_bench.yaml | 18 | # Usage: | COMMENT |
| LOW | …/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16-AutoQuantize.yaml | 4 | # Usage: | COMMENT |
| LOW | …VIDIA-Nemotron-3-Nano-30B-A3B-BF16/offline_kd_qad.yaml | 35 | # Usage: | COMMENT |
| LOW | …VIDIA-Nemotron-3-Nano-30B-A3B-BF16/nemo_evaluator.yaml | 17 | # Usage: | COMMENT |
| LOW | …IDIA-Nemotron-3-Nano-30B-A3B-BF16/megatron_lm_ptq.yaml | 31 | # Usage: | COMMENT |
| LOW | …A-Nemotron-3-Super-120B-A12B-BF16/megatron_lm_ptq.yaml | 12 | # Usage: | COMMENT |
| LOW | …ron-3-Super-120B-A12B-BF16/megatron_bridge_import.yaml | 7 | # Usage: | COMMENT |
| LOW | …A-Nemotron-3-Ultra-550B-A55B-BF16/megatron_lm_ptq.yaml | 13 | # Usage: | COMMENT |
| LOW | …tron-3.5-Lightning-30B-A3B-BF16/hf_dspark_ptq_fp8.yaml | 13 | # Usage: | COMMENT |
| LOW | …on-3.5-Lightning-30B-A3B-BF16/hf_dspark_ptq_nvfp4.yaml | 12 | # Usage: | COMMENT |
| LOW⚡ | …/meta-llama/Llama-3.2-1B-Instruct/megatron_lm_qad.yaml | 21 | # Usage: | COMMENT |
| LOW | …openai/gpt-oss-20b/hf_streaming_eagle3_multi_node.yaml | 21 | # Usage: | COMMENT |
| LOW | …openai/gpt-oss-20b/hf_streaming_dflash_multi_node.yaml | 26 | # Usage: | COMMENT |
| LOW | tools/debugger/client.sh | 21 | # Usage: | COMMENT |
| LOW | plugins/modelopt/skills/common/remote_exec.sh | 19 | # Usage: | COMMENT |
| LOW | …t/skills/evaluation/recipes/examples/example_eval.yaml | 12 | # Usage: | COMMENT |
| LOW | .agents/scripts/sync-upstream-skills.sh | 23 | # Usage: | COMMENT |
| LOW | examples/speculative_decoding/launch_train.sh | 17 | # Usage: | COMMENT |
| LOW | …VIDIA-Nemotron-3-Nano-30B-A3B-BF16/nemo_evaluator.yaml | 17 | # Usage: | COMMENT |
| LOW | …amples/llm_qat/configs/train/lr/lr_config_example.yaml | 14 | # Usage: | COMMENT |
| LOW | …es/diffusers/distillation/configs/accelerate/fsdp.yaml | 6 | # Usage: | COMMENT |
| LOW | examples/llm_eval/run_lm_eval_vllm.sh | 21 | # Usage: | COMMENT |
| 2 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | tools/mcp/modelopt_mcp/bridge.py | 865 | argv += [target, "whoami && hostname"] | CODE |
| HIGH | …ugins/modelopt/skills/day0-release/scripts/gate_run.py | 30 | "score": float | null, # canonical score, if extracted | STRING |
| HIGH | modelopt/torch/speculative/plugins/hf_lilicorr.py | 579 | """``KL(p|K || q)`` from the target's candidate distribution to the head's. | STRING |
| HIGH | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 37 | {"Conv_88 || Conv_59": 0.01}, | CODE |
| HIGH⚡ | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 145 | {"Conv_757 || Conv_635": 0.01, "Conv_757 || Conv_635 output reformatter 0": 0.02}, | CODE |
| HIGH⚡ | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 146 | {"Conv_757 || Conv_635": 0.03}, | CODE |
| HIGH⚡ | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 164 | "Conv_705 || Conv_733": 0.01, | CODE |
| HIGH⚡ | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 165 | "Reformatting CopyNode for Output Tensor 0 to Conv_705 || Conv_733": 0.02, | CODE |
| HIGH⚡ | …h/deploy/_runtime/tensorrt/test_layerwise_profiling.py | 167 | {"Conv_705 || Conv_733": 0.03}, | CODE |
| HIGH⚡ | tests/examples/specdec_bench/test_upload_to_s3.py | 135 | (run / "configuration.json").write_text('{"container_image": null}') | CODE |
| HIGH | …mples/speculative_decoding/scripts/quantize_drafter.py | 127 | be null (``[null, "tokenization_x.XTokenizerFast"]``); a ``repo--`` prefix points at | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | README.md | 40 | - [2026/03/11] Model Optimizer quantized Nemotron-3-Super checkpoints are available on Hugging Face for download: [FP8]( | CODE |
| MEDIUM | tools/launcher/common/query.py | 196 | # Restore the full reasoning trace for the last generated assistant turn. | COMMENT |
| MEDIUM | …gins/modelopt/skills/evaluation/references/nel-next.md | 1 | # nel-next (nemo-evaluator 0.4.x) — shared reference for harbor / agentic benchmarks | COMMENT |
| MEDIUM | …/evaluation/recipes/tasks/aa_next/swebench_verified.md | 96 | ### Deployment proxy (multi-turn agentic) | COMMENT |
| MEDIUM | …lls/evaluation/recipes/examples/example_eval_next.yaml | 1 | # nel-next (nemo-evaluator 0.4.x) agentic AA eval template — NOT the 0.2.6 schema. | COMMENT |
| MEDIUM | …lls/evaluation/recipes/examples/example_eval_next.yaml | 28 | - "--enable-auto-tool-choice" # agentic = tool-calling | CODE |
| MEDIUM | plugins/modelopt/skills/evaluation/scripts/nel-next.sh | 45 | # [harbor] = agentic/sandbox deps; [export] pulls mlflow for `mlflow-push`. | COMMENT |
| MEDIUM | modelopt/torch/utils/plugins/megatron_mmlu.py | 118 | # Group dev examples by subject for few-shot prompt construction. | COMMENT |
| MEDIUM | …t/torch/prune/importance_hooks/expert_removal_hooks.py | 294 | # Compute zero-shot expert ranks (double argsort converts counts to rank positions) | COMMENT |
| MEDIUM | examples/speculative_decoding/README.md | 252 | # Direct SFT training mix (~3.9M rows, includes agentic/tool-use datasets): | COMMENT |
| MEDIUM | …rials/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16/ABLATIONS.md | 4 | > Every benchmark table in this document is from evaluations run **without** eval-time tool calling (no Python sandbox). | CODE |
| MEDIUM | examples/dataset/MEGATRON_DATA_PREP.md | 261 | Datasets below are from the [Nemotron Post-Training v3 collection](https://huggingface.co/collections/nvidia/nemotron-po | CODE |
| MEDIUM | examples/dataset/augmentations.yaml | 85 | # /no_think: suppresses chain-of-thought in models that support it (e.g. Qwen3). | COMMENT |
| MEDIUM | examples/llm_eval/mmlu.py | 207 | # Skip examples that do not fit even at zero-shot, otherwise the backend rejects | COMMENT |
| MEDIUM | examples/llm_eval/modeling.py | 546 | ctx = prompt # Special format has lower few-shot performance | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| CRITICAL | modelopt/torch/fastgen/methods/dmd.py | 108 | "modelopt.torch.fastgen.plugins.qwen_image.attach_feature_capture(teacher, ...) " | CODE |
| CRITICAL | tests/unit/torch/quantization/test_autoquant_shapley.py | 1052 | shared = model.layer.mlp.shared_experts.gate_proj.get_hparam("quant_recipe") | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | modelopt/torch/trace/analyzer.py | 1235 | inputs = [node for node in graph.nodes if node.op == "placeholder"] | CODE |
| LOW⚡ | modelopt/torch/utils/graph.py | 38 | "placeholder": lambda _: None, | CODE |
| LOW | modelopt/torch/utils/graph.py | 78 | if nx.op == "placeholder": | CODE |
| LOW⚡ | tests/_test_utils/torch/puzzletron/utils.py | 247 | {"role": "user", "content": "please cite Lorem Ipsum?"}, | CODE |
| LOW⚡ | tests/_test_utils/torch/puzzletron/utils.py | 251 | "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed in blandit ante. " | CODE |
| LOW⚡ | tests/_test_utils/torch/puzzletron/utils.py | 251 | "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed in blandit ante. " | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | examples/hf_ptq/scripts/parser.sh | 152 | # Now you can use the variables $GPU, $MODEL, and $TASKS in your script | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | modelopt/torch/peft/conversion.py | 27 | # TODO: Add test cases to cover these functions | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | modelopt/torch/nas/autonas.py | 247 | def process_data(data): | CODE |