Go ahead and axolotl questions
This report summarizes scanner evidence for axolotl-ai-cloud/axolotl, a Python project with 12,554 GitHub stars. SynthScan examined 368,570 lines of code across 1843 source files, recording 6413 pattern matches distributed across 25 syntactic categories. The legacy adjusted pattern signal is 22.2 (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? 5097 finding(s).
Are imports declared and package/API names verified? 527 finding(s).
Are wrappers, guards, and repeated structures necessary? 5272 finding(s).
Which changes need focused human review? 6413 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 6413 distinct pattern matches across 25 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 | tests/test_bailing_hybrid_modeling.py | 93 | def test_layer_types_alternate_kda_and_mla(self, config, model): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 107 | def test_published_checkpoint_loads_unchanged(self, config, model, tmp_path): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 120 | def test_save_restores_published_layout(self, config, model, tmp_path): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 126 | def test_layer_types_ignore_a_stale_serialized_value(self, config): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 136 | def test_layer_types_survive_a_round_trip(self, config, tmp_path): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 143 | def test_multi_token_prediction_weights_are_not_reported_missing(self): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 172 | def test_matches_the_lower_bound_activation(self, attention): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 186 | def test_zero_pre_activation_is_half_the_bound(self, attention): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 200 | def test_stays_within_the_bound(self, attention): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 212 | def test_splits_on_every_restart(self): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 217 | def test_unpacked_batch_needs_no_varlen(self): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 223 | def test_chunk_starting_mid_document(self): | CODE |
| LOW⚡ | tests/test_bailing_hybrid_modeling.py | 229 | def test_batched_packing_splits_rows_and_restarts(self): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 243 | def test_both_attention_branches_receive_position_ids(self, model): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 275 | def test_packed_documents_match_separate_forwards(self, model): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 286 | def test_second_document_is_unaffected_by_the_first(self, model): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 311 | def test_packed_batch_splits_documents(self, model): | CODE |
| LOW | tests/test_bailing_hybrid_modeling.py | 336 | def test_segment_id_mask_still_splits_documents(self, model): | CODE |
| LOW | tests/test_processing_strategy_dispatch.py | 41 | def test_model_type_is_preferred_over_chat_template(monkeypatch): | CODE |
| LOW | tests/test_processing_strategy_dispatch.py | 69 | def test_absent_model_type_falls_back_to_chat_template(monkeypatch): | CODE |
| LOW | tests/test_processing_strategy_dispatch.py | 93 | def test_no_registry_key_collides_with_a_chat_template_name(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 97 | def test_native_descriptors_publish_immutable_specs_and_generic_peft_wrapper(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 124 | def test_native_lora_validation_rejects_unsafe_modes_and_unscoped_targets(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 154 | def test_diffusion_gemma_self_conditioning_module_is_trainable_and_saved(tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 195 | def test_tied_text_lora_shares_factors_and_rejects_unequal_loaded_duplicates(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 263 | def test_packed_diffusion_gemma_subclass_keeps_encoder_cache_in_one_forward(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 303 | def test_packed_diffusion_gemma_subclass_unrolls_inside_ddp_forward(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 336 | def test_packed_diffusion_gemma_bypasses_native_decoder_mask_creator(monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 345 | def mask_creator_must_not_run(*args, **kwargs): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 382 | def test_tied_lora_full_native_save_reload_merge_and_export(tmp_path, monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 442 | def test_tied_lora_rejects_partially_targeted_actual_alias_group(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 463 | def test_dream_factory_pins_code_and_weights(source, revision, monkeypatch, tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 520 | def test_dream_factory_from_config_removes_loader_only_kwargs(monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 554 | def test_dream_factory_builds_tiny_cached_audited_source(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 659 | def test_dream_factory_from_pretrained_resets_nonpersistent_rope_buffers(tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 715 | def test_native_diffusion_gemma_merge_dispatches_to_loaded_peft_path(monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 733 | def test_native_diffusion_gemma_rejects_explicit_efficient_merge(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 747 | def test_trainable_token_adapter_routes_to_efficient_merge(monkeypatch, tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 822 | def test_merged_export_preserves_valid_decision_manifest(tmp_path, monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 872 | def test_native_generation_dispatches_and_preserves_legacy_seam(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 919 | def test_native_nemotron_generation_dispatches_configurable_steps(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 931 | def generate_with_denoising_steps(self, prompt, **kwargs): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 964 | def test_native_generation_runs_tiny_diffusion_gemma_model_output(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 985 | def test_native_diffusion_gemma_infill_preserves_pinned_canvas_tokens(steps): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1029 | def test_native_generation_runs_tiny_dream_tensor_output(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1087 | def test_native_cli_uses_completion_by_default_and_routes_native_infill(monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1160 | def test_native_cli_commands_preserve_completion_and_reject_mask_infill(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1178 | def test_native_generation_callback_samples_sft_completion_suffix(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1225 | def test_nemotron_factory_accepts_flex_attention(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1270 | def test_nemotron_configurable_steps_preserve_source_default_and_allow_large_k(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1348 | def test_nemotron_tiny_source_mask_loss_and_generic_peft_contract(tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1512 | def test_nemotron_trainable_token_rows_survive_loaded_peft_merge(tmp_path): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1595 | def test_dream_generation_config_resolves_remote_utils_through_peft(monkeypatch): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1635 | def test_axolotl_subclass_built_first_keeps_grouped_mm_experts(): | CODE |
| LOW | tests/test_native_diffusion_model_support.py | 1656 | def test_diffusion_gemma_lora_targets_may_be_listed_per_path(): | CODE |
| LOW | tests/test_glm_dsa_cp_validation.py | 21 | def test_config_validation_recovers_inherited_defaults(): | CODE |
| LOW | tests/test_glm_dsa_cp_validation.py | 40 | def test_config_validation_recovers_known_none_defaults(): | CODE |
| LOW | tests/test_glm_dsa_cp_validation.py | 56 | def test_config_validation_retry_accepts_field_names(): | CODE |
| LOW | tests/test_glm_dsa_cp_validation.py | 77 | def test_dsa_cp_skips_flash_and_ring_requirement(self, monkeypatch): | CODE |
| LOW | tests/test_glm_dsa_cp_validation.py | 90 | def test_cp_without_dsa_uses_builtin_plugin(self, monkeypatch): | CODE |
| 4358 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | tests/test_ebft_kernels.py | 22 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 24 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 98 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 100 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 164 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 166 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 237 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tests/test_ebft_kernels.py | 239 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | tests/integrations/kernels/test_gemma4_moe.py | 175 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tests/integrations/kernels/test_gemma4_moe.py | 177 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tests/integrations/kernels/test_gemma4_moe.py | 446 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tests/integrations/kernels/test_gemma4_moe.py | 448 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 24 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 26 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 120 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 122 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 500 | # ============================================================================ | COMMENT |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 502 | # ============================================================================ | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 45 | # ─── Helpers ───────────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 121 | # ─── Model shape configs ──────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 139 | # ─── Forward tests ────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 186 | # ─── Backward dX tests ────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 220 | # ─── Backward LoRA gradient tests ─────────────────────────────────────────── | COMMENT |
| MEDIUM | …egrations/kernels/scattermoe_lora/test_lora_kernels.py | 291 | # ─── Full autograd tests ──────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …/scattermoe_lora/test_scattermoe_lora_int64_indices.py | 63 | # ─── Helpers ───────────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM | …/scattermoe_lora/test_scattermoe_lora_int64_indices.py | 79 | # ─── Parity tests at non-overflow shapes (bitwise identity) ────────────────── | COMMENT |
| MEDIUM | …/scattermoe_lora/test_scattermoe_lora_int64_indices.py | 300 | # ─── Overflow correctness at the bench shape ───────────────────────────────── | COMMENT |
| MEDIUM | …/scattermoe_lora/test_scattermoe_lora_int64_indices.py | 463 | # ─── Smaller-shape overflow (runs on L40S / 24 GiB GPUs) ───────────────────── | COMMENT |
| MEDIUM | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 190 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 192 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 295 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 297 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …s/kernels/scattermoe_lora/test_mxfp4_expert_weights.py | 187 | # ─── Strategy A — bitwise identity vs bf16 baseline ─────────────────────────── | COMMENT |
| MEDIUM | …s/kernels/scattermoe_lora/test_mxfp4_expert_weights.py | 292 | # ─── Strategy A — backward through fused dX/gather paths ────────────────────── | COMMENT |
| MEDIUM | …s/kernels/scattermoe_lora/test_mxfp4_expert_weights.py | 410 | # ─── Strategy B — fused MXFP4 Triton kernel ────────────────────────────────── | COMMENT |
| MEDIUM | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 16 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 18 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 85 | # ========================================================================= | COMMENT |
| MEDIUM | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 87 | # ========================================================================= | COMMENT |
| MEDIUM⚡ | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 211 | # ========================================================================= | COMMENT |
| MEDIUM⚡ | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 213 | # ========================================================================= | COMMENT |
| MEDIUM⚡ | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 366 | # ========================================================================= | COMMENT |
| MEDIUM⚡ | …ons/kernels/scattermoe_lora/test_autotune_telemetry.py | 368 | # ========================================================================= | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 156 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 158 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 201 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 204 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 254 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 256 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 272 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 274 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 72 | # =========================================================================== | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 74 | # =========================================================================== | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 159 | # =========================================================================== | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 161 | # =========================================================================== | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 351 | # =========================================================================== | COMMENT |
| MEDIUM | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 353 | # =========================================================================== | COMMENT |
| MEDIUM⚡ | …ions/kernels/scattermoe_lora/test_grouped_fp4_train.py | 218 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …ions/kernels/scattermoe_lora/test_grouped_fp4_train.py | 221 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | …ions/kernels/scattermoe_lora/test_grouped_fp4_train.py | 229 | # =========================================================================== | COMMENT |
| 323 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | tests/conftest.py | 225 | except LocalEntryNotFoundError: | CODE |
| HIGH | tests/core/test_async_grpo.py | 236 | except ValueError: | CODE |
| HIGH | tests/core/test_builders.py | 404 | except Exception: | CODE |
| HIGH | tests/utils/lora/test_merge_lora.py | 2338 | except RuntimeError: | CODE |
| HIGH | tests/integrations/test_swanlab.py | 1333 | except ValueError: | CODE |
| HIGH | tests/telemetry/test_errors.py | 406 | except ValueError: | CODE |
| HIGH | tests/e2e/utils.py | 55 | except Exception: # pylint: disable=broad-except | CODE |
| HIGH | tests/e2e/multigpu/solo/test_grpo.py | 140 | except FileNotFoundError: | CODE |
| HIGH | tests/e2e/integrations/test_hooks.py | 30 | except FileNotFoundError: | CODE |
| HIGH | tests/e2e/integrations/test_hooks.py | 186 | except FileNotFoundError: | CODE |
| HIGH | tests/monkeypatch/test_kernelize_fixes.py | 183 | except (TypeError, ValueError, AttributeError): | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn_robustness.py | 18 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_nf4_loading.py | 1936 | except BaseException: # pylint: disable=broad-except | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn.py | 17 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn.py | 198 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn.py | 232 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn.py | 211 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_fused_attn_defensive.py | 19 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_5_fused_attn.py | 17 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_5_fused_attn.py | 168 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_5_fused_attn.py | 329 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_5_fused_attn.py | 182 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_5_fused_attn.py | 340 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_qwen3_vl_fused_attn.py | 16 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_gemma4_fused_attn.py | 105 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_gemma4_fused_attn.py | 145 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_gemma4_fused_attn.py | 117 | except AttributeError: | CODE |
| HIGH | tests/monkeypatch/test_gemma4_fused_attn.py | 169 | except AttributeError: | CODE |
| HIGH | scripts/analyze_profile.py | 1388 | except Exception: | CODE |
| HIGH | src/axolotl/train.py | 517 | except (AttributeError, UnicodeDecodeError, OfflineModeIsEnabled): | CODE |
| HIGH | src/axolotl/train.py | 419 | except FileNotFoundError: | CODE |
| HIGH | src/axolotl/core/builders/causal.py | 564 | except (AttributeError, KeyError, TypeError): | CODE |
| HIGH | src/axolotl/core/trainers/base.py | 839 | except (ValueError, TypeError, FileNotFoundError): | CODE |
| HIGH | src/axolotl/core/trainers/ebft/strided.py | 51 | except ImportError: | CODE |
| HIGH | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 337 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 341 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 445 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/diffusion_lm/tokens.py | 23 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/diffusion_lm/tokens.py | 106 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/diffusion_lm/tokens.py | 100 | except Exception: | CODE |
| HIGH | src/axolotl/core/trainers/diffusion_lm/trainer.py | 144 | except Exception: | CODE |
| HIGH | src/axolotl/utils/tee.py | 162 | except Exception: | CODE |
| HIGH | src/axolotl/utils/tee.py | 41 | except Exception: | CODE |
| HIGH | src/axolotl/utils/tee.py | 70 | except Exception: | CODE |
| HIGH | src/axolotl/utils/quantization.py | 37 | except (ImportError, RuntimeError): | CODE |
| HIGH | src/axolotl/utils/quantization.py | 46 | except (ImportError, RuntimeError): | CODE |
| HIGH | src/axolotl/utils/quantization.py | 53 | except (ImportError, RuntimeError): | CODE |
| HIGH | src/axolotl/utils/quantization.py | 62 | except (ImportError, RuntimeError): | CODE |
| HIGH | src/axolotl/utils/quantization.py | 231 | except ImportError: | CODE |
| HIGH | src/axolotl/utils/callbacks/generation.py | 83 | except (ImportError, Exception): | CODE |
| HIGH | src/axolotl/utils/callbacks/swanlab.py | 245 | except ImportError: | CODE |
| HIGH | src/axolotl/utils/data/shared.py | 191 | except (FileNotFoundError, ConnectionError): | CODE |
| HIGH | src/axolotl/utils/data/lock.py | 70 | except FileNotFoundError: | CODE |
| HIGH | src/axolotl/cli/config.py | 81 | except ValueError: | CODE |
| HIGH | src/axolotl/cli/config.py | 85 | except ValueError: | CODE |
| HIGH | src/axolotl/cli/chat.py | 1235 | except ImportError: | CODE |
| HIGH | src/axolotl/cli/utils/lora_merge.py | 363 | except ImportError: | CODE |
| HIGH | src/axolotl/integrations/nemo_gym/server.py | 98 | except (requests.exceptions.ConnectionError, requests.exceptions.Timeout): | CODE |
| HIGH | src/axolotl/integrations/nemo_gym/plugin.py | 542 | except Exception: | CODE |
| HIGH | src/axolotl/integrations/nemo_gym/plugin.py | 244 | except (ValueError, KeyError, TypeError): | CODE |
| 42 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tests/test_native_diffusion_model_support.py | 3 | CODE | |
| LOW | tests/test_diffusion_model_support.py | 3 | CODE | |
| LOW | tests/test_fp32_norms.py | 3 | CODE | |
| LOW | tests/test_vectorized_scanner.py | 21 | CODE | |
| LOW | tests/test_train_seed_initialization.py | 3 | CODE | |
| LOW | tests/test_process_labels_fusion.py | 9 | CODE | |
| LOW | tests/test_mm_collator_warnings.py | 3 | CODE | |
| LOW | tests/test_no_legacy_attn_reads.py | 8 | CODE | |
| LOW | tests/native_source_fixtures.py | 3 | CODE | |
| LOW | tests/core/trainers/diffusion_lm/test_weighting.py | 3 | CODE | |
| LOW | tests/core/trainers/diffusion_lm/test_encoder_canvas.py | 3 | CODE | |
| LOW | tests/utils/test_cuda13.py | 3 | CODE | |
| LOW | tests/utils/test_quantization_init_guard.py | 81 | CODE | |
| LOW | tests/integrations/test_context_parallel_glm_dsa_cp.py | 19 | CODE | |
| LOW | tests/integrations/test_diffusion_config.py | 3 | CODE | |
| LOW | tests/integrations/kernels/test_gemma4_moe.py | 453 | CODE | |
| LOW | …rnels/scattermoe_lora/test_scattermoe_lora_m_bucket.py | 22 | CODE | |
| LOW | tests/integrations/kernels/scattermoe_lora/conftest.py | 19 | CODE | |
| LOW | …/scattermoe_lora/test_scattermoe_lora_int64_indices.py | 33 | CODE | |
| LOW | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 25 | CODE | |
| LOW | …ns/kernels/scattermoe_lora/test_nvfp4_fsdp_sharding.py | 27 | CODE | |
| LOW | …s/kernels/scattermoe_lora/test_nvfp4_layout_offline.py | 17 | CODE | |
| LOW | …ons/kernels/scattermoe_lora/test_grouped_fp4_guards.py | 14 | CODE | |
| LOW | …grations/kernels/scattermoe_lora/bench_int64_kernel.py | 26 | CODE | |
| LOW | …kernels/scattermoe_lora/test_nvfp4_weight_converter.py | 22 | CODE | |
| LOW | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 30 | CODE | |
| LOW | …ions/kernels/scattermoe_lora/test_grouped_fp4_train.py | 23 | CODE | |
| LOW | …termoe_lora/test_parallel_experts_large_batch_repro.py | 32 | CODE | |
| LOW | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 15 | CODE | |
| LOW | …/scattermoe_lora/test_gates_backward_autocast_dtype.py | 13 | CODE | |
| LOW | …tions/kernels/scattermoe_lora/test_grouped_fp4_perf.py | 23 | CODE | |
| LOW | …els/scattermoe_lora/test_grouped_fp4_weight_scale_2.py | 24 | CODE | |
| LOW | …s/integrations/kernels/dsv4/test_dsv4_attention_ops.py | 36 | CODE | |
| LOW | …ntegrations/diffusion_decision/test_decision_unroll.py | 3 | CODE | |
| LOW | tests/integrations/diffusion_decision/test_images.py | 3 | CODE | |
| LOW | …egrations/diffusion_decision/test_training_collator.py | 1 | CODE | |
| LOW | …tions/diffusion_decision/test_decision_grad_through.py | 3 | CODE | |
| LOW | …integrations/diffusion_decision/test_slot_collation.py | 3 | CODE | |
| LOW | …/integrations/diffusion_decision/test_slot_sampling.py | 3 | CODE | |
| LOW | …ations/diffusion_decision/test_nemotron_free_varlen.py | 3 | CODE | |
| LOW | tests/integrations/diffusion_decision/test_trainer.py | 3 | CODE | |
| LOW | tests/integrations/diffusion_decision/helpers.py | 1 | CODE | |
| LOW | …grations/diffusion_decision/test_fixed_slot_runtime.py | 3 | CODE | |
| LOW | …/integrations/diffusion_decision/test_dataset_slots.py | 3 | CODE | |
| LOW | …rations/diffusion_decision/test_jsonl_normalization.py | 1 | CODE | |
| LOW | tests/integrations/liger/test_plugin.py | 374 | CODE | |
| LOW | tests/e2e/multigpu/_cpu_ram_efficient_load_probe.py | 10 | CODE | |
| LOW | tests/e2e/multigpu/test_fsdp2_fp32_norms.py | 11 | CODE | |
| LOW | tests/e2e/multigpu/_fp32_norms_dtype_capture.py | 9 | CODE | |
| LOW | tests/e2e/multigpu/solo/_sonicmoe_nvfp4_fsdp2_worker.py | 3 | CODE | |
| LOW | tests/monkeypatch/test_float8_fsdp_sharding.py | 20 | CODE | |
| LOW | tests/monkeypatch/test_torchao_ddp.py | 3 | CODE | |
| LOW | tests/monkeypatch/test_fsdp2_state_dict_naming.py | 20 | CODE | |
| LOW | tests/prompt_strategies/test_chat_templates_mistral.py | 9 | CODE | |
| LOW | cicd/deps_hash.py | 11 | CODE | |
| LOW | cicd/select_e2e_tests.py | 26 | CODE | |
| LOW | scripts/diffusion_lm/build_public_decision_mix.py | 8 | CODE | |
| LOW | scripts/diffusion_lm/build_public_image_decision_mix.py | 8 | CODE | |
| LOW | src/axolotl/train.py | 3 | CODE | |
| LOW | src/axolotl/train.py | 47 | CODE | |
| 421 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | .runpod/src/utils.py | 40 | CODE | |
| LOW | .runpod/src/utils.py | 59 | CODE | |
| LOW | tests/test_bailing_hybrid_modeling.py | 63 | CODE | |
| LOW | tests/test_vectorized_scanner.py | 364 | CODE | |
| LOW | tests/test_vectorized_scanner.py | 548 | CODE | |
| LOW | tests/test_process_labels_fusion.py | 232 | CODE | |
| LOW | tests/test_weighted_loss.py | 15 | CODE | |
| LOW | tests/core/test_async_grpo.py | 300 | CODE | |
| LOW | tests/core/trainers/diffusion_lm/test_full_sequence.py | 163 | CODE | |
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 178 | CODE | |
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 215 | CODE | |
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 255 | CODE | |
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 301 | CODE | |
| LOW | tests/utils/lora/_lora_tying_distributed.py | 150 | CODE | |
| LOW | tests/utils/lora/test_merge_lora.py | 2568 | CODE | |
| LOW | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 331 | CODE | |
| LOW | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 55 | CODE | |
| LOW | tests/integrations/diffusion_decision/test_trainer.py | 1748 | CODE | |
| LOW | …ts/integrations/diffusion_decision/test_permutation.py | 45 | CODE | |
| LOW | …s/integrations/context_parallel/_fla_mamba_cp_probe.py | 13 | CODE | |
| LOW | …ations/expert_parallel/_ep_checkpoint_resume_worker.py | 184 | CODE | |
| LOW | tests/e2e/multigpu/_ringmaster_parity.py | 20 | CODE | |
| LOW | tests/e2e/multigpu/_fla_cp_parity.py | 80 | CODE | |
| LOW | …u/solo/_native_nvfp4_deepspeed_lora_merge_lifecycle.py | 237 | CODE | |
| LOW | …ultigpu/solo/_native_nvfp4_fsdp2_lora_parity_worker.py | 258 | CODE | |
| LOW | …/solo/_native_nvfp4_fsdp2_merge_aware_resume_worker.py | 315 | CODE | |
| LOW | …tigpu/solo/_torchao_lora_deepspeed_zero3_checkpoint.py | 62 | CODE | |
| LOW | …lo/test_native_nvfp4_deepspeed_lora_merge_lifecycle.py | 22 | CODE | |
| LOW | …s/e2e/multigpu/solo/_native_nvfp4_tp_hf_load_worker.py | 17 | CODE | |
| LOW | tests/e2e/multigpu/solo/test_grpo.py | 25 | CODE | |
| LOW | …2e/multigpu/solo/_torchao_lora_deepspeed_checkpoint.py | 52 | CODE | |
| LOW | tests/monkeypatch/test_lora_kernels_nemotron_h.py | 29 | CODE | |
| LOW | tests/monkeypatch/test_gemma4_unified_fused_attn.py | 25 | CODE | |
| LOW | tests/monkeypatch/test_nf4_loading.py | 426 | CODE | |
| LOW | tests/monkeypatch/test_nf4_loading.py | 1160 | CODE | |
| LOW | tests/monkeypatch/_fsdp2_full_checkpoint.py | 385 | CODE | |
| LOW | tests/monkeypatch/_fsdp2_full_checkpoint.py | 734 | CODE | |
| LOW | …monkeypatch/test_peft_checkpoint_wrapper_state_dict.py | 15 | CODE | |
| LOW | tests/monkeypatch/_fsdp2_bnb_checkpoint.py | 349 | CODE | |
| LOW | tests/monkeypatch/_fsdp2_bnb_checkpoint.py | 433 | CODE | |
| LOW | tests/monkeypatch/test_fla_mamba.py | 431 | CODE | |
| LOW | tests/monkeypatch/_lora_fp32_deepspeed.py | 30 | CODE | |
| LOW | tests/monkeypatch/test_parallelism_mesh_order.py | 74 | CODE | |
| LOW | tests/prompt_strategies/test_chat_templates_advanced.py | 728 | CODE | |
| LOW | tests/prompt_strategies/test_chat_templates_advanced.py | 1409 | CODE | |
| LOW | docs/scripts/generate_examples_docs.py | 94 | CODE | |
| LOW | docs/scripts/generate_examples_docs.py | 215 | CODE | |
| LOW | docs/scripts/generate_examples_docs.py | 286 | CODE | |
| LOW | docs/scripts/generate_examples_docs.py | 107 | CODE | |
| LOW | docs/scripts/generate_config_docs.py | 50 | CODE | |
| LOW | docs/scripts/generate_config_docs.py | 126 | CODE | |
| LOW | docs/scripts/generate_config_docs.py | 263 | CODE | |
| LOW | docs/scripts/generate_config_docs.py | 438 | CODE | |
| LOW | docs/scripts/generate_config_docs.py | 591 | CODE | |
| LOW | cicd/select_e2e_tests.py | 196 | CODE | |
| LOW | cicd/select_e2e_tests.py | 261 | CODE | |
| LOW | cicd/select_e2e_tests.py | 304 | CODE | |
| LOW | cicd/select_e2e_tests.py | 762 | CODE | |
| LOW | cicd/select_e2e_tests.py | 816 | CODE | |
| LOW | scripts/analyze_profile.py | 338 | CODE | |
| 259 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | tests/test_attn_implementation.py | 38 | def _xformers_available(): | CODE |
| LOW | tests/test_attn_implementation.py | 56 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | tests/core/test_builders.py | 404 | except Exception: | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 2277 | except Exception: # pragma: no cover | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 2287 | except Exception as ex: # pragma: no cover - torchao API / device gaps | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 2383 | except Exception: # pragma: no cover | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 2393 | except Exception as ex: # pragma: no cover | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 3329 | except Exception: # pragma: no cover | CODE |
| LOW | tests/utils/lora/test_merge_lora.py | 3342 | except Exception as ex: # pragma: no cover | CODE |
| MEDIUM | tests/integrations/kernels/test_gemma4_moe.py | 451 | def _can_import_sonicmoe(): | CODE |
| LOW⚡ | tests/integrations/kernels/test_gemma4_moe.py | 456 | except Exception: | CODE |
| LOW | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 65 | except Exception: | CODE |
| MEDIUM | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 56 | def gpu_name() -> str: | CODE |
| LOW | …grations/kernels/scattermoe_lora/bench_int64_kernel.py | 59 | except Exception: | CODE |
| MEDIUM | …grations/kernels/scattermoe_lora/bench_int64_kernel.py | 51 | def gpu_name() -> str: | CODE |
| LOW | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 61 | except Exception: | CODE |
| LOW | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 371 | except Exception: | CODE |
| LOW | …els/scattermoe_lora/test_grouped_fp4_weight_scale_2.py | 52 | except Exception: | CODE |
| LOW | …sts/integrations/expert_parallel/test_torch_backend.py | 101 | except Exception: # pylint: disable=broad-exception-caught | CODE |
| LOW | …sts/integrations/expert_parallel/test_torch_backend.py | 104 | except Exception: # pylint: disable=broad-exception-caught | CODE |
| LOW | …sts/integrations/expert_parallel/test_torch_backend.py | 1241 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | …sts/integrations/expert_parallel/test_torch_backend.py | 1312 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | tests/integrations/expert_parallel/test_nvfp4_wrap.py | 134 | except Exception: # pylint: disable=broad-except | CODE |
| MEDIUM | tests/integrations/expert_parallel/test_nvfp4_wrap.py | 65 | def _nvfp4_worker(rank, world_size, port, q, device="cpu"): | CODE |
| LOW | tests/e2e/utils.py | 55 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | tests/e2e/utils.py | 66 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | …sts/e2e/kernels/test_lora_mlp_geglu_moe_correctness.py | 22 | except Exception: # pragma: no cover | CODE |
| MEDIUM | tests/e2e/kernels/test_scattermoe_lora_kernels.py | 42 | def wrapper(*args, **kwargs): | CODE |
| LOW⚡ | tests/e2e/kernels/test_scattermoe_lora_kernels.py | 45 | except Exception as exc: # pylint: disable=broad-except | CODE |
| LOW | …e/multigpu/solo/_native_nvfp4_tp_lora_parity_worker.py | 130 | except Exception as exc: | CODE |
| LOW | …e/multigpu/solo/_native_nvfp4_tp_lora_parity_worker.py | 139 | except Exception as exc: | CODE |
| LOW | …e/multigpu/solo/_native_nvfp4_tp_lora_parity_worker.py | 310 | except Exception as error: | CODE |
| LOW | tests/monkeypatch/_fsdp2_full_checkpoint.py | 560 | except Exception as exc: | CODE |
| LOW | tests/monkeypatch/test_sdpa_varlen.py | 173 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | docs/scripts/generate_config_docs.py | 503 | except Exception: | CODE |
| LOW | docs/scripts/generate_config_docs.py | 607 | except Exception as e: | CODE |
| LOW | cicd/select_e2e_tests.py | 953 | except Exception: # pylint: disable=broad-except | CODE |
| LOW | examples/orpheus/README.md | 79 | except Exception as e: | CODE |
| LOW | examples/orpheus/README.md | 240 | except Exception as e: | CODE |
| LOW | scripts/analyze_profile.py | 1388 | except Exception: | CODE |
| LOW | scripts/check_installed_axolotl.py | 35 | except Exception as exc: # noqa: BLE001 | CODE |
| LOW | scripts/check_installed_axolotl.py | 75 | except Exception as exc: # noqa: BLE001 | CODE |
| LOW | src/axolotl/train.py | 91 | except Exception: # pylint: disable=broad-exception-caught | CODE |
| LOW | src/axolotl/kernels/quantize.py | 176 | except Exception: | CODE |
| LOW | src/axolotl/kernels/autotune_telemetry.py | 38 | except Exception: # pylint: disable=broad-exception-caught | CODE |
| LOW | src/axolotl/kernels/op_registry.py | 50 | except Exception as exc: # pylint: disable=broad-exception-caught | CODE |
| LOW | src/axolotl/core/trainers/base.py | 962 | except Exception as exc: # pylint: disable=broad-except | CODE |
| LOW | src/axolotl/core/trainers/base.py | 995 | except Exception as exc: # pylint: disable=broad-except | CODE |
| LOW | …c/axolotl/core/trainers/mixins/distributed_parallel.py | 215 | except Exception as exc: # pylint: disable=broad-except | CODE |
| LOW | src/axolotl/core/trainers/ebft/trainer.py | 501 | except Exception as e: | CODE |
| LOW⚡ | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 208 | except Exception: | CODE |
| LOW⚡ | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 337 | except Exception: | CODE |
| LOW⚡ | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 341 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 445 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 23 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 40 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 45 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 63 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 100 | except Exception: | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/tokens.py | 106 | except Exception: | CODE |
| 156 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH⚡ | examples/swanlab/dpo-swanlab-full-featured.yml | 13 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/dpo-swanlab-full-featured.yml | 145 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/dpo-swanlab-full-featured.yml | 147 | # swanlab_api_key: your-api-key | COMMENT |
| HIGH | examples/swanlab/lora-swanlab-profiling.yml | 12 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/lora-swanlab-profiling.yml | 92 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/lora-swanlab-profiling.yml | 94 | # swanlab_api_key: your-api-key | COMMENT |
| HIGH | examples/swanlab/README.md | 22 | export SWANLAB_API_KEY=your-api-key | CODE |
| HIGH | examples/swanlab/README.md | 43 | export SWANLAB_API_KEY=your-api-key | CODE |
| HIGH | examples/swanlab/README.md | 66 | export SWANLAB_API_KEY=your-api-key | CODE |
| HIGH | examples/swanlab/README.md | 149 | export SWANLAB_API_KEY=your-api-key | CODE |
| HIGH | examples/swanlab/README.md | 156 | swanlab_api_key: your-api-key | CODE |
| HIGH | examples/swanlab/custom_trainer_profiling.py | 247 | export SWANLAB_API_KEY=your-api-key | STRING |
| HIGH | examples/swanlab/dpo-swanlab-completions.yml | 12 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/dpo-swanlab-completions.yml | 104 | # export SWANLAB_API_KEY=your-api-key | COMMENT |
| HIGH⚡ | examples/swanlab/dpo-swanlab-completions.yml | 106 | # swanlab_api_key: your-api-key | COMMENT |
| HIGH | src/axolotl/integrations/swanlab/plugins.py | 88 | " export SWANLAB_API_KEY=your-api-key\n" | CODE |
| HIGH | src/axolotl/integrations/swanlab/plugins.py | 89 | " 2. Add 'swanlab_api_key: your-api-key' to config (less secure)\n" | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 46 | swanlab_api_key: YOUR_API_KEY # Optional: can also use env var SWANLAB_API_KEY | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 53 | export SWANLAB_API_KEY=your-api-key-here | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 841 | 1. Set environment variable: `export SWANLAB_API_KEY=your-api-key` | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 842 | 2. Add to config (less secure): `swanlab_api_key: your-api-key` | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 938 | export SWANLAB_API_KEY=your-api-key-here | CODE |
| HIGH | src/axolotl/integrations/swanlab/README.md | 949 | swanlab_api_key: your-api-key-here | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | .runpod/requirements.txt | 1 | # Required Python packages get listed here, one per line. | COMMENT |
| LOW | .runpod/src/config/config.yaml | 1 | # # This is the huggingface model that contains *.pt, *.safetensors, or *.bin files | COMMENT |
| LOW | .runpod/src/config/config.yaml | 21 | # tokenizer_use_fast: | COMMENT |
| LOW | .runpod/src/config/config.yaml | 41 | COMMENT | |
| LOW | .runpod/src/config/config.yaml | 61 | # float16: true | COMMENT |
| LOW | .runpod/src/config/config.yaml | 81 | # - path: repo | COMMENT |
| LOW | .runpod/src/config/config.yaml | 101 | COMMENT | |
| LOW | .runpod/src/config/config.yaml | 121 | # dataset_shard_num: | COMMENT |
| LOW | .runpod/src/config/config.yaml | 141 | # sample_packing_eff_est: | COMMENT |
| LOW | .runpod/src/config/config.yaml | 161 | # # - gate_proj | COMMENT |
| LOW | .runpod/src/config/config.yaml | 181 | # # Must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed | COMMENT |
| LOW | .runpod/src/config/config.yaml | 201 | # # Training hyperparameters | COMMENT |
| LOW | .runpod/src/config/config.yaml | 221 | # max_steps: | COMMENT |
| LOW | .runpod/src/config/config.yaml | 241 | # lr_scheduler: # 'one_cycle' | empty for cosine | COMMENT |
| LOW | .runpod/src/config/config.yaml | 261 | # # - adafactor | COMMENT |
| LOW | .runpod/src/config/config.yaml | 281 | COMMENT | |
| LOW | .runpod/src/config/config.yaml | 301 | # # xpos RoPE see https://github.com/kaiokendev/cutoff-len-is-context-len/blob/main/util/xpos_rope_llama_monkey_patch.py | COMMENT |
| LOW | .runpod/src/config/config.yaml | 321 | # # Add extra tokens. | COMMENT |
| LOW | tests/conftest.py | 361 | def download_argilla_ultrafeedback_binarized_preferences_cleaned_kto_dataset(): | COMMENT |
| LOW | tests/conftest.py | 381 | # @pytest.fixture(scope="session") | COMMENT |
| LOW | tests/integrations/test_swanlab.py | 1 | # Copyright 2024 Axolotl AI. All rights reserved. | COMMENT |
| LOW | tests/integrations/test_mixlora.py | 1 | # Copyright 2025 Axolotl AI. All rights reserved. | COMMENT |
| LOW | …s/kernels/scattermoe_lora/test_mxfp4_expert_weights.py | 61 | # Per-shape Strategy-B tolerances. Forward outputs accumulate K dot-products | COMMENT |
| LOW | …rnels/scattermoe_lora/test_grouped_fp4_dequant_path.py | 341 | dn.float().pow(2).mean().backward() | COMMENT |
| LOW | tests/e2e/multigpu/solo/__init__.py | 1 | # Tests under this directory should get run "solo" on their own as they | COMMENT |
| LOW | examples/gemma4/26b-a4b-moe-bnb-lora.yaml | 1 | # Gemma-4-26B-A4B MoE QLoRA on a bf16 checkpoint quantized to bnb-4bit at load — the throughput- | COMMENT |
| LOW | examples/ebft/qwen35-4b-ebft-structured-async.yaml | 1 | # EBFT Structured Mode: Qwen3.5-4B (hybrid linear attention) | COMMENT |
| LOW | examples/ebft/qwen35-4b-ebft-structured.yaml | 1 | # EBFT Structured Mode: Qwen3.5-4B (hybrid linear attention) | COMMENT |
| LOW | examples/ebft/llama-1b-ebft-opencode.yaml | 1 | # EBFT: Energy-Based Fine-Tuning with Llama-3.2-1B on OpenCodeInstruct | COMMENT |
| LOW | examples/cloud/nebius.yaml | 21 | # HF_TOKEN: mbsec-YOUR_SECRET_ID | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 1 | # SwanLab Full-Featured DPO Training Example | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 141 | # Workspace for team collaboration (shared experiments) | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 161 | swanlab_completion_log_interval: 100 # Log every 100 steps | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 181 | # export SWANLAB_LARK_WEBHOOK_URL=https://open.feishu.cn/... | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 201 | # - Identify bottlenecks in training loop | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 221 | COMMENT | |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 241 | # Expected Training Behavior | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 261 | # INFO: Only rank 0 will initialize SwanLab | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 281 | # ============================================================================ | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 301 | # If SwanLab initialization fails: | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-full-featured.yml | 321 | # 1. Verify use_swanlab is true | COMMENT |
| LOW | examples/swanlab/lora-swanlab-profiling.yml | 1 | # SwanLab LoRA Training Example with Performance Profiling | COMMENT |
| LOW | examples/swanlab/lora-swanlab-profiling.yml | 101 | # ============================================================================ | COMMENT |
| LOW | examples/swanlab/lora-swanlab-profiling.yml | 121 | COMMENT | |
| LOW | examples/swanlab/lora-swanlab-profiling.yml | 141 | # Then compare profiling metrics in SwanLab dashboard to see performance impact | COMMENT |
| LOW | examples/swanlab/lora-swanlab-profiling.yml | 161 | # | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-completions.yml | 1 | # SwanLab DPO Training Example with Completion Logging | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-completions.yml | 101 | COMMENT | |
| LOW | examples/swanlab/dpo-swanlab-completions.yml | 121 | # View the table in SwanLab dashboard under "rlhf_completions" | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-completions.yml | 141 | # Get real-time training notifications in your team chat | COMMENT |
| LOW | examples/swanlab/dpo-swanlab-completions.yml | 161 | COMMENT | |
| LOW | scripts/airgap/build_bundle.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | src/axolotl/train.py | 441 | else: | COMMENT |
| LOW | src/axolotl/kernels/lora.py | 1161 | # d_magnitude = sum_t(grad * combined) * (1 / weight_norm) | COMMENT |
| LOW | src/axolotl/core/training_args_base.py | 201 | chat_template: Optional[str] = field( | COMMENT |
| LOW | src/axolotl/core/builders/base.py | 1 | # Copyright 2024 Axolotl AI. All rights reserved. | COMMENT |
| LOW | src/axolotl/core/trainers/grpo/async_trainer.py | 1361 | if is_bg and self.use_vllm: | COMMENT |
| LOW | src/axolotl/core/trainers/grpo/async_trainer.py | 1381 | # ``if not async_prefetch: return``, so we MUST wire the stock | COMMENT |
| LOW | src/axolotl/core/trainers/grpo/fast_async_trainer.py | 1 | # Copyright 2020-2026 Axolotl AI. All rights reserved. | COMMENT |
| LOW | src/axolotl/core/trainers/prm/prm_trainer.py | 1 | # Copyright 2020-2026 The HuggingFace Team. All rights reserved. | COMMENT |
| 57 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | .axolotl-complete.bash | 18 | # Check if previous word is in our list | COMMENT |
| LOW | tests/e2e/kernels/test_scattermoe_lora_kernels.py | 809 | # Set weights to match test data | COMMENT |
| LOW | tests/e2e/patched/test_peft_embeddings.py | 63 | # Check if the embeddings are upcast correctly | COMMENT |
| LOW | tests/prompt_strategies/test_chat_templates_advanced.py | 644 | # Check if system message is not present in input_ids | COMMENT |
| LOW | tests/prompt_strategies/test_chat_templates_advanced.py | 715 | # Check if responses are labeled correctly based on role | COMMENT |
| LOW | docs/scripts/generate_examples_docs.py | 123 | # Check if target is outside examples_dir | COMMENT |
| LOW | docs/scripts/generate_examples_docs.py | 153 | # Check if this exact path is in allowlist | COMMENT |
| LOW | docs/scripts/generate_examples_docs.py | 243 | # Check if this is a parent page that has sub-pages | COMMENT |
| LOW | docs/scripts/generate_config_docs.py | 180 | # Check if this type itself is a Pydantic model | COMMENT |
| LOW | docs/scripts/generate_config_docs.py | 395 | # Check if this starts a new group (blank line before or significant gap) | COMMENT |
| LOW | docs/scripts/generate_config_docs.py | 674 | # Check if this field has nested models | COMMENT |
| LOW | examples/neutrino/8b-ternary-qat-sft-fsdp2.yml | 61 | # Set this to true if you have less than 80GB per GPU. | COMMENT |
| LOW | examples/neutrino/8b-ternary-qat-cpt-fsdp2.yml | 68 | # Set this to true if you have less than 80GB per GPU. | COMMENT |
| LOW | examples/glm_moe_dsa/glm-5.2-nvfp4-lora.yaml | 17 | # Set expert_parallel_size to the total GPU count (256 must divide it evenly), dp_shard_size 1. | COMMENT |
| LOW | scripts/cloud-entrypoint.sh | 29 | # Check if the word looks like the start of a key | COMMENT |
| LOW | scripts/cloud-entrypoint.sh | 72 | # Check if JUPYTER_PASSWORD is set and not empty | COMMENT |
| LOW | scripts/cloud-entrypoint.sh | 74 | # Set JUPYTER_TOKEN to the value of JUPYTER_PASSWORD | COMMENT |
| LOW | scripts/analyze_profile.py | 1110 | # Check if this small block is surrounded by inactive | COMMENT |
| LOW | scripts/cloud-entrypoint-term.sh | 25 | # Check if the word looks like the start of a key | COMMENT |
| LOW | scripts/cloud-entrypoint-term.sh | 63 | # Check if JUPYTER_PASSWORD is set and not empty | COMMENT |
| LOW | scripts/cloud-entrypoint-term.sh | 65 | # Set JUPYTER_TOKEN to the value of JUPYTER_PASSWORD | COMMENT |
| LOW | src/axolotl/train.py | 603 | # Check if the `token_ids_to_fix` kwarg exists in the fix_untrained_tokens args | COMMENT |
| LOW | src/axolotl/core/trainers/ebft/strided.py | 408 | # Check if we can share weights (PEFT model on single GPU) | COMMENT |
| LOW | src/axolotl/core/trainers/grpo/async_trainer.py | 1322 | # Check if model has FP8 weights | COMMENT |
| LOW | src/axolotl/utils/tee.py | 157 | # Close the file handle if open | COMMENT |
| LOW | src/axolotl/utils/freeze.py | 39 | # Check if any path component matches a vision/audio prefix | COMMENT |
| LOW | src/axolotl/utils/trainer.py | 226 | # Check if single example or batched by looking at the first element | COMMENT |
| LOW | src/axolotl/utils/trainer.py | 272 | # Check if single example or batch | COMMENT |
| LOW | src/axolotl/utils/callbacks/swanlab.py | 48 | # Check if SwanLab run is initialized | COMMENT |
| LOW | src/axolotl/utils/callbacks/swanlab.py | 163 | # Check if SwanLab is initialized | COMMENT |
| LOW | src/axolotl/utils/callbacks/lisa.py | 59 | # Check if it's time to switch active layers, including at step 0 | COMMENT |
| LOW⚡ | src/axolotl/utils/schemas/datasets.py | 227 | # Set chat_template to tokenizer_default if not set | COMMENT |
| LOW | src/axolotl/utils/samplers/multipack.py | 237 | # Assign bins to ranks - each rank gets every n-th bin | COMMENT |
| LOW | src/axolotl/utils/data/shared.py | 182 | # Check if it's a HuggingFace dataset | COMMENT |
| LOW | src/axolotl/utils/data/shared.py | 185 | # Check if it's a cloud storage path and get appropriate filesystem | COMMENT |
| LOW | src/axolotl/utils/data/streaming.py | 209 | # Set this to 1 so downstream data_loader doesn't try to increase the batch size | COMMENT |
| LOW | src/axolotl/utils/mistral/mistral_tokenizer.py | 68 | # Check if MistralRequestValidator has a _mode attribute. | COMMENT |
| LOW | src/axolotl/cli/config.py | 114 | # Check if the config is a valid HTTPS URL to a .yml or .yaml file | COMMENT |
| LOW | src/axolotl/cli/checks.py | 42 | # Verify if token is valid | COMMENT |
| LOW | src/axolotl/cli/utils/fetch.py | 35 | # Check if file exists and needs updating | COMMENT |
| LOW⚡ | src/axolotl/cli/utils/train.py | 27 | # Check if rdzv_endpoint is present | COMMENT |
| LOW⚡ | src/axolotl/cli/utils/train.py | 31 | # Check if rdzv_backend is already provided | COMMENT |
| LOW⚡ | src/axolotl/cli/utils/train.py | 36 | # Check if rdzv_id is already provided | COMMENT |
| LOW | src/axolotl/integrations/nemo_gym/server.py | 93 | # Check if already running | COMMENT |
| LOW | src/axolotl/integrations/cut_cross_entropy/__init__.py | 63 | # Check if cut_cross_entropy is installed | COMMENT |
| LOW | src/axolotl/integrations/cut_cross_entropy/__init__.py | 76 | # Check if Axolotl's cce fork is installed | COMMENT |
| LOW | src/axolotl/integrations/swanlab/plugins.py | 50 | # Check if SwanLab is enabled | COMMENT |
| LOW | src/axolotl/integrations/swanlab/plugins.py | 446 | # Check if Lark webhook URL is configured | COMMENT |
| LOW⚡ | src/axolotl/integrations/swanlab/plugins.py | 503 | # Check if completion logging is enabled | COMMENT |
| LOW⚡ | src/axolotl/integrations/swanlab/plugins.py | 509 | # Check if trainer is an RLHF trainer | COMMENT |
| LOW | src/axolotl/integrations/swanlab/profiling.py | 40 | # Check if SwanLab is enabled and initialized | COMMENT |
| LOW | src/axolotl/integrations/swanlab/profiling.py | 184 | # Check if should log based on config | COMMENT |
| LOW | src/axolotl/integrations/swanlab/profiling.py | 186 | # Check if SwanLab is enabled | COMMENT |
| LOW | src/axolotl/telemetry/callbacks.py | 131 | # Check if we should report metrics | COMMENT |
| LOW | src/axolotl/loaders/utils.py | 76 | # Check if image_size is not set and load image size from model config if available | COMMENT |
| LOW | src/axolotl/monkeypatch/deepspeed_utils.py | 33 | # Check if already patched | COMMENT |
| LOW | src/axolotl/monkeypatch/lora_kernels.py | 271 | # Check if already patched | STRING |
| LOW | src/axolotl/monkeypatch/attention/xformers.py | 45 | # Check if we have sliding window attention | COMMENT |
| LOW⚡ | …/axolotl/monkeypatch/transformers/trainer_loss_calc.py | 49 | # Check if already patched | COMMENT |
| LOW⚡ | …/axolotl/monkeypatch/transformers/trainer_loss_calc.py | 54 | # Check if the patterns exist | COMMENT |
| 7 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | …tigpu/solo/_torchao_lora_deepspeed_zero3_checkpoint.py | 0 | two-rank native torchao nvfp4 + ordinary lora deepspeed smoke. | STRING |
| HIGH | …2e/multigpu/solo/_torchao_lora_deepspeed_checkpoint.py | 0 | two-rank native torchao nvfp4 + ordinary lora deepspeed smoke. | STRING |
| HIGH | tests/e2e/multigpu/solo/_torchao_lora_deepspeed.py | 0 | two-rank native torchao nvfp4 + ordinary lora deepspeed smoke. | STRING |
| HIGH | src/axolotl/cli/preprocess.py | 0 | parses `axolotl` config, cli args, and calls `do_evaluate`. args: config: path to `axolotl` config yaml file. kwargs: ad | STRING |
| HIGH | src/axolotl/cli/merge_sharded_fsdp_weights.py | 0 | parses `axolotl` config, cli args, and calls `do_evaluate`. args: config: path to `axolotl` config yaml file. kwargs: ad | STRING |
| HIGH | src/axolotl/cli/train.py | 0 | parses `axolotl` config, cli args, and calls `do_evaluate`. args: config: path to `axolotl` config yaml file. kwargs: ad | STRING |
| HIGH | src/axolotl/cli/evaluate.py | 0 | parses `axolotl` config, cli args, and calls `do_evaluate`. args: config: path to `axolotl` config yaml file. kwargs: ad | STRING |
| HIGH | src/axolotl/integrations/liger/models/qwen3_5.py | 0 | args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked | STRING |
| HIGH | src/axolotl/integrations/liger/models/qwen3_5_moe.py | 0 | args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked | STRING |
| HIGH | src/axolotl/integrations/liger/models/base.py | 0 | args: labels (`torch.longtensor` of shape `(batch_size, sequence_length)`, *optional*): labels for computing the masked | STRING |
| HIGH | src/axolotl/monkeypatch/models/qwen3/fused_attn.py | 0 | unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight. | STRING |
| HIGH | …c/axolotl/monkeypatch/models/qwen3_5_moe/fused_attn.py | 0 | unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight. | STRING |
| HIGH | src/axolotl/monkeypatch/models/qwen3_5/fused_attn.py | 0 | unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight. | STRING |
| HIGH | src/axolotl/monkeypatch/models/qwen3_moe/fused_attn.py | 0 | unwrap peft ``modulestosavewrapper`` so the kernel reads the active adapter's weight. | STRING |
| HIGH | src/axolotl/prompt_strategies/bradley_terry/llama3.py | 0 | chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab | STRING |
| HIGH | src/axolotl/prompt_strategies/dpo/chatml.py | 0 | chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab | STRING |
| HIGH | src/axolotl/prompt_strategies/dpo/llama3.py | 0 | chatml transforms for datasets with system, input, chosen, rejected ex. https://huggingface.co/datasets/argilla/distilab | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | tests/conftest.py | 694 | # Create a temporary directory | COMMENT |
| MEDIUM⚡ | tests/test_packed_pretraining.py | 38 | # Create a list of 2000 random texts rather than just using it within the | COMMENT |
| MEDIUM⚡ | tests/test_packed_pretraining.py | 42 | # Create an IterableDataset | COMMENT |
| MEDIUM | tests/integrations/diffusion/test_plugin.py | 43 | # Create a minimal trainer instance just for testing methods | COMMENT |
| MEDIUM | tests/e2e/utils.py | 25 | # Create a temporary directory | COMMENT |
| MEDIUM | tests/e2e/kernels/test_lora.py | 44 | # Create main state with nested state | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 684 | # Create a new dataset with modified role names | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 758 | # Create a new dataset with the train and train_detail fields | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 1214 | # Create a situation where eot_tokens contains eos_token | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 1299 | # Create a non-existent token that definitely won't be in the template | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 1350 | # Create a custom chat template that uses our EOT token | COMMENT |
| MEDIUM | tests/prompt_strategies/test_chat_templates_advanced.py | 1425 | # Create a list to test different train_on_eot settings | COMMENT |
| MEDIUM | docs/scripts/generate_examples_docs.py | 83 | # Create a unique asset path based on source directory name | COMMENT |
| MEDIUM | scripts/cloud-entrypoint.sh | 21 | # Create the authorized_keys file if it doesn't exist | COMMENT |
| MEDIUM | scripts/cloud-entrypoint-term.sh | 17 | # Create the authorized_keys file if it doesn't exist | COMMENT |
| MEDIUM | src/axolotl/train.py | 205 | # Define the context managers to use | COMMENT |
| MEDIUM | src/axolotl/core/trainers/grpo/trainer.py | 251 | # Create the dataloader | COMMENT |
| MEDIUM | src/axolotl/core/trainers/prm/prm_trainer.py | 352 | # Create the label | STRING |
| MEDIUM | src/axolotl/utils/tokenization.py | 90 | # Create a delimiter based on text_only flag | COMMENT |
| MEDIUM | src/axolotl/utils/schemas/utils.py | 31 | data = data.copy() # Create a copy to avoid modifying the original | CODE |
| MEDIUM | src/axolotl/utils/samplers/multipack.py | 100 | # Create a new bin if needed and if we haven't reached the limit | COMMENT |
| MEDIUM | src/axolotl/cli/delinearize_llama4.py | 117 | # Create a new state dict for the converted model | COMMENT |
| MEDIUM | src/axolotl/integrations/mixlora/model.py | 124 | # Create a one-hot mask for selected experts and average over tokens | COMMENT |
| MEDIUM | src/axolotl/loaders/tokenizer.py | 52 | # Create the tokenizer directory in output_dir if it doesn't exist | COMMENT |
| MEDIUM | src/axolotl/monkeypatch/attention/flex_attn.py | 35 | # Create a new instance if one doesn't already exist | COMMENT |
| MEDIUM | …onkeypatch/models/mistral3/mistral_common_tokenizer.py | 22 | # Define the replacement | COMMENT |
| MEDIUM | src/axolotl/prompt_strategies/chat_template.py | 393 | # Create a new list to store tokens that should be kept | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tests/core/trainers/diffusion_lm/test_encoder_canvas.py | 503 | def _set_attention_backend(model, backend: str) -> None: | CODE |
| LOW | src/axolotl/core/builders/__init__.py | 6 | __all__ = ["HFCausalTrainerBuilder", "HFRLTrainerBuilder"] | CODE |
| LOW | src/axolotl/core/trainers/__init__.py | 10 | __all__ = [ | CODE |
| LOW | src/axolotl/core/trainers/grpo/async_trainer.py | 694 | def set_trainer(self, trainer) -> None: | CODE |
| LOW | src/axolotl/core/trainers/grpo/sampler.py | 166 | def set_epoch(self, epoch: int) -> None: | CODE |
| LOW | src/axolotl/core/trainers/prm/__init__.py | 5 | __all__ = ["PRMConfig"] | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/__init__.py | 11 | __all__ = ["get_diffusion_config"] | CODE |
| LOW | src/axolotl/core/trainers/diffusion_lm/callbacks.py | 15 | logger = logging.getLogger(__name__) | CODE |
| LOW | …xolotl/core/trainers/diffusion_lm/backends/__init__.py | 5 | __all__ = ["FullSequenceBackend"] | CODE |
| LOW | src/axolotl/utils/optimizers/adamc.py | 16 | __all__ = ["AdamC"] | CODE |
| LOW | src/axolotl/utils/optimizers/adopt.py | 36 | __all__ = ["ADOPT", "adopt"] | CODE |
| LOW | src/axolotl/utils/chat_templates/__init__.py | 14 | __all__ = [ | CODE |
| LOW | src/axolotl/utils/data/__init__.py | 14 | __all__ = [ | CODE |
| LOW | src/axolotl/utils/mistral/__init__.py | 6 | __all__ = ["HFMistralTokenizer", "Mistral3Processor"] | CODE |
| LOW | src/axolotl/utils/generation/__init__.py | 5 | __all__ = ["generate_samples", "format_generation_for_logging"] | CODE |
| LOW | src/axolotl/utils/collators/__init__.py | 11 | __all__ = [ | CODE |
| LOW | src/axolotl/cli/utils/__init__.py | 15 | __all__ = [ | CODE |
| LOW | src/axolotl/cli/utils/lora_merge.py | 1807 | def _update_config_vocab_size(output_path: Path, vocab_size: int) -> None: | CODE |
| LOW | src/axolotl/cli/cloud/modal_.py | 5 | __all__ = ["ModalCloud", "run_cmd"] | CODE |
| LOW | src/axolotl/cli/cloud/baseten/__init__.py | 5 | __all__ = ["BasetenCloud"] | CODE |
| LOW | src/axolotl/integrations/nemo_gym/__init__.py | 20 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/kernels/__init__.py | 4 | __all__ = [ | CODE |
| LOW | …/integrations/kernels/libs/scattermoe_lora/__init__.py | 17 | __all__ = [ | CODE |
| LOW⚡ | …tegrations/kernels/libs/scattermoe_lora/chunked_bnb.py | 26 | def set_chunk_size_override(n) -> None: | CODE |
| LOW⚡ | …tegrations/kernels/libs/scattermoe_lora/chunked_bnb.py | 30 | def set_layer_gc_active(flag) -> None: | CODE |
| LOW⚡ | …tegrations/kernels/libs/scattermoe_lora/chunked_bnb.py | 34 | def set_bnb_fast(flag) -> None: | CODE |
| LOW | …grations/kernels/libs/scattermoe_lora/grouped_train.py | 36 | def set_grouped_backend_override(backend) -> None: | CODE |
| LOW | …tions/kernels/libs/scattermoe_lora/kernels/__init__.py | 12 | __all__ = ["ops", "lora_ops"] | CODE |
| LOW | src/axolotl/integrations/kernels/libs/dsv4/__init__.py | 11 | __all__ = [ | CODE |
| LOW | …/axolotl/integrations/kernels/libs/glm_dsa/__init__.py | 15 | __all__ = [ | CODE |
| LOW | …olotl/integrations/kernels/libs/sonicmoe/nvfp4_lora.py | 45 | def set_merge_aware_enabled(enabled: bool) -> None: | CODE |
| LOW | …axolotl/integrations/kernels/libs/sonicmoe/__init__.py | 26 | __all__ = [ | CODE |
| LOW | …/axolotl/integrations/kernels/libs/sonicmoe/experts.py | 451 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/kd/kernels/__init__.py | 7 | __all__ = ["LigerFusedLinearKLTopKLogprobLoss"] | CODE |
| LOW | src/axolotl/integrations/diffusion_decision/__init__.py | 10 | __all__ = [ | CODE |
| LOW | …xolotl/integrations/diffusion_decision/slot_runtime.py | 29 | def _set_value(value: Any, name: str, item: Any) -> None: | CODE |
| LOW | src/axolotl/integrations/diffusion_decision/template.py | 53 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/diffusion_decision/trainer.py | 1429 | def set_epoch(self, epoch: int) -> None: | CODE |
| LOW | src/axolotl/integrations/diffusion_decision/trainer.py | 1481 | def set_epoch(self, epoch: int) -> None: | CODE |
| LOW | …l/integrations/diffusion_decision/adapters/__init__.py | 44 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/diffusion/generation.py | 11 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/diffusion/__init__.py | 10 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/diffusion/utils.py | 9 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/diffusion/callbacks.py | 6 | __all__ = ["DiffusionGenerationCallback", "generate_samples"] | CODE |
| LOW | src/axolotl/integrations/diffusion/args.py | 17 | __all__ = ["DiffusionArgs", "DiffusionConfig"] | CODE |
| LOW | src/axolotl/integrations/diffusion/trainer.py | 8 | __all__ = ["AxolotlDiffusionTrainer", "DiffusionTrainer"] | CODE |
| LOW | src/axolotl/integrations/mixlora/__init__.py | 27 | __all__ = ["MixLoraPlugin"] | CODE |
| LOW | src/axolotl/integrations/densemixer/__init__.py | 5 | __all__ = ["DenseMixerPlugin"] | CODE |
| LOW | src/axolotl/integrations/hatchery/__init__.py | 15 | __all__ = ["HatcheryArgs", "HatcheryConfig", "HatcheryPlugin"] | CODE |
| LOW | src/axolotl/integrations/swanlab/__init__.py | 6 | __all__ = ["SwanLabConfig", "SwanLabPlugin"] | CODE |
| LOW | src/axolotl/integrations/liger/__init__.py | 25 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/context_parallel/__init__.py | 17 | __all__ = [ | CODE |
| LOW | src/axolotl/integrations/llm_compressor/__init__.py | 5 | __all__ = ["LLMCompressorPlugin"] | CODE |
| LOW | src/axolotl/integrations/mora/__init__.py | 6 | __all__ = ["MoraArgs", "MoraConfig", "MoraPlugin", "MoraType"] | CODE |
| LOW | src/axolotl/integrations/expert_parallel/experts_fn.py | 37 | def set_token_capacity(cap: int | None) -> None: | CODE |
| LOW | src/axolotl/integrations/expert_parallel/experts_fn.py | 46 | def set_dispatch_chunks(chunks: int) -> None: | CODE |
| LOW | src/axolotl/integrations/expert_parallel/experts_fn.py | 91 | def set_valid_token_mask(mask: torch.Tensor | None) -> None: | CODE |
| LOW | src/axolotl/integrations/expert_parallel/experts_fn.py | 108 | def set_backend(backend: str | None) -> None: | CODE |
| LOW | src/axolotl/integrations/expert_parallel/experts_fn.py | 119 | def set_local_implementation(name: str) -> None: | CODE |
| LOW | src/axolotl/integrations/expert_parallel/__init__.py | 25 | __all__ = [ | CODE |
| 11 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | src/axolotl/core/trainers/grpo/__init__.py | 302 | Returns the reward function from the given fully qualified name, or the path to the reward function model. | STRING |
| HIGH | src/axolotl/utils/quantization.py | 71 | This function is used to build a post-training quantization config. Args: weight_dtype: The dtype to u | STRING |
| HIGH | src/axolotl/utils/schemas/utils.py | 9 | Handle backwards compatibility between legacy message field mapping and new property mapping system. Previousl | STRING |
| HIGH | src/axolotl/utils/chat_templates/base.py | 31 | Finds the correct chat_template based on the user's choice, jinja_template, and tokenizer. Args: user_ | STRING |
| HIGH | src/axolotl/utils/data/rl.py | 130 | Filter out samples that exceed maximum sequence length. Args: sample: Dataset sample to check. rl: | STRING |
| HIGH | src/axolotl/utils/data/rl.py | 187 | Check sequence length and raise ValueError if exceeded. Used as a filter function for ``excess_length_strategy: rai | STRING |
| HIGH | src/axolotl/cli/config.py | 97 | First, determines if the passed config is a valid HTTPS URL. Then, attempts to query for it and parse its conte | STRING |
| HIGH | src/axolotl/cli/config.py | 163 | Helper method for choosing a `axolotl` config YAML file (considering only files ending with `.yml` or `.yaml`). | STRING |
| HIGH | src/axolotl/cli/agent_docs/__init__.py | 84 | Return the content of an agent doc by topic name. Args: topic: One of the keys in TOPICS, or "overview" (de | STRING |
| HIGH | src/axolotl/integrations/base.py | 375 | Loads a plugin based on the given plugin name. The plugin name should be in the format "module_name.class_name". Th | STRING |
| HIGH | src/axolotl/integrations/expert_parallel/shard.py | 164 | Slice expert weights along dim 0 per the EP rank. Args: model: A built (but not yet FSDP-wrapped) HuggingFa | STRING |
| HIGH | src/axolotl/loaders/utils.py | 151 | Loads and configures a model configuration from HuggingFace or local sources. This function determines the appropri | STRING |
| HIGH | src/axolotl/monkeypatch/lora_kernels.py | 156 | Get the appropriate attention class by inspecting the model config. Uses dynamic import to support any model ar | STRING |
| HIGH | src/axolotl/monkeypatch/lora_kernels.py | 475 | Applies optimized Triton kernel patches to a PEFT model. Patches a PEFT model with optimized implementations f | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| CRITICAL⚡ | src/axolotl/integrations/kernels/autotune_callback.py | 49 | props = triton.runtime.driver.active.utils.get_device_properties(idx) | CODE |
| CRITICAL⚡ | …tions/kernels/libs/scattermoe_lora/kernels/lora_ops.py | 192 | props = triton.runtime.driver.active.utils.get_device_properties( | CODE |
| CRITICAL | src/axolotl/integrations/kernels/libs/dsv4/attention.py | 50 | return triton.runtime.driver.active.utils.get_device_properties(dev)[ | CODE |
| CRITICAL | …axolotl/integrations/kernels/libs/glm_dsa/_autotune.py | 21 | props = triton.runtime.driver.active.utils.get_device_properties(dev) | CODE |
| CRITICAL | src/axolotl/monkeypatch/attention/xformers.py | 72 | xformers.ops.fmha.attn_bias.BlockDiagonalCausalMask.from_seqlens( | CODE |
| CRITICAL | src/axolotl/monkeypatch/attention/xformers.py | 94 | attn_bias = xformers.ops.fmha.attn_bias.BlockDiagonalCausalMask.from_seqlens( | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | …ions/kernels/scattermoe_lora/test_mxfp4_integration.py | 225 | CODE | |
| LOW | tests/integrations/diffusion_decision/helpers.py | 126 | CODE | |
| LOW | tests/integrations/diffusion_decision/helpers.py | 155 | CODE | |
| LOW | tests/integrations/diffusion_decision/helpers.py | 257 | CODE | |
| LOW | tests/e2e/kernels/test_scattermoe_lora_kernels.py | 182 | CODE | |
| LOW | tests/e2e/kernels/test_scattermoe_lora_kernels.py | 1573 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1729 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 78 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 772 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 819 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 895 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1179 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1232 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1265 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1351 | CODE | |
| LOW | src/axolotl/processing_strategies.py | 1404 | CODE | |
| LOW | src/axolotl/kernels/lora.py | 854 | CODE | |
| LOW | src/axolotl/kernels/lora.py | 915 | CODE | |
| LOW | src/axolotl/kernels/lora.py | 1914 | CODE | |
| LOW | src/axolotl/core/trainers/grpo/async_trainer.py | 2914 | CODE | |
| LOW | src/axolotl/core/trainers/grpo/trainer.py | 92 | CODE | |
| LOW | src/axolotl/core/trainers/prm/prm_trainer.py | 154 | CODE | |
| LOW | src/axolotl/core/trainers/diffusion_lm/generation.py | 18 | CODE | |
| LOW | src/axolotl/core/trainers/diffusion_lm/generation.py | 347 | CODE | |
| LOW | …/core/trainers/diffusion_lm/backends/encoder_canvas.py | 282 | CODE | |
| LOW | src/axolotl/utils/optimizers/qgalore.py | 61 | CODE | |
| LOW | src/axolotl/utils/optimizers/sinkgd_triton.py | 381 | CODE | |
| LOW | src/axolotl/utils/optimizers/adopt.py | 40 | CODE | |
| LOW | src/axolotl/utils/optimizers/sinkgd.py | 249 | CODE | |
| LOW | src/axolotl/utils/samplers/multipack.py | 257 | CODE | |
| LOW | src/axolotl/utils/mistral/mistral_tokenizer.py | 190 | CODE | |
| LOW | src/axolotl/cli/chat.py | 851 | CODE | |
| LOW | src/axolotl/cli/utils/lora_merge.py | 1966 | CODE | |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2318 | CODE | |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2563 | CODE | |
| LOW | …s/kernels/libs/scattermoe_lora/parallel_linear_lora.py | 569 | CODE | |
| LOW | …l/integrations/kernels/libs/scattermoe_lora/experts.py | 567 | CODE | |
| LOW | …l/integrations/kernels/libs/scattermoe_lora/experts.py | 809 | CODE | |
| LOW | …els/libs/scattermoe_lora/cutlass_fp4/grouped_kernel.py | 1426 | CODE | |
| LOW | src/axolotl/integrations/lm_eval/cli.py | 31 | CODE | |
| LOW | src/axolotl/integrations/kd/chat_template.py | 34 | CODE | |
| LOW | src/axolotl/integrations/kd/kernels/liger.py | 120 | CODE | |
| LOW | src/axolotl/integrations/kd/kernels/liger.py | 180 | CODE | |
| LOW | src/axolotl/integrations/kd/kernels/liger.py | 429 | CODE | |
| LOW | …lotl/integrations/diffusion_decision/prepared_cache.py | 318 | CODE | |
| LOW | src/axolotl/integrations/diffusion_decision/slots.py | 39 | CODE | |
| LOW | …olotl/integrations/diffusion_decision/preprocessing.py | 35 | CODE | |
| LOW | …l/integrations/diffusion_decision/training_collator.py | 55 | CODE | |
| LOW | src/axolotl/integrations/mixlora/trainer.py | 60 | CODE | |
| LOW | src/axolotl/integrations/liger/models/jamba.py | 16 | CODE | |
| LOW | src/axolotl/integrations/liger/models/qwen3_5.py | 15 | CODE | |
| LOW | src/axolotl/integrations/liger/models/deepseekv2.py | 15 | CODE | |
| LOW | src/axolotl/integrations/liger/models/qwen3_5_moe.py | 14 | CODE | |
| LOW | src/axolotl/monkeypatch/stablelm_attn_hijack_flash.py | 247 | CODE | |
| LOW | src/axolotl/monkeypatch/attention/xformers.py | 19 | CODE | |
| LOW | src/axolotl/monkeypatch/models/qwen3_next/modeling.py | 384 | CODE | |
| LOW | src/axolotl/monkeypatch/models/voxtral/modeling.py | 18 | CODE | |
| LOW | src/axolotl/monkeypatch/models/falcon_h1/modeling.py | 301 | CODE | |
| LOW | …support/bailing_hybrid/configuration_bailing_moe_v3.py | 23 | CODE | |
| LOW | …odel_support/bailing_hybrid/modeling_bailing_moe_v3.py | 574 | CODE | |
| 5 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | pyproject.toml | 61 | # lm eval harness | COMMENT |
| MEDIUM⚡ | …ts/integrations/kernels/scattermoe_lora/bench_mxfp4.py | 296 | # Timing harness | COMMENT |
| MEDIUM⚡ | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 181 | # 2. Memory bound across skew (THE skew guard) — a robust op's peak must NOT scale with routing | COMMENT |
| MEDIUM⚡ | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 183 | # BALANCED peak (scale-robust, no magic absolute). A capacity-padded impl (cap=max) blows up at | COMMENT |
| MEDIUM⚡ | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 212 | # robust: skew peak within 1.5x of balanced. capacity-padded all_to_one is many x -> fails. | COMMENT |
| MEDIUM⚡ | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 229 | # cap this process to a fraction that comfortably fits the robust path but not an E*cap blowup | COMMENT |
| MEDIUM⚡ | …tions/kernels/scattermoe_lora/test_grouped_fp4_skew.py | 234 | ) # robust needs ~12x routed + model; blowup needs ~Ex more | CODE |
| MEDIUM | tests/monkeypatch/test_nf4_loading.py | 372 | # ModelLoader.load() does this before prepare; this harness builds the | COMMENT |
| MEDIUM | tests/monkeypatch/test_nf4_loading.py | 1092 | # ModelLoader.load() does this before prepare; this harness builds the | COMMENT |
| MEDIUM | examples/archived/xgen-7b/xgen-7b-8k-qlora.yml | 31 | # "We find hyperparameters to be largely robust across datasets" | COMMENT |
| MEDIUM | examples/archived/falcon/config-7b-qlora.yml | 33 | # "We find hyperparameters to be largely robust across datasets" | COMMENT |
| MEDIUM | .github/workflows/tests-nightly.yml | 5 | # to this file or the cicd/ harness. | COMMENT |
| LOW | src/axolotl/cli/utils/sweeps.py | 61 | # If no paired values, just use regular combinations | COMMENT |
| MEDIUM | …ions/kernels/libs/scattermoe_lora/nvfp4_moe_loading.py | 134 | # robust NVFP4 signal (uint8 qdata) — distinguishing NVFP4 from FP8 (which also carries a | COMMENT |
| MEDIUM | …tions/kernels/libs/scattermoe_lora/kernels/lora_ops.py | 2562 | (128, 32, 64, 8, 3), # robust fallback within the stock search space | CODE |
| MEDIUM | …/axolotl/integrations/kernels/libs/glm_dsa/dispatch.py | 1 | """Length-aware dispatch for GLM-5.2 DSA attention — robust across short and long sequences. | STRING |
| LOW | src/axolotl/integrations/kd/collator.py | 211 | # just pass straight to parent. | COMMENT |
| LOW | src/axolotl/prompt_strategies/chat_template.py | 677 | # grid info, etc.); just set the fields we computed locally. | COMMENT |
| LOW | src/axolotl/prompt_strategies/ebft/ebft_strided_chat.py | 20 | # For preview: just return the first user message | COMMENT |
| MEDIUM | src/axolotl/prompt_strategies/ebft/ebft_reasoning.py | 280 | # directly in the input_ids (robust to tokenization alignment). | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | tests/integrations/nebius/test_cloud.py | 567 | SimpleNamespace(returncode=0, stdout='{"metadata": {}, "status": null}'), | CODE |
| HIGH | src/axolotl/core/trainers/grpo/async_trainer.py | 877 | VLLMGeneration._init_vllm = _orig_init_vllm # type: ignore[possibly-undefined] | CODE |
| HIGH | src/axolotl/utils/schemas/config.py | 922 | "flash_attn dependency. Left unset (null) it auto-enables when supported (torch >= " | CODE |
| HIGH | src/axolotl/cli/config_options.py | 2681 | "With sample packing + attn_implementation=sdpa, route packed rows through torch.nn.attention.varlen.varlen_attn | CODE |
| HIGH | src/axolotl/integrations/nemo_gym/server.py | 69 | ["bash", "-c", "source .venv/bin/activate && uv sync"], | CODE |
| HIGH | src/axolotl/integrations/kd/kernels/liger.py | 36 | 0.0 for Forward KL (P_teacher || P_student). | STRING |
| HIGH | src/axolotl/integrations/kd/kernels/liger.py | 37 | 1.0 for Reverse KL (P_student || P_teacher). | STRING |
| HIGH | src/axolotl/integrations/cut_cross_entropy/__init__.py | 40 | '`pip uninstall -y cut-cross-entropy && pip install "cut-cross-entropy[transformers] @ git+https://github.com/axolot | CODE |
| HIGH | src/axolotl/integrations/nebius/runner.py | 128 | 'source /workspace/axolotl/scripts/cuda13_env.sh && exec "$@"', | CODE |
| HIGH | src/axolotl/integrations/modal/cloud.py | 181 | f"RUN cd /workspace/axolotl && git fetch && git checkout {self.config.branch} && git pull", | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | tests/test_processing_strategies.py | 61 | # Generic fake tokenizer/processor scaffold | COMMENT |
| MEDIUM | examples/ebft/llama-1b-ebft-strided.yaml | 12 | context_length: 8 # context window per block | CODE |
| MEDIUM | examples/ebft/llama-1b-ebft-strided-structured.yaml | 12 | context_length: 8 # context window per block | CODE |
| MEDIUM | examples/ebft/README.md | 100 | context_length: 8 # context window per block (paper default: 8) | CODE |
| MEDIUM | examples/muse-glimmer/README.md | 134 | - [Meta AI Research announcement](https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model) | CODE |
| MEDIUM | examples/mistral-medium-3_5/README.md | 3 | [Mistral Medium 3.5](https://huggingface.co/mistralai/Mistral-Medium-3.5-128B) is a 128B parameter dense multimodal mode | CODE |
| MEDIUM | src/axolotl/core/trainers/ebft/strided.py | 153 | # Rule 1: Generated token attends to its context window in the prompt | COMMENT |
| MEDIUM | src/axolotl/core/trainers/ebft/strided.py | 917 | # Last token of the context window predicts the first rollout token | COMMENT |
| MEDIUM | src/axolotl/integrations/nemo_gym/README.md | 79 | # Agent server (orchestrates multi-turn via /run) | COMMENT |
| MEDIUM | …ntegrations/nemo_gym/examples/nemo_gym_multi_turn.yaml | 4 | # using NeMo Gym's agentic environments. The model generates responses, | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 139 | # Step 5 - shouldn't check (not divisible by 10) | COMMENT |
| LOW | tests/utils/callbacks/test_dynamic_checkpoint.py | 145 | # Step 10 - should check | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2475 | # Step 1: Fuse — MergeModulelist (stack experts) per source pattern | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2521 | # Step 2: Build the fused key name and merge LoRA | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2547 | # Step 3: Save in fused format (runtime format) so that the merged | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2951 | # Step 0: dequantize quantized weights so the LoRA delta folds into the TRUE weight (a raw read | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2968 | # Step 1: Handle fused weight conversions (MoE experts) if applicable | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2996 | # Step 2: Merge remaining (non-fused) tensors with LoRA | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 363 | # Step 1: Compute shared base FFN outputs (optimization from paper) | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 367 | # Step 2: Route tokens to experts | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 371 | # Step 3: Compute base FFN output (shared SwiGLU + down_proj) | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 375 | # Step 4: Compute expert LoRA deltas and aggregate | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | src/axolotl/cli/utils/lora_merge.py | 2475 | # Step 1: Fuse — MergeModulelist (stack experts) per source pattern | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2521 | # Step 2: Build the fused key name and merge LoRA | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2547 | # Step 3: Save in fused format (runtime format) so that the merged | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2951 | # Step 0: dequantize quantized weights so the LoRA delta folds into the TRUE weight (a raw read | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2968 | # Step 1: Handle fused weight conversions (MoE experts) if applicable | COMMENT |
| LOW | src/axolotl/cli/utils/lora_merge.py | 2996 | # Step 2: Merge remaining (non-fused) tensors with LoRA | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 363 | # Step 1: Compute shared base FFN outputs (optimization from paper) | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 367 | # Step 2: Route tokens to experts | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 371 | # Step 3: Compute base FFN output (shared SwiGLU + down_proj) | COMMENT |
| LOW⚡ | src/axolotl/integrations/mixlora/model.py | 375 | # Step 4: Compute expert LoRA deltas and aggregate | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | .runpod/src/config/config.yaml | 65 | # # HuggingFace dataset repo | s3://,gs:// path | "json" for local dataset, make sure to fill data_files | COMMENT |
| LOW | src/axolotl/utils/callbacks/profiler.py | 101 | # make sure to record if we happen to have more steps than steps to profile | COMMENT |
| LOW | src/axolotl/cli/utils/train.py | 187 | # make sure to flush stdout and stderr before replacing the process | COMMENT |
| LOW | src/axolotl/cli/utils/train.py | 213 | # make sure to flush stdout and stderr before replacing the process | COMMENT |
| LOW | src/axolotl/integrations/kd/collator_online_teacher.py | 531 | # TODO make sure to figure out which input in sub_batch_features to update the batch in the original | COMMENT |
| LOW | src/axolotl/integrations/liger/models/jamba.py | 124 | ) # make sure to reside in the same device | CODE |
| MEDIUM | src/axolotl/prompt_strategies/context_qa.py | 107 | "Therefore, I'm unable to answer your question based on the given context.", | CODE |
| MEDIUM | src/axolotl/prompt_strategies/creative_acr.py | 33 | refusal: whether the Response refuses to answer the Question. Responses containing language similar to "As an AI languag | CODE |
| MEDIUM | src/axolotl/prompt_strategies/creative_acr.py | 90 | refusal: whether the Response refuses to answer the Question. Responses containing language similar to "As an AI languag | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | scripts/airgap/build_bundle.sh | 5 | # Usage: | COMMENT |
| LOW | …tl/integrations/nemo_gym/examples/nemo_gym_sudoku.yaml | 11 | # Usage: | COMMENT |
| LOW | …ntegrations/nemo_gym/examples/nemo_gym_multi_turn.yaml | 15 | # Usage: | COMMENT |
| LOW | …integrations/nemo_gym/examples/nemo_gym_multi_env.yaml | 12 | # Usage: | COMMENT |
| LOW | src/axolotl/integrations/hatchery/__init__.py | 17 | # Usage: | COMMENT |
| LOW | …axolotl/integrations/hatchery/examples/tinker_sft.yaml | 3 | # Usage: | COMMENT |
| LOW | src/axolotl/loaders/tokenizer.py | 289 | # Usage: | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | tests/test_data.py | 44 | "lorem ipsum dolor sit amet.", | CODE |
| LOW | tests/test_data.py | 44 | "lorem ipsum dolor sit amet.", | CODE |
| LOW | tests/test_prompt_tokenizers.py | 75 | "instruction": "hello cruel. lorem ipsum dolor sit amet.", | CODE |
| LOW | tests/test_prompt_tokenizers.py | 75 | "instruction": "hello cruel. lorem ipsum dolor sit amet.", | CODE |
| LOW⚡ | src/axolotl/utils/data/sft.py | 263 | f.write("lorem ipsum dolor sit amet\n") | CODE |
| LOW⚡ | src/axolotl/utils/data/sft.py | 263 | f.write("lorem ipsum dolor sit amet\n") | CODE |
| LOW | src/axolotl/integrations/nemo_gym/README.md | 74 | api_key: dummy_key | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | README.md | 189 | - [API Reference](https://docs.axolotl.ai/docs/api/) - Auto-generated code documentation | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | tests/scripts/test_select_e2e_tests.py | 234 | {"tests/e2e/utils.py": "def helper():\n return 1\n"}, | STRING |