Home / axolotl-ai-cloud / axolotl
Repository Analysis

axolotl-ai-cloud/axolotl

20.6 Moderate AI signal View on GitHub
20.6
Adjusted Score
20.6
Raw Score
100%
Time Factor
—
Last Push
—
Stars
—
Language
193,462
Lines of Code
1134
Files
2711
Pattern Hits
2026-05-31
Scan Date

Score History

Severity Breakdown

CRITICAL 3HIGH 67MEDIUM 393LOW 2248

Pattern Findings

2711 matches across 20 categories. Click a row to expand file-level details.

Hyper-Verbose Identifiers1575 hits · 1583 pts
SeverityFileLineSnippet
LOWtests/test_fp32_norms.py53def test_suffix_matches_multiple_norm_families():
LOWtests/test_fp32_norms.py60def test_suffix_does_not_match_non_norm_modules():
LOWtests/test_fp32_norms.py67def test_explicit_classname_matches_custom_norm():
LOWtests/test_fp32_norms.py71def test_fully_qualified_pattern_matches_exact_path():
LOWtests/test_fp32_norms.py77def test_mixed_patterns_suffix_and_qualified():
LOWtests/test_fp32_norms.py107def test_meta_device_is_supported(monkeypatch):
LOWtests/test_fp32_norms.py128def test_passthrough_fully_shard_kwargs_are_used(monkeypatch):
LOWtests/test_fp32_norms.py159def test_no_matches_warns_and_returns_zero(caplog):
LOWtests/test_fp32_norms.py177def test_explicit_classes_override_defaults(monkeypatch):
LOWtests/test_utils_tee.py25def test_file_only_stream_writes_after_prepare(monkeypatch):
LOWtests/test_utils_tee.py50def test_stdout_is_mirrored_after_prepare(capsys, monkeypatch):
LOWtests/test_utils_tee.py72def test_truncate_vs_append_behavior(monkeypatch):
LOWtests/conftest.py226# def download_fozzie_alpaca_dpo_dataset():
LOWtests/conftest.py128def download_tiny_llama_model():
LOWtests/conftest.py133def download_tiny_mistral_model():
LOWtests/conftest.py138def download_tiny_mixtral_model():
LOWtests/conftest.py148def download_tiny_falcon_model():
LOWtests/conftest.py202def download_argilla_distilabel_intel_orca_dpo_dataset():
LOWtests/conftest.py210def download_argilla_ultrafeedback_binarized_preferences_cleaned_dataset():
LOWtests/conftest.py218def download_argilla_ultrafeedback_binarized_preferences_cleaned_kto_dataset():
LOWtests/conftest.py240# def dataset_fozzie_alpaca_dpo_dataset(
LOWtests/conftest.py257def download_arcee_ai_distilabel_intel_orca_dpo_pairs_dataset():
LOWtests/conftest.py265def download_argilla_dpo_pairs_dataset():
LOWtests/conftest.py273def download_tiny_shakespeare_dataset():
LOWtests/conftest.py279def download_evolkit_kd_sample_dataset():
LOWtests/conftest.py287def download_deepseek_model_fixture():
LOWtests/conftest.py292def download_huggyllama_model_fixture():
LOWtests/conftest.py56def retry_on_request_exceptions(max_retries=3, delay=1):
LOWtests/conftest.py80def snapshot_download_w_retry(*args, **kwargs):
LOWtests/conftest.py96def download_ds_fixture_bundle():
LOWtests/conftest.py168def download_tatsu_lab_alpaca_dataset():
LOWtests/conftest.py453def tokenizer_huggyllama_w_special_tokens(
LOWtests/conftest.py581def dataset_winglian_tiny_shakespeare(
LOWtests/conftest.py605def dataset_argilla_ultrafeedback_binarized_preferences_cleaned(
LOWtests/test_tokenizers.py44 def test_special_tokens_modules_to_save(self):
LOWtests/test_tokenizers.py79 def test_add_additional_special_tokens(self):
LOWtests/test_tokenizers.py96 def test_added_tokens_overrides(self, temp_dir):
LOWtests/test_tokenizers.py147 def test_added_tokens_overrides_with_toolargeid(self, temp_dir):
LOWtests/test_ebft_kernels.py289 def test_single_sample_returns_zeros(self):
LOWtests/test_triton_kernels.py34def _ref_selective_log_softmax(logits, index):
LOWtests/test_triton_kernels.py157 def test_many_rows_beyond_max_grid(self):
LOWtests/test_triton_kernels.py166 def test_entropy_non_negative(self):
LOWtests/test_triton_kernels.py379 def test_negative_indices_no_crash(self):
LOWtests/test_triton_kernels.py62 def test_correctness_various_shapes(self, B, L):
LOWtests/test_triton_kernels.py91 def test_uniform_distribution(self):
LOWtests/test_triton_kernels.py191 def test_correctness_various_shapes(self, B, L, K):
LOWtests/test_triton_kernels.py270 def test_output_always_nonpositive(self):
LOWtests/test_triton_kernels.py280 def test_many_rows_beyond_max_grid(self):
LOWtests/test_triton_kernels.py306 def test_gradient_matches_reference(self, B, L, V, K):
LOWtests/test_triton_kernels.py395 def test_index_exceeds_vocab_no_crash(self):
LOWtests/test_triton_kernels.py415 def test_mixed_valid_invalid_multi_index(self):
LOWtests/test_triton_kernels.py438 def test_oob_backward_no_crash(self):
LOWtests/test_triton_kernels.py454 def test_oob_backward_valid_rows_correct(self):
LOWtests/test_revision_parameter.py18 def test_load_tokenizer_passes_revision(
LOWtests/test_revision_parameter.py43 def test_load_tokenizer_omits_revision_when_unset(
LOWtests/test_revision_parameter.py65 def test_modify_tokenizer_files_passes_revision(
LOWtests/test_revision_parameter.py81 def test_modify_tokenizer_files_defaults_revision_to_main(
LOWtests/test_revision_parameter.py95 def test_load_processor_passes_revision(self, mock_auto_processor):
LOWtests/test_revision_parameter.py117 def test_load_processor_omits_revision_when_unset(self, mock_auto_processor):
LOWtests/test_revision_parameter.py138 def test_load_processor_forwards_processor_kwargs(self, mock_auto_processor):
1515 more matches not shown…
Decorative Section Separators341 hits · 1145 pts
SeverityFileLineSnippet
MEDIUMtests/test_ebft_kernels.py22# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py24# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py98# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py100# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py164# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py166# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py237# ---------------------------------------------------------------------------
MEDIUMtests/test_ebft_kernels.py239# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py23# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py25# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py174# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py176# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py371# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py373# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py46# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py48# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py291# ---------------------------------------------------------------------------
MEDIUMtests/test_triton_kernels.py293# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py17# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py19# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py61# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py63# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py111# ---------------------------------------------------------------------------
MEDIUMtests/test_http_weight_sync.py113# ---------------------------------------------------------------------------
MEDIUMtests/integrations/test_gemma4_moe.py173# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py175# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py444# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py446# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py22# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py24# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py118# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py120# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py498# ============================================================================
MEDIUMtests/integrations/test_gemma4_moe.py500# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py45# ─── Helpers ─────────────────────────────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py121# ─── Model shape configs ────────────────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py139# ─── Forward tests ──────────────────────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py186# ─── Backward dX tests ──────────────────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py220# ─── Backward LoRA gradient tests ───────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora_kernels.py291# ─── Full autograd tests ────────────────────────────────────────────────────
MEDIUMtests/integrations/test_scattermoe_lora.py24# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py26# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py254# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py256# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py288# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py290# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py634# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py636# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py186# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py188# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py328# ============================================================================
MEDIUMtests/integrations/test_scattermoe_lora.py330# ============================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py11# ---------------------------------------------------------------------------
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py13# ---------------------------------------------------------------------------
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py65# =========================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py67# =========================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py190# =========================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py192# =========================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py364# =========================================================================
MEDIUM…sts/integrations/test_scattermoe_autotune_telemetry.py366# =========================================================================
281 more matches not shown…
Deep Nesting208 hits · 198 pts
SeverityFileLineSnippet
LOW.runpod/src/utils.py40
LOW.runpod/src/utils.py59
LOWtests/core/test_async_grpo.py275
LOWtests/utils/callbacks/test_dynamic_checkpoint.py178
LOWtests/utils/callbacks/test_dynamic_checkpoint.py215
LOWtests/utils/callbacks/test_dynamic_checkpoint.py255
LOWtests/utils/callbacks/test_dynamic_checkpoint.py301
LOW…ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py331
LOWtests/e2e/multigpu/solo/test_grpo.py25
LOWtests/prompt_strategies/test_chat_templates_advanced.py728
LOWtests/prompt_strategies/test_chat_templates_advanced.py1409
LOWdocs/scripts/generate_examples_docs.py94
LOWdocs/scripts/generate_examples_docs.py215
LOWdocs/scripts/generate_examples_docs.py286
LOWdocs/scripts/generate_examples_docs.py107
LOWdocs/scripts/generate_config_docs.py50
LOWdocs/scripts/generate_config_docs.py126
LOWdocs/scripts/generate_config_docs.py263
LOWdocs/scripts/generate_config_docs.py438
LOWdocs/scripts/generate_config_docs.py591
LOWscripts/analyze_profile.py338
LOWscripts/analyze_profile.py565
LOWscripts/analyze_profile.py651
LOWscripts/analyze_profile.py774
LOWscripts/analyze_profile.py888
LOWscripts/analyze_profile.py975
LOWscripts/analyze_profile.py1079
LOWscripts/analyze_profile.py1338
LOWsrc/axolotl/processing_strategies.py378
LOWsrc/axolotl/processing_strategies.py493
LOWsrc/axolotl/processing_strategies.py174
LOWsrc/axolotl/prompters.py48
LOWsrc/axolotl/train.py254
LOWsrc/axolotl/train.py391
LOWsrc/axolotl/kernels/autotune_telemetry.py46
LOWsrc/axolotl/core/chat/messages.py134
LOWsrc/axolotl/core/builders/rl.py39
LOWsrc/axolotl/core/builders/rl.py101
LOWsrc/axolotl/core/builders/causal.py452
LOWsrc/axolotl/core/builders/base.py289
LOWsrc/axolotl/core/builders/base.py290
LOWsrc/axolotl/core/trainers/utils.py4
LOWsrc/axolotl/core/trainers/utils.py20
LOW…olotl/core/trainers/mixins/activation_checkpointing.py73
LOWsrc/axolotl/core/trainers/mixins/layer_offloading.py24
LOWsrc/axolotl/core/trainers/mixins/optimizer.py22
LOWsrc/axolotl/core/trainers/mixins/optimizer.py107
LOWsrc/axolotl/core/trainers/mixins/scheduler.py27
LOWsrc/axolotl/core/trainers/ebft/__init__.py79
LOWsrc/axolotl/core/trainers/ebft/trainer.py144
LOWsrc/axolotl/core/trainers/ebft/trainer.py186
LOWsrc/axolotl/core/trainers/ebft/trainer.py429
LOWsrc/axolotl/core/trainers/ebft/strided.py252
LOWsrc/axolotl/core/trainers/ebft/strided.py334
LOWsrc/axolotl/core/trainers/ebft/strided.py562
LOWsrc/axolotl/core/trainers/ebft/strided.py859
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py526
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py692
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py970
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py1205
148 more matches not shown…
Unused Imports186 hits · 184 pts
SeverityFileLineSnippet
LOWtests/test_fp32_norms.py3
LOWtests/test_no_legacy_attn_reads.py8
LOWtests/test_attn_implementation.py42
LOWtests/utils/test_cuda13.py3
LOWtests/integrations/test_gemma4_moe.py451
LOW…rnels/scattermoe_lora/test_scattermoe_lora_m_bucket.py22
LOWtests/integrations/kernels/scattermoe_lora/conftest.py19
LOW…/scattermoe_lora/test_scattermoe_lora_int64_indices.py33
LOW…ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py25
LOW…grations/kernels/scattermoe_lora/bench_int64_kernel.py26
LOW…termoe_lora/test_parallel_experts_large_batch_repro.py32
LOWtests/e2e/multigpu/test_fsdp2_fp32_norms.py11
LOWtests/e2e/multigpu/_fp32_norms_dtype_capture.py9
LOWtests/prompt_strategies/test_chat_templates_mistral.py9
LOWsrc/axolotl/train.py3
LOWsrc/axolotl/train.py46
LOWsrc/axolotl/train.py46
LOWsrc/axolotl/core/training_args.py5
LOWsrc/axolotl/core/builders/__init__.py3
LOWsrc/axolotl/core/builders/__init__.py4
LOWsrc/axolotl/core/trainers/__init__.py7
LOWsrc/axolotl/core/trainers/base.py3
LOWsrc/axolotl/core/trainers/mixins/__init__.py5
LOWsrc/axolotl/core/trainers/mixins/__init__.py6
LOWsrc/axolotl/core/trainers/mixins/__init__.py7
LOWsrc/axolotl/core/trainers/mixins/__init__.py8
LOWsrc/axolotl/core/trainers/mixins/__init__.py9
LOWsrc/axolotl/core/trainers/mixins/__init__.py10
LOWsrc/axolotl/core/trainers/mixins/__init__.py11
LOWsrc/axolotl/core/trainers/mixins/__init__.py12
LOWsrc/axolotl/core/trainers/ebft/strided.py24
LOWsrc/axolotl/core/trainers/ebft/strided.py24
LOWsrc/axolotl/core/trainers/ebft/strided.py24
LOWsrc/axolotl/core/trainers/ebft/strided.py24
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py24
LOWsrc/axolotl/core/trainers/grpo/trainer.py54
LOWsrc/axolotl/utils/tee.py6
LOWsrc/axolotl/utils/cuda13.py3
LOWsrc/axolotl/utils/fp32_norms.py3
LOWsrc/axolotl/utils/ctx_managers/__init__.py5
LOWsrc/axolotl/utils/callbacks/swanlab.py3
LOWsrc/axolotl/utils/callbacks/__init__.py3
LOWsrc/axolotl/utils/callbacks/comet_.py12
LOWsrc/axolotl/utils/callbacks/lisa.py18
LOWsrc/axolotl/utils/callbacks/mlflow_.py15
LOWsrc/axolotl/utils/callbacks/trackio_.py13
LOWsrc/axolotl/utils/optimizers/qgalore.py3
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/optimizers/adopt.py16
LOWsrc/axolotl/utils/schemas/validation.py1614
LOWsrc/axolotl/utils/chat_templates/__init__.py6
LOWsrc/axolotl/utils/chat_templates/__init__.py6
LOWsrc/axolotl/utils/chat_templates/__init__.py6
LOWsrc/axolotl/utils/chat_templates/__init__.py6
LOWsrc/axolotl/utils/chat_templates/__init__.py6
126 more matches not shown…
Magic Placeholder Names23 hits · 161 pts
SeverityFileLineSnippet
HIGHexamples/swanlab/dpo-swanlab-full-featured.yml13# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/dpo-swanlab-full-featured.yml145# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/dpo-swanlab-full-featured.yml147# swanlab_api_key: your-api-key
HIGHexamples/swanlab/lora-swanlab-profiling.yml12# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/lora-swanlab-profiling.yml92# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/lora-swanlab-profiling.yml94# swanlab_api_key: your-api-key
HIGHexamples/swanlab/README.md22export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/README.md43export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/README.md66export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/README.md149export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/README.md156swanlab_api_key: your-api-key
HIGHexamples/swanlab/custom_trainer_profiling.py247 export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/dpo-swanlab-completions.yml12# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/dpo-swanlab-completions.yml104# export SWANLAB_API_KEY=your-api-key
HIGHexamples/swanlab/dpo-swanlab-completions.yml106# swanlab_api_key: your-api-key
HIGHsrc/axolotl/integrations/swanlab/plugins.py88 " export SWANLAB_API_KEY=your-api-key\n"
HIGHsrc/axolotl/integrations/swanlab/plugins.py89 " 2. Add 'swanlab_api_key: your-api-key' to config (less secure)\n"
HIGHsrc/axolotl/integrations/swanlab/README.md46swanlab_api_key: YOUR_API_KEY # Optional: can also use env var SWANLAB_API_KEY
HIGHsrc/axolotl/integrations/swanlab/README.md53export SWANLAB_API_KEY=your-api-key-here
HIGHsrc/axolotl/integrations/swanlab/README.md8411. Set environment variable: `export SWANLAB_API_KEY=your-api-key`
HIGHsrc/axolotl/integrations/swanlab/README.md8422. Add to config (less secure): `swanlab_api_key: your-api-key`
HIGHsrc/axolotl/integrations/swanlab/README.md938export SWANLAB_API_KEY=your-api-key-here
HIGHsrc/axolotl/integrations/swanlab/README.md949swanlab_api_key: your-api-key-here
Excessive Try-Catch Wrapping105 hits · 121 pts
SeverityFileLineSnippet
LOWtests/core/test_builders.py604 except Exception:
MEDIUMtests/integrations/test_gemma4_moe.py449def _can_import_sonicmoe():
LOWtests/integrations/test_gemma4_moe.py454 except Exception:
LOW…ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py65 except Exception:
MEDIUM…ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py56def gpu_name() -> str:
LOW…grations/kernels/scattermoe_lora/bench_int64_kernel.py59 except Exception:
MEDIUM…grations/kernels/scattermoe_lora/bench_int64_kernel.py51def gpu_name() -> str:
MEDIUMtests/e2e/integrations/test_scattermoe_lora_kernels.py42def wrapper(*args, **kwargs):
LOWtests/e2e/integrations/test_scattermoe_lora_kernels.py45 except Exception as exc: # pylint: disable=broad-except
LOWdocs/scripts/generate_config_docs.py503 except Exception:
LOWdocs/scripts/generate_config_docs.py607 except Exception as e:
LOWexamples/orpheus/README.md79 except Exception as e:
LOWexamples/orpheus/README.md240 except Exception as e:
LOWscripts/analyze_profile.py1388 except Exception:
LOWsrc/axolotl/train.py91 except Exception: # pylint: disable=broad-exception-caught
LOWsrc/axolotl/kernels/autotune_telemetry.py42 except Exception: # pylint: disable=broad-exception-caught
LOWsrc/axolotl/core/trainers/ebft/trainer.py501 except Exception as e:
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py1177 except Exception as exc:
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py207 except Exception:
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py336 except Exception:
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py340 except Exception:
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py444 except Exception:
LOWsrc/axolotl/utils/tee.py41 except Exception:
LOWsrc/axolotl/utils/tee.py70 except Exception:
LOWsrc/axolotl/utils/tee.py162 except Exception:
LOWsrc/axolotl/utils/config/__init__.py48 except Exception:
MEDIUMsrc/axolotl/utils/config/__init__.py36def get_device():
LOWsrc/axolotl/utils/callbacks/generation.py38 except Exception as e:
LOWsrc/axolotl/utils/callbacks/swanlab.py90 except Exception as err:
LOWsrc/axolotl/utils/callbacks/swanlab.py124 except Exception as err:
LOWsrc/axolotl/utils/callbacks/__init__.py374 except Exception as exc:
LOWsrc/axolotl/utils/callbacks/__init__.py435 except Exception:
LOWsrc/axolotl/utils/callbacks/opentelemetry.py72 except Exception as e:
LOWsrc/axolotl/utils/callbacks/opentelemetry.py132 except Exception as e:
LOWsrc/axolotl/utils/callbacks/opentelemetry.py221 except Exception as e:
LOWsrc/axolotl/utils/schemas/config.py1799 except Exception: # pylint: disable=broad-exception-caught
LOWsrc/axolotl/utils/data/shared.py503 except Exception:
LOWsrc/axolotl/utils/generation/sft.py128 except Exception as e:
LOWsrc/axolotl/utils/generation/sft.py132 except Exception as e:
LOWsrc/axolotl/cli/config.py158 except Exception as err:
LOWsrc/axolotl/cli/preprocess.py88 except Exception: # nosec B110
MEDIUMsrc/axolotl/cli/utils/fetch.py66 print(f"Error downloading {file_path}: {str(request_error)}")
MEDIUMsrc/axolotl/cli/utils/fetch.py133 print(f"Error processing {file_path}: {str(request_error)}")
LOWsrc/axolotl/cli/utils/diffusion.py116 except Exception:
LOWsrc/axolotl/cli/utils/diffusion.py127 except Exception:
LOWsrc/axolotl/cli/utils/lora_merge.py52 except Exception:
LOWsrc/axolotl/cli/utils/lora_merge.py78 except Exception: # noqa: BLE001
MEDIUMsrc/axolotl/integrations/base.py418 print(f"Error: {exc}")
LOWsrc/axolotl/integrations/base.py746 except Exception:
LOWsrc/axolotl/integrations/nemo_gym/plugin.py88 except Exception as exc:
LOWsrc/axolotl/integrations/nemo_gym/plugin.py218 except Exception as exc:
LOWsrc/axolotl/integrations/nemo_gym/plugin.py575 except Exception:
LOWsrc/axolotl/integrations/nemo_gym/plugin.py662 except Exception as exc:
LOWsrc/axolotl/integrations/nemo_gym/rewards.py104 except Exception as exc:
LOWsrc/axolotl/integrations/nemo_gym/rewards.py168 except Exception as exc:
LOWsrc/axolotl/integrations/nemo_gym/data_producer.py141 except Exception:
LOWsrc/axolotl/integrations/nemo_gym/data_producer.py249 except Exception as exc: # never let metric logging break training
LOWsrc/axolotl/integrations/kernels/autotune_callback.py31 except Exception: # pylint: disable=broad-exception-caught
LOWsrc/axolotl/integrations/kernels/autotune_callback.py49 except Exception: # pylint: disable=broad-exception-caught
LOWsrc/axolotl/integrations/kd/collator_online_teacher.py253 except Exception as e: # Catch other potential errors during processing
45 more matches not shown…
Redundant / Tautological Comments69 hits · 108 pts
SeverityFileLineSnippet
LOW.axolotl-complete.bash18 # Check if previous word is in our list
LOWtests/e2e/integrations/test_scattermoe_lora_kernels.py809 # Set weights to match test data
LOWtests/e2e/patched/test_peft_embeddings.py59 # Check if the embeddings are upcast correctly
LOWtests/prompt_strategies/test_chat_templates_advanced.py644 # Check if system message is not present in input_ids
LOWtests/prompt_strategies/test_chat_templates_advanced.py715 # Check if responses are labeled correctly based on role
LOWdocs/scripts/generate_examples_docs.py123 # Check if target is outside examples_dir
LOWdocs/scripts/generate_examples_docs.py153 # Check if this exact path is in allowlist
LOWdocs/scripts/generate_examples_docs.py243 # Check if this is a parent page that has sub-pages
LOWdocs/scripts/generate_config_docs.py180 # Check if this type itself is a Pydantic model
LOWdocs/scripts/generate_config_docs.py395 # Check if this starts a new group (blank line before or significant gap)
LOWdocs/scripts/generate_config_docs.py674 # Check if this field has nested models
LOWscripts/cloud-entrypoint.sh29 # Check if the word looks like the start of a key
LOWscripts/cloud-entrypoint.sh72# Check if JUPYTER_PASSWORD is set and not empty
LOWscripts/cloud-entrypoint.sh74 # Set JUPYTER_TOKEN to the value of JUPYTER_PASSWORD
LOWscripts/analyze_profile.py1110 # Check if this small block is surrounded by inactive
LOWscripts/cloud-entrypoint-term.sh25 # Check if the word looks like the start of a key
LOWscripts/cloud-entrypoint-term.sh63# Check if JUPYTER_PASSWORD is set and not empty
LOWscripts/cloud-entrypoint-term.sh65 # Set JUPYTER_TOKEN to the value of JUPYTER_PASSWORD
LOWsrc/axolotl/train.py506 # Check if the `token_ids_to_fix` kwarg exists in the fix_untrained_tokens args
LOWsrc/axolotl/core/trainers/ebft/strided.py408 # Check if we can share weights (PEFT model on single GPU)
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py1274 # Check if model has FP8 weights
LOWsrc/axolotl/utils/tee.py157 # Close the file handle if open
LOWsrc/axolotl/utils/freeze.py39 # Check if any path component matches a vision/audio prefix
LOWsrc/axolotl/utils/trainer.py48 # Check if the tensor is 2D. If not, unsqueeze it to make it 2D
LOWsrc/axolotl/utils/trainer.py230 # Check if single example or batched by looking at the first element
LOWsrc/axolotl/utils/trainer.py276 # Check if single example or batch
LOWsrc/axolotl/utils/callbacks/swanlab.py48 # Check if SwanLab run is initialized
LOWsrc/axolotl/utils/callbacks/swanlab.py163 # Check if SwanLab is initialized
LOWsrc/axolotl/utils/callbacks/lisa.py59 # Check if it's time to switch active layers, including at step 0
LOWsrc/axolotl/utils/schemas/datasets.py210 # Set chat_template to tokenizer_default if not set
LOWsrc/axolotl/utils/samplers/multipack.py237 # Assign bins to ranks - each rank gets every n-th bin
LOWsrc/axolotl/utils/data/shared.py120 # Check if it's a HuggingFace dataset
LOWsrc/axolotl/utils/data/shared.py123 # Check if it's a cloud storage path and get appropriate filesystem
LOWsrc/axolotl/utils/data/streaming.py209 # Set this to 1 so downstream data_loader doesn't try to increase the batch size
LOWsrc/axolotl/utils/mistral/mistral_tokenizer.py62 # Check if MistralRequestValidator has a _mode attribute.
LOWsrc/axolotl/cli/config.py115 # Check if the config is a valid HTTPS URL to a .yml or .yaml file
LOWsrc/axolotl/cli/checks.py42 # Verify if token is valid
LOWsrc/axolotl/cli/utils/fetch.py35 # Check if file exists and needs updating
LOWsrc/axolotl/cli/utils/train.py27 # Check if rdzv_endpoint is present
LOWsrc/axolotl/cli/utils/train.py31 # Check if rdzv_backend is already provided
LOWsrc/axolotl/cli/utils/train.py36 # Check if rdzv_id is already provided
LOWsrc/axolotl/integrations/nemo_gym/server.py93 # Check if already running
LOW…ions/kernels/libs/scattermoe_lora/selective_dequant.py260 # Check if the parameter is BnB-quantized via parametrize
LOWsrc/axolotl/integrations/cut_cross_entropy/__init__.py61 # Check if cut_cross_entropy is installed
LOWsrc/axolotl/integrations/cut_cross_entropy/__init__.py74 # Check if Axolotl's cce fork is installed
LOWsrc/axolotl/integrations/swanlab/plugins.py50 # Check if SwanLab is enabled
LOWsrc/axolotl/integrations/swanlab/plugins.py446 # Check if Lark webhook URL is configured
LOWsrc/axolotl/integrations/swanlab/plugins.py503 # Check if completion logging is enabled
LOWsrc/axolotl/integrations/swanlab/plugins.py509 # Check if trainer is an RLHF trainer
LOWsrc/axolotl/integrations/swanlab/profiling.py40 # Check if SwanLab is enabled and initialized
LOWsrc/axolotl/integrations/swanlab/profiling.py184 # Check if should log based on config
LOWsrc/axolotl/integrations/swanlab/profiling.py186 # Check if SwanLab is enabled
LOWsrc/axolotl/scripts/vllm_worker_ext.py67 # Check if this is a simple direct param (exists as-is)
LOWsrc/axolotl/telemetry/callbacks.py102 # Check if we should report metrics
LOWsrc/axolotl/telemetry/errors.py50 # Check if this line contains a file path
LOWsrc/axolotl/loaders/utils.py77 # Check if image_size is not set and load image size from model config if available
LOWsrc/axolotl/monkeypatch/deepspeed_utils.py33 # Check if already patched
LOWsrc/axolotl/monkeypatch/lora_kernels.py244 # Check if already patched
LOWsrc/axolotl/monkeypatch/attention/xformers.py45 # Check if we have sliding window attention
LOW…/axolotl/monkeypatch/transformers/trainer_loss_calc.py49 # Check if already patched
9 more matches not shown…
Cross-File Repetition21 hits · 105 pts
SeverityFileLineSnippet
HIGHtests/e2e/multigpu/test_fsdp2.py0verify that training completed successfully — artifacts, no-nan, loss stayed in qwen2-pretraining scale (tiny-qwen2-129m
HIGHtests/e2e/multigpu/test_dist_muon_fsdp2.py0verify that training completed successfully — artifacts, no-nan, loss stayed in qwen2-pretraining scale (tiny-qwen2-129m
HIGHtests/e2e/multigpu/test_fsdp1.py0verify that training completed successfully — artifacts, no-nan, loss stayed in qwen2-pretraining scale (tiny-qwen2-129m
HIGHsrc/axolotl/utils/collators/batching.py0collator for multipack specific to the using the batchsampler
HIGHsrc/axolotl/utils/collators/batching.py0collator for multipack specific to the using the batchsampler
HIGHsrc/axolotl/utils/collators/batching.py0collator for multipack specific to the using the batchsampler
HIGHsrc/axolotl/integrations/liger/models/llama4.py0args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked
HIGHsrc/axolotl/integrations/liger/models/qwen3_moe.py0args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked
HIGHsrc/axolotl/integrations/liger/models/qwen3_5.py0args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked
HIGHsrc/axolotl/integrations/liger/models/qwen3_5_moe.py0args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked
HIGHsrc/axolotl/integrations/liger/models/qwen3.py0args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked
HIGHsrc/axolotl/integrations/liger/models/llama4.py0apply liger kernels to replace original implementation in huggingface llama models (2 and 3) args: cross_entropy (bool):
HIGHsrc/axolotl/integrations/liger/models/qwen3_moe.py0apply liger kernels to replace original implementation in huggingface llama models (2 and 3) args: cross_entropy (bool):
HIGHsrc/axolotl/integrations/liger/models/qwen3.py0apply liger kernels to replace original implementation in huggingface llama models (2 and 3) args: cross_entropy (bool):
HIGHsrc/axolotl/monkeypatch/models/qwen3/fused_attn.py0unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight.
HIGH…c/axolotl/monkeypatch/models/qwen3_5_moe/fused_attn.py0unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight.
HIGHsrc/axolotl/monkeypatch/models/qwen3_5/fused_attn.py0unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight.
HIGHsrc/axolotl/monkeypatch/models/qwen3_moe/fused_attn.py0unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight.
HIGHsrc/axolotl/prompt_strategies/bradley_terry/llama3.py0chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab
HIGHsrc/axolotl/prompt_strategies/dpo/chatml.py0chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab
HIGHsrc/axolotl/prompt_strategies/dpo/llama3.py0chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab
Self-Referential Comments30 hits · 92 pts
SeverityFileLineSnippet
MEDIUMtests/conftest.py504 # Create a temporary directory
MEDIUMtests/test_packed_pretraining.py38 # Create a list of 2000 random texts rather than just using it within the
MEDIUMtests/test_packed_pretraining.py42 # Create an IterableDataset
MEDIUMtests/integrations/test_diffusion.py43 # Create a minimal trainer instance just for testing methods
MEDIUMtests/e2e/utils.py23 # Create a temporary directory
MEDIUMtests/e2e/kernels/test_lora.py39 # Create main state with nested state
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py684 # Create a new dataset with modified role names
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py758 # Create a new dataset with the train and train_detail fields
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py1214 # Create a situation where eot_tokens contains eos_token
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py1299 # Create a non-existent token that definitely won't be in the template
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py1350 # Create a custom chat template that uses our EOT token
MEDIUMtests/prompt_strategies/test_chat_templates_advanced.py1425 # Create a list to test different train_on_eot settings
MEDIUMdocs/scripts/generate_examples_docs.py83 # Create a unique asset path based on source directory name
MEDIUMscripts/cloud-entrypoint.sh21 # Create the authorized_keys file if it doesn't exist
MEDIUMscripts/cloud-entrypoint-term.sh17 # Create the authorized_keys file if it doesn't exist
MEDIUMsrc/axolotl/train.py195 # Define the context managers to use
MEDIUMsrc/axolotl/core/trainers/grpo/trainer.py245 # Create the dataloader
MEDIUMsrc/axolotl/utils/tokenization.py90 # Create a delimiter based on text_only flag
MEDIUMsrc/axolotl/utils/trainer.py56 # Create a tensor to track group ids
MEDIUMsrc/axolotl/utils/ctx_managers/sequence_parallel.py74 # Create a boolean mask tensor for this rank's chunk
MEDIUMsrc/axolotl/utils/ctx_managers/sequence_parallel.py224 # Create a partially applied version of the apply_sequence_parallelism function
MEDIUMsrc/axolotl/utils/schemas/utils.py31 data = data.copy() # Create a copy to avoid modifying the original
MEDIUMsrc/axolotl/utils/samplers/multipack.py100 # Create a new bin if needed and if we haven't reached the limit
MEDIUMsrc/axolotl/cli/delinearize_llama4.py117 # Create a new state dict for the converted model
MEDIUMsrc/axolotl/loaders/tokenizer.py54 # Create the tokenizer directory in output_dir if it doesn't exist
MEDIUMsrc/axolotl/monkeypatch/fsdp2_qlora.py29 # Define the replacement
MEDIUMsrc/axolotl/monkeypatch/fsdp2_qlora.py106 # Define the replacement
MEDIUMsrc/axolotl/monkeypatch/attention/flex_attn.py35 # Create a new instance if one doesn't already exist
MEDIUM…onkeypatch/models/mistral3/mistral_common_tokenizer.py22 # Define the replacement
MEDIUMsrc/axolotl/prompt_strategies/chat_template.py362 # Create a new list to store tokens that should be kept
Over-Commented Block89 hits · 82 pts
SeverityFileLineSnippet
LOW.runpod/requirements.txt1# Required Python packages get listed here, one per line.
LOW.runpod/src/config/config.yaml1# # This is the huggingface model that contains *.pt, *.safetensors, or *.bin files
LOW.runpod/src/config/config.yaml21# tokenizer_use_fast:
LOW.runpod/src/config/config.yaml41
LOW.runpod/src/config/config.yaml61# float16: true
LOW.runpod/src/config/config.yaml81# - path: repo
LOW.runpod/src/config/config.yaml101
LOW.runpod/src/config/config.yaml121# dataset_shard_num:
LOW.runpod/src/config/config.yaml141# sample_packing_eff_est:
LOW.runpod/src/config/config.yaml161# # - gate_proj
LOW.runpod/src/config/config.yaml181# # Must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed
LOW.runpod/src/config/config.yaml201# # Training hyperparameters
LOW.runpod/src/config/config.yaml221# max_steps:
LOW.runpod/src/config/config.yaml241# lr_scheduler: # 'one_cycle' | empty for cosine
LOW.runpod/src/config/config.yaml261# # - adafactor
LOW.runpod/src/config/config.yaml281
LOW.runpod/src/config/config.yaml301# # xpos RoPE see https://github.com/kaiokendev/cutoff-len-is-context-len/blob/main/util/xpos_rope_llama_monkey_patch.py
LOW.runpod/src/config/config.yaml321# # Add extra tokens.
LOWtests/conftest.py221 "argilla/ultrafeedback-binarized-preferences-cleaned-kto", repo_type="dataset"
LOWtests/conftest.py241# download_fozzie_alpaca_dpo_dataset,
LOWtests/integrations/test_swanlab.py1# Copyright 2024 Axolotl AI. All rights reserved.
LOW…s/kernels/scattermoe_lora/test_mxfp4_expert_weights.py61# Per-shape Strategy-B tolerances. Forward outputs accumulate K dot-products
LOWtests/e2e/multigpu/solo/__init__.py1# Tests under this directory should get run "solo" on their own as they
LOWexamples/ebft/qwen35-4b-ebft-structured-async.yaml1# EBFT Structured Mode: Qwen3.5-4B (hybrid linear attention)
LOWexamples/ebft/qwen35-4b-ebft-structured.yaml1# EBFT Structured Mode: Qwen3.5-4B (hybrid linear attention)
LOWexamples/ebft/llama-1b-ebft-opencode.yaml1# EBFT: Energy-Based Fine-Tuning with Llama-3.2-1B on OpenCodeInstruct
LOWexamples/swanlab/dpo-swanlab-full-featured.yml1# SwanLab Full-Featured DPO Training Example
LOWexamples/swanlab/dpo-swanlab-full-featured.yml141# Workspace for team collaboration (shared experiments)
LOWexamples/swanlab/dpo-swanlab-full-featured.yml161swanlab_completion_log_interval: 100 # Log every 100 steps
LOWexamples/swanlab/dpo-swanlab-full-featured.yml181# export SWANLAB_LARK_WEBHOOK_URL=https://open.feishu.cn/...
LOWexamples/swanlab/dpo-swanlab-full-featured.yml201# - Identify bottlenecks in training loop
LOWexamples/swanlab/dpo-swanlab-full-featured.yml221
LOWexamples/swanlab/dpo-swanlab-full-featured.yml241# Expected Training Behavior
LOWexamples/swanlab/dpo-swanlab-full-featured.yml261# INFO: Only rank 0 will initialize SwanLab
LOWexamples/swanlab/dpo-swanlab-full-featured.yml281# ============================================================================
LOWexamples/swanlab/dpo-swanlab-full-featured.yml301# If SwanLab initialization fails:
LOWexamples/swanlab/dpo-swanlab-full-featured.yml321# 1. Verify use_swanlab is true
LOWexamples/swanlab/lora-swanlab-profiling.yml1# SwanLab LoRA Training Example with Performance Profiling
LOWexamples/swanlab/lora-swanlab-profiling.yml101# ============================================================================
LOWexamples/swanlab/lora-swanlab-profiling.yml121
LOWexamples/swanlab/lora-swanlab-profiling.yml141# Then compare profiling metrics in SwanLab dashboard to see performance impact
LOWexamples/swanlab/lora-swanlab-profiling.yml161#
LOWexamples/swanlab/dpo-swanlab-completions.yml1# SwanLab DPO Training Example with Completion Logging
LOWexamples/swanlab/dpo-swanlab-completions.yml101
LOWexamples/swanlab/dpo-swanlab-completions.yml121# View the table in SwanLab dashboard under "rlhf_completions"
LOWexamples/swanlab/dpo-swanlab-completions.yml141# Get real-time training notifications in your team chat
LOWexamples/swanlab/dpo-swanlab-completions.yml161
LOW.github/workflows/base.yml81 - cuda: "130"
LOWsrc/axolotl/kernels/lora.py1061 Q_combined = Q_combined.view(-1, Q_combined.shape[-1])
LOWsrc/axolotl/core/training_args_base.py201 chat_template: Optional[str] = field(
LOWsrc/axolotl/core/builders/base.py1# Copyright 2024 Axolotl AI. All rights reserved.
LOWsrc/axolotl/core/trainers/grpo/async_trainer.py1321 # sync to avoid merge_adapter (fails on FP8 / races with training).
LOWsrc/axolotl/core/trainers/grpo/fast_async_trainer.py1# Copyright 2020-2026 Axolotl AI. All rights reserved.
LOWsrc/axolotl/utils/lora.py1# Copyright 2025 Axolotl AI. All rights reserved.
LOWsrc/axolotl/integrations/config.py1# Copyright 2024 Axolotl AI. All rights reserved.
LOWsrc/axolotl/integrations/base.py1# Copyright 2024 Axolotl AI. All rights reserved.
LOW…tl/integrations/nemo_gym/examples/nemo_gym_sudoku.yaml1# Axolotl + NeMo Gym: Sudoku RL Training Example
LOW…ntegrations/nemo_gym/examples/nemo_gym_multi_turn.yaml1# Axolotl + NeMo Gym: Multi-Turn Tool-Use RL Training Example
LOW…integrations/nemo_gym/examples/nemo_gym_multi_env.yaml1# Axolotl + NeMo Gym: Multi-Environment RL Training Example
LOW…tions/kernels/libs/scattermoe_lora/parallel_experts.py1# SPDX-License-Identifier: Apache-2.0
29 more matches not shown…
Docstring Block Structure15 hits · 75 pts
SeverityFileLineSnippet
HIGHsrc/axolotl/kernels/quantize.py64 Fast NF4 dequantization using `bitsandbytes` CUDA kernels. Performs efficient dequantization of weights from N
HIGHsrc/axolotl/core/trainers/grpo/__init__.py249 Returns the reward function from the given fully qualified name, or the path to the reward function model.
HIGHsrc/axolotl/utils/quantization.py61 This function is used to build a post-training quantization config. Args: weight_dtype: The dtype to u
HIGHsrc/axolotl/utils/schemas/utils.py9 Handle backwards compatibility between legacy message field mapping and new property mapping system. Previousl
HIGHsrc/axolotl/utils/chat_templates/base.py31 Finds the correct chat_template based on the user's choice, jinja_template, and tokenizer. Args: user_
HIGHsrc/axolotl/utils/data/rl.py129Filter out samples that exceed maximum sequence length. Args: sample: Dataset sample to check. rl:
HIGHsrc/axolotl/utils/data/rl.py186Check sequence length and raise ValueError if exceeded. Used as a filter function for ``excess_length_strategy: rai
HIGHsrc/axolotl/cli/config.py98 First, determines if the passed config is a valid HTTPS URL. Then, attempts to query for it and parse its conte
HIGHsrc/axolotl/cli/config.py164 Helper method for choosing a `axolotl` config YAML file (considering only files ending with `.yml` or `.yaml`).
HIGHsrc/axolotl/cli/agent_docs/__init__.py84Return the content of an agent doc by topic name. Args: topic: One of the keys in TOPICS, or "overview" (de
HIGHsrc/axolotl/integrations/base.py316Loads a plugin based on the given plugin name. The plugin name should be in the format "module_name.class_name". Th
HIGHsrc/axolotl/integrations/expert_parallel/shard.py79Slice expert weights along dim 0 per the EP rank. Args: model: A built (but not yet FSDP-wrapped) HuggingFa
HIGHsrc/axolotl/loaders/utils.py152Loads and configures a model configuration from HuggingFace or local sources. This function determines the appropri
HIGHsrc/axolotl/monkeypatch/lora_kernels.py153 Get the appropriate attention class by inspecting the model config. Uses dynamic import to support any model ar
HIGHsrc/axolotl/monkeypatch/lora_kernels.py380 Applies optimized Triton kernel patches to a PEFT model. Patches a PEFT model with optimized implementations f
Hallucination Indicators3 hits · 35 pts
SeverityFileLineSnippet
CRITICAL…tions/kernels/libs/scattermoe_lora/kernels/lora_ops.py184 props = triton.runtime.driver.active.utils.get_device_properties(
CRITICALsrc/axolotl/monkeypatch/attention/xformers.py72 xformers.ops.fmha.attn_bias.BlockDiagonalCausalMask.from_seqlens(
CRITICALsrc/axolotl/monkeypatch/attention/xformers.py94 attn_bias = xformers.ops.fmha.attn_bias.BlockDiagonalCausalMask.from_seqlens(
Cross-Language Confusion7 hits · 30 pts
SeverityFileLineSnippet
HIGHsrc/axolotl/core/trainers/grpo/async_trainer.py803 VLLMGeneration._init_vllm = _orig_init_vllm # type: ignore[possibly-undefined]
HIGHsrc/axolotl/cli/cloud/modal_.py116 f"RUN cd /workspace/axolotl && git fetch && git checkout {self.config.branch} && git pull",
HIGHsrc/axolotl/integrations/nemo_gym/server.py69 ["bash", "-c", "source .venv/bin/activate && uv sync"],
HIGHsrc/axolotl/integrations/kd/kernels/liger.py36 0.0 for Forward KL (P_teacher || P_student).
HIGHsrc/axolotl/integrations/kd/kernels/liger.py37 1.0 for Reverse KL (P_student || P_teacher).
HIGHsrc/axolotl/integrations/cut_cross_entropy/__init__.py38 '`pip uninstall -y cut-cross-entropy && pip install "cut-cross-entropy[transformers] @ git+https://github.com/axolot
HIGH…xolotl/monkeypatch/models/kimi_linear/modeling_kimi.py45 "Plese run `pip uninstall fla-core flash-linear-attention -y && pip install git+https://github.com/fla-org/flash
AI Slop Vocabulary9 hits · 22 pts
SeverityFileLineSnippet
MEDIUMpyproject.toml62 # lm eval harness
MEDIUM…ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py296# Timing harness
MEDIUMexamples/archived/xgen-7b/xgen-7b-8k-qlora.yml31# "We find hyperparameters to be largely robust across datasets"
MEDIUMexamples/archived/falcon/config-7b-qlora.yml33# "We find hyperparameters to be largely robust across datasets"
LOWsrc/axolotl/cli/utils/sweeps.py61 # If no paired values, just use regular combinations
LOWsrc/axolotl/integrations/kd/collator.py211 # just pass straight to parent.
LOWsrc/axolotl/prompt_strategies/chat_template.py642 # grid info, etc.); just set the fields we computed locally.
LOWsrc/axolotl/prompt_strategies/ebft/ebft_strided_chat.py20 # For preview: just return the first user message
MEDIUMsrc/axolotl/prompt_strategies/ebft/ebft_reasoning.py280 # directly in the input_ids (robust to tokenization alignment).
Slop Phrases10 hits · 15 pts
SeverityFileLineSnippet
LOW.runpod/src/config/config.yaml65# # HuggingFace dataset repo | s3://,gs:// path | "json" for local dataset, make sure to fill data_files
LOWsrc/axolotl/utils/callbacks/profiler.py89 # make sure to record if we happen to have more steps than steps to profile
LOWsrc/axolotl/cli/utils/train.py187 # make sure to flush stdout and stderr before replacing the process
LOWsrc/axolotl/cli/utils/train.py213 # make sure to flush stdout and stderr before replacing the process
LOWsrc/axolotl/integrations/kd/collator_online_teacher.py531 # TODO make sure to figure out which input in sub_batch_features to update the batch in the original
LOWsrc/axolotl/integrations/liger/models/qwen3_moe.py119 ) # make sure to reside in the same device
LOWsrc/axolotl/integrations/liger/models/jamba.py146 ) # make sure to reside in the same device
MEDIUMsrc/axolotl/prompt_strategies/context_qa.py107 "Therefore, I'm unable to answer your question based on the given context.",
MEDIUMsrc/axolotl/prompt_strategies/creative_acr.py33refusal: whether the Response refuses to answer the Question. Responses containing language similar to "As an AI languag
MEDIUMsrc/axolotl/prompt_strategies/creative_acr.py90refusal: whether the Response refuses to answer the Question. Responses containing language similar to "As an AI languag
Example Usage Blocks6 hits · 9 pts
SeverityFileLineSnippet
LOW…tl/integrations/nemo_gym/examples/nemo_gym_sudoku.yaml11# Usage:
LOW…ntegrations/nemo_gym/examples/nemo_gym_multi_turn.yaml15# Usage:
LOW…integrations/nemo_gym/examples/nemo_gym_multi_env.yaml12# Usage:
LOWsrc/axolotl/integrations/hatchery/__init__.py17# Usage:
LOW…axolotl/integrations/hatchery/examples/tinker_sft.yaml3# Usage:
LOWsrc/axolotl/loaders/tokenizer.py291 # Usage:
Verbosity Indicators5 hits · 8 pts
SeverityFileLineSnippet
LOWsrc/axolotl/cli/utils/lora_merge.py837 # Step 1: Fuse — MergeModulelist (stack experts) per source pattern
LOWsrc/axolotl/cli/utils/lora_merge.py883 # Step 2: Build the fused key name and merge LoRA
LOWsrc/axolotl/cli/utils/lora_merge.py948 # Step 3: Save in fused format (runtime format) so that the merged
LOWsrc/axolotl/cli/utils/lora_merge.py1129 # Step 1: Handle fused weight conversions (MoE experts) if applicable
LOWsrc/axolotl/cli/utils/lora_merge.py1149 # Step 2: Merge remaining (non-fused) tensors with LoRA
Fake / Example Data6 hits · 7 pts
SeverityFileLineSnippet
LOWtests/test_data.py38 "lorem ipsum dolor sit amet.",
LOWtests/test_data.py38 "lorem ipsum dolor sit amet.",
LOWtests/test_prompt_tokenizers.py75 "instruction": "hello cruel. lorem ipsum dolor sit amet.",
LOWtests/test_prompt_tokenizers.py75 "instruction": "hello cruel. lorem ipsum dolor sit amet.",
LOWsrc/axolotl/utils/data/sft.py255 f.write("lorem ipsum dolor sit amet\n")
LOWsrc/axolotl/utils/data/sft.py255 f.write("lorem ipsum dolor sit amet\n")
Synthetic Comment Markers1 hit · 5 pts
SeverityFileLineSnippet
HIGHREADME.md168- [API Reference](https://docs.axolotl.ai/docs/api/) - Auto-generated code documentation
Dead Code2 hits · 4 pts
SeverityFileLineSnippet
MEDIUMsrc/axolotl/models/mamba/modeling_mamba.py89
MEDIUMsrc/axolotl/models/mamba/modeling_mamba.py90