Enjoy the magic of Diffusion models!
This report summarizes scanner evidence for modelscope/DiffSynth-Studio, a Python project with 13,216 GitHub stars. SynthScan examined 166,936 lines of code across 1578 source files, recording 1238 pattern matches distributed across 19 syntactic categories. The legacy adjusted pattern signal is 8.8 (Moderate 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? 760 finding(s).
Are imports declared and package/API names verified? 387 finding(s).
Are wrappers, guards, and repeated structures necessary? 602 finding(s).
Which changes need focused human review? 1238 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 1238 distinct pattern matches across 19 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 | docs/zh/Research_Tutorial/train_from_scratch.py | 8 | CODE | |
| LOW | docs/en/Research_Tutorial/train_from_scratch.py | 8 | CODE | |
| LOW | diffsynth/__init__.py | 1 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 1 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 2 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 3 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 4 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 5 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 6 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 7 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 8 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 9 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 10 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 11 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 12 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 13 | CODE | |
| LOW | diffsynth/metrics/__init__.py | 14 | CODE | |
| LOW | diffsynth/metrics/lpips.py | 5 | CODE | |
| LOW | diffsynth/core/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/__init__.py | 3 | CODE | |
| LOW | diffsynth/core/__init__.py | 4 | CODE | |
| LOW | diffsynth/core/__init__.py | 5 | CODE | |
| LOW | diffsynth/core/__init__.py | 6 | CODE | |
| LOW | diffsynth/core/__init__.py | 7 | CODE | |
| LOW | diffsynth/core/__init__.py | 8 | CODE | |
| LOW | diffsynth/core/gradient/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/quant/__init__.py | 3 | CODE | |
| LOW | diffsynth/core/attention/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/attention/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/offload_training/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/loader/__init__.py | 3 | CODE | |
| LOW | diffsynth/core/loader/model.py | 6 | CODE | |
| LOW | diffsynth/core/vram/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/vram/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 1 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/device/__init__.py | 2 | CODE | |
| LOW | diffsynth/core/device/npu_compatible_device.py | 14 | CODE | |
| LOW | diffsynth/core/data/__init__.py | 1 | CODE | |
| 326 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | docs/zh/Research_Tutorial/train_from_scratch.py | 155 | CODE | |
| LOW | docs/en/Research_Tutorial/train_from_scratch.py | 155 | CODE | |
| LOW | diffsynth/metrics/hpsv3.py | 15 | CODE | |
| LOW | diffsynth/metrics/unified_reward_edit.py | 20 | CODE | |
| LOW | diffsynth/metrics/qwen_image_bench.py | 17 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 116 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 101 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 160 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 171 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 163 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 182 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 174 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 191 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 185 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 202 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 194 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 210 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 218 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 227 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 221 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 230 | CODE | |
| LOW | diffsynth/core/loader/model.py | 11 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 10 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 38 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 95 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 214 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 296 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 375 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 446 | CODE | |
| LOW | diffsynth/core/data/unified_dataset.py | 29 | CODE | |
| LOW | diffsynth/core/data/unified_dataset.py | 41 | CODE | |
| LOW | diffsynth/pipelines/anima_image.py | 74 | CODE | |
| LOW | diffsynth/pipelines/stable_diffusion_xl.py | 90 | CODE | |
| LOW | diffsynth/pipelines/ltx2_audio_video.py | 714 | CODE | |
| LOW | diffsynth/pipelines/ltx2_audio_video.py | 188 | CODE | |
| LOW | diffsynth/pipelines/lingbot_video.py | 83 | CODE | |
| LOW | diffsynth/pipelines/flux_image.py | 1298 | CODE | |
| LOW | diffsynth/pipelines/flux_image.py | 131 | CODE | |
| LOW | diffsynth/pipelines/flux_image.py | 194 | CODE | |
| LOW | diffsynth/pipelines/hidream_o1_image.py | 60 | CODE | |
| LOW | diffsynth/pipelines/krea2.py | 280 | CODE | |
| LOW | diffsynth/pipelines/krea2.py | 248 | CODE | |
| LOW | diffsynth/pipelines/krea2.py | 63 | CODE | |
| LOW | diffsynth/pipelines/minimax_h3_audio_video.py | 944 | CODE | |
| LOW | diffsynth/pipelines/minimax_h3_audio_video.py | 94 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 1455 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 1817 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 1908 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 120 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 198 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 937 | CODE | |
| LOW | diffsynth/pipelines/ace_step.py | 573 | CODE | |
| LOW | diffsynth/pipelines/ace_step.py | 90 | CODE | |
| LOW | diffsynth/pipelines/mova_audio_video.py | 348 | CODE | |
| LOW | diffsynth/pipelines/mova_audio_video.py | 115 | CODE | |
| LOW | diffsynth/pipelines/qwen_video_edit.py | 356 | CODE | |
| LOW | diffsynth/pipelines/qwen_video_edit.py | 293 | CODE | |
| LOW | diffsynth/pipelines/qwen_video_edit.py | 90 | CODE | |
| LOW | diffsynth/pipelines/stable_diffusion.py | 66 | CODE | |
| LOW | diffsynth/pipelines/ideogram4.py | 270 | CODE | |
| 219 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | docs/zh/API_Reference/core/quant.md | 249 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | docs/zh/Diffusion_Templates/Template_Model_Training.md | 31 | class DataProcessor: | CODE |
| LOW | docs/zh/Diffusion_Templates/Template_Model_Training.md | 122 | class DataProcessor: | CODE |
| LOW | docs/zh/Diffusion_Templates/Template_Model_Training.md | 147 | class DataProcessor: | CODE |
| LOW | …zh/Developer_Guide/Integrating_Quantization_Backend.md | 140 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | docs/en/API_Reference/core/quant.md | 249 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | docs/en/Diffusion_Templates/Template_Model_Training.md | 31 | class DataProcessor: | CODE |
| LOW | docs/en/Diffusion_Templates/Template_Model_Training.md | 122 | class DataProcessor: | CODE |
| LOW | docs/en/Diffusion_Templates/Template_Model_Training.md | 142 | class DataProcessor: | CODE |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 140 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | diffsynth/core/gradient/gradient_checkpoint.py | 17 | def create_custom_forward_use_reentrant(module): | CODE |
| LOW | diffsynth/core/gradient/gradient_checkpoint.py | 30 | def gradient_checkpoint_forward( | CODE |
| LOW | diffsynth/core/quant/config.py | 213 | def prepare_for_prequantized_load(self, model: torch.nn.Module, compute_dtype: torch.dtype = torch.bfloat16): | CODE |
| LOW | diffsynth/core/quant/config.py | 260 | def measure_quantization_error(self, model: torch.nn.Module, compute_device="cuda", verbose=True) -> list: | CODE |
| LOW | diffsynth/core/quant/config.py | 469 | def prepare_for_prequantized_load(self, model: torch.nn.Module, compute_dtype: torch.dtype = torch.bfloat16): | CODE |
| LOW | diffsynth/core/quant/config.py | 528 | def measure_quantization_error(self, model: torch.nn.Module, compute_device="cuda", verbose=True) -> list: | CODE |
| LOW | diffsynth/core/quant/config.py | 540 | def _distinct_backend_configs(self): | CODE |
| LOW | diffsynth/core/quant/base.py | 74 | def create_quantized_linear_shell(self, linear: torch.nn.Linear, compute_dtype: torch.dtype) -> torch.nn.Module: | CODE |
| LOW | diffsynth/core/quant/backends/entropack.py | 90 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | diffsynth/core/quant/backends/torchao.py | 70 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | diffsynth/core/quant/backends/torchao.py | 134 | def _mx_dynamic_activation_mx_weight(elem_dtype): | CODE |
| LOW | diffsynth/core/quant/backends/bitsandbytes.py | 91 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW | diffsynth/core/quant/backends/comfy_kitchen.py | 233 | def create_quantized_linear_shell(self, linear, compute_dtype): | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 61 | def initialize_attention_priority(): | CODE |
| LOW | diffsynth/core/attention/attention.py | 119 | def torch_sdpa_sliding_window(q: torch.Tensor, k: torch.Tensor, v: torch.Tensor, sliding_window: int, q_pattern="b n s d | CODE |
| LOW | diffsynth/core/attention/attention.py | 194 | def flash_attention_customized_kernel(q: torch.Tensor, k: torch.Tensor, v: torch.Tensor, q_pattern="b n s d", k_pattern= | CODE |
| LOW | diffsynth/core/offload_training/manager.py | 138 | def _find_orphan_params_and_buffers(self, model: nn.Module, managed_param_ids: set): | CODE |
| LOW | diffsynth/core/npu_patch/npu_fused_operator.py | 16 | def rms_norm_forward_transformers_npu(self, hidden_states): | CODE |
| LOW | diffsynth/core/loader/config.py | 67 | def parse_original_file_pattern(self): | CODE |
| LOW | diffsynth/core/loader/model.py | 130 | def load_model_with_disk_offload(model_class, path, config=None, torch_dtype=torch.bfloat16, device="cpu", state_dict_co | CODE |
| LOW | diffsynth/core/loader/file.py | 26 | def load_state_dict_from_safetensors(file_path, torch_dtype=None, device="cpu"): | CODE |
| LOW | diffsynth/core/loader/file.py | 36 | def load_metadata_from_safetensors(file_path): | CODE |
| LOW | diffsynth/core/loader/file.py | 73 | def convert_state_dict_keys_to_single_str(state_dict, with_shape=True): | CODE |
| LOW | diffsynth/core/loader/file.py | 107 | def load_keys_dict_from_safetensors(file_path): | CODE |
| LOW | diffsynth/core/loader/file.py | 115 | def convert_state_dict_to_keys_dict(state_dict): | CODE |
| LOW | diffsynth/core/loader/file.py | 131 | def convert_keys_dict_to_single_str(state_dict, with_shape=True): | CODE |
| LOW | diffsynth/core/vram/initialization.py | 6 | def skip_model_initialization(device=torch.device("meta")): | CODE |
| LOW | diffsynth/core/vram/layers.py | 563 | def _materialize_root_params_from_disk(model: torch.nn.Module, disk_map: DiskMap): | CODE |
| LOW | diffsynth/core/vram/layers.py | 574 | def enable_vram_management_recursively(model: torch.nn.Module, module_map: dict, vram_config: dict, vram_limit=None, nam | CODE |
| LOW | diffsynth/core/device/npu_compatible_device.py | 108 | def get_available_device_type(): | CODE |
| LOW | diffsynth/core/data/unified_dataset.py | 63 | def search_for_cached_data_files(self, path): | CODE |
| LOW | diffsynth/pipelines/stable_diffusion_xl.py | 311 | def model_fn_stable_diffusion_xl( | CODE |
| LOW | diffsynth/pipelines/ltx2_audio_video.py | 387 | def build_video_keyframes_mask(self, pipe, video_latent_shape): | CODE |
| LOW | diffsynth/pipelines/ltx2_audio_video.py | 551 | def apply_input_images_to_latents(self, latents, input_latents, input_indexes, input_strength=1.0, input_latents_vid | CODE |
| LOW | diffsynth/pipelines/flux_image.py | 31 | def process_single_controlnet(self, controlnet_input: ControlNetInput, conditioning: torch.Tensor, **kwargs): | CODE |
| LOW | diffsynth/pipelines/flux_image.py | 481 | def apply_controlnet_mask_on_latents(self, pipe, latents, mask): | CODE |
| LOW | diffsynth/pipelines/flux_image.py | 488 | def apply_controlnet_mask_on_image(self, pipe, image, mask): | CODE |
| LOW | diffsynth/pipelines/flux_image.py | 1117 | def extract_arcface_bgr_embedding(self, in_image, landmark, device): | CODE |
| LOW | diffsynth/pipelines/flux_image.py | 1151 | def parse_lora_encoder_inputs(self, lora_encoder_inputs): | CODE |
| LOW⚡ | diffsynth/pipelines/minimax_h3_audio_video.py | 478 | def _resolve_reference_image_shape(self, pipe: MiniMaxH3Pipeline, width: int, height: int, short_edge: int): | CODE |
| LOW⚡ | diffsynth/pipelines/minimax_h3_audio_video.py | 482 | def _resolve_reference_video_shape(self, pipe: MiniMaxH3Pipeline, width: int, height: int, short_edge: int, max_pixe | CODE |
| LOW⚡ | diffsynth/pipelines/minimax_h3_audio_video.py | 486 | def _trim_reference_video_length(self, pipe: MiniMaxH3Pipeline, frame_count: int) -> int: | CODE |
| LOW | diffsynth/pipelines/wan_video.py | 1438 | def wantodance_get_single_freqs(freqs, frame_num, fps): | CODE |
| LOW | diffsynth/pipelines/wan_video.py | 1722 | def create_custom_forward_vap(block, vap): | CODE |
| LOW | diffsynth/pipelines/ace_step.py | 193 | def switch_noncover_condition(self, inputs_shared, inputs_posi, inputs_nega, progress_id): | CODE |
| LOW | diffsynth/pipelines/ace_step.py | 422 | def _get_silence_latent_slice(self, pipe, length: int) -> torch.Tensor: | CODE |
| LOW | diffsynth/pipelines/mova_audio_video.py | 348 | def model_fn_mova_audio_video( | CODE |
| LOW | diffsynth/pipelines/stable_diffusion.py | 210 | def model_fn_stable_diffusion( | CODE |
| LOW | diffsynth/pipelines/qwen_image.py | 500 | def apply_controlnet_mask_on_latents(self, pipe, latents, mask): | CODE |
| LOW | diffsynth/pipelines/qwen_image.py | 507 | def apply_controlnet_mask_on_image(self, pipe, image, mask): | CODE |
| 172 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | docs/zh/Model_Details/Ideogram-4.md | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | docs/en/Model_Details/Ideogram-4.md | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | …/model_inference_low_vram/ideogram-4-bf16-repackage.py | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | …ideogram4/model_inference/ideogram-4-bf16-repackage.py | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | examples/ideogram4/model_inference/ideogram-4-fp8.py | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | examples/ideogram4/model_inference/ideogram-4-nf4.py | 0 | { "high_level_description": "a medium-shot photograph of formula 1 driver max verstappen wearing his red bull racing rac | STRING |
| HIGH | docs/en/Model_Details/MiniMax-H3.md | 0 | widen the half-open frame range [start, end) to whole clips. frames count from 0. | STRING |
| HIGH | …nimax_h3/model_inference_low_vram/MiniMax-H3-Retake.py | 0 | widen the half-open frame range [start, end) to whole clips. frames count from 0. | STRING |
| HIGH | …amples/minimax_h3/model_inference/MiniMax-H3-Retake.py | 0 | widen the half-open frame range [start, end) to whole clips. frames count from 0. | STRING |
| HIGH | diffsynth/models/ltx2_dit.py | 0 | this matches the implementation in denoising diffusion probabilistic models: create sinusoidal timestep embeddings. args | STRING |
| HIGH | diffsynth/models/flux2_dit.py | 0 | this matches the implementation in denoising diffusion probabilistic models: create sinusoidal timestep embeddings. args | STRING |
| HIGH | diffsynth/models/boogu_image_dit.py | 0 | this matches the implementation in denoising diffusion probabilistic models: create sinusoidal timestep embeddings. args | STRING |
| HIGH | diffsynth/configs/model_configs.py | 0 | offical model repo: https://www.modelscope.cn/models/lightricks/ltx-2 repackaged model repo: https://www.modelscope.cn/m | STRING |
| HIGH | …s/ltx2/model_inference_low_vram/LTX-2-T2AV-OneStage.py | 0 | offical model repo: https://www.modelscope.cn/models/lightricks/ltx-2 repackaged model repo: https://www.modelscope.cn/m | STRING |
| HIGH | …s/ltx2/model_inference_low_vram/LTX-2-T2AV-TwoStage.py | 0 | offical model repo: https://www.modelscope.cn/models/lightricks/ltx-2 repackaged model repo: https://www.modelscope.cn/m | STRING |
| HIGH | examples/ltx2/model_inference/LTX-2-T2AV-OneStage.py | 0 | offical model repo: https://www.modelscope.cn/models/lightricks/ltx-2 repackaged model repo: https://www.modelscope.cn/m | STRING |
| HIGH | examples/ltx2/model_inference/LTX-2-T2AV-TwoStage.py | 0 | offical model repo: https://www.modelscope.cn/models/lightricks/ltx-2 repackaged model repo: https://www.modelscope.cn/m | STRING |
| HIGH | …1/model_inference_low_vram/Qwen-Image-2.1-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …n_image_21/model_inference/Qwen-Image-2.1-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …ax_h3/model_inference_low_vram/MiniMax-H3-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …les/minimax_h3/model_inference/MiniMax-H3-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …ge/model_inference_low_vram/Z-Image-Turbo-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …les/z_image/model_inference/Z-Image-Turbo-EntroPack.py | 0 | download the pre-quantized package and its json sidecar (a ready-to-use model_configs entry), return a loadable modelcon | STRING |
| HIGH | …odel_inference_low_vram/acestep15xlsft-vocals2music.py | 0 | [verse 1] 深夜的屏幕，微光在闪烁 指尖敲击着，未知的脉络 不再是孤岛，独自去摸索 这里有一片海，等待你停泊 从零到一的距离，不再遥远 丰富的模型，静候被点燃 打破围墙的界限，推倒高墙 让智慧的火花，自由地碰撞 [pre-choru | STRING |
| HIGH | …ce_step/model_inference/acestep15xlsft-vocals2music.py | 0 | [verse 1] 深夜的屏幕，微光在闪烁 指尖敲击着，未知的脉络 不再是孤岛，独自去摸索 这里有一片海，等待你停泊 从零到一的距离，不再遥远 丰富的模型，静候被点燃 打破围墙的界限，推倒高墙 让智慧的火花，自由地碰撞 [pre-choru | STRING |
| HIGH | …_training/validate_full/acestep15xlsft-vocals2music.py | 0 | [verse 1] 深夜的屏幕，微光在闪烁 指尖敲击着，未知的脉络 不再是孤岛，独自去摸索 这里有一片海，等待你停泊 从零到一的距离，不再遥远 丰富的模型，静候被点燃 打破围墙的界限，推倒高墙 让智慧的火花，自由地碰撞 [pre-choru | STRING |
| HIGH | examples/yue2/model_inference_low_vram/YuE2-Edit.py | 0 | [verse] neon fades along the lane footsteps keep the time of rain fold the night and leave it here morning has a sky to | STRING |
| HIGH | examples/yue2/model_inference_low_vram/YuE2.py | 0 | [verse] neon fades along the lane footsteps keep the time of rain fold the night and leave it here morning has a sky to | STRING |
| HIGH | examples/yue2/model_inference/YuE2-Edit.py | 0 | [verse] neon fades along the lane footsteps keep the time of rain fold the night and leave it here morning has a sky to | STRING |
| HIGH | examples/yue2/model_inference/YuE2.py | 0 | [verse] neon fades along the lane footsteps keep the time of rain fold the night and leave it here morning has a sky to | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | diffsynth/core/quant/base.py | 163 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 61 | CODE | |
| LOW | diffsynth/core/attention/attention.py | 230 | CODE | |
| LOW | diffsynth/core/loader/config.py | 84 | CODE | |
| LOW | diffsynth/core/loader/model.py | 11 | CODE | |
| LOW | diffsynth/core/loader/file.py | 57 | CODE | |
| LOW | diffsynth/core/loader/file.py | 73 | CODE | |
| LOW | diffsynth/core/loader/file.py | 131 | CODE | |
| LOW | diffsynth/core/vram/layers.py | 574 | CODE | |
| LOW | diffsynth/core/data/operators.py | 325 | CODE | |
| LOW | diffsynth/core/data/unified_dataset.py | 71 | CODE | |
| LOW | diffsynth/core/data/unified_dataset.py | 90 | CODE | |
| LOW | diffsynth/pipelines/ltx2_audio_video.py | 562 | CODE | |
| LOW | diffsynth/pipelines/flux_image.py | 1298 | CODE | |
| LOW | diffsynth/pipelines/flux_image.py | 115 | CODE | |
| LOW | diffsynth/pipelines/hidream_o1_image.py | 166 | CODE | |
| LOW | diffsynth/pipelines/minimax_music3.py | 42 | CODE | |
| LOW | diffsynth/pipelines/minimax_h3_audio_video.py | 58 | CODE | |
| LOW | diffsynth/pipelines/minimax_h3_audio_video.py | 260 | CODE | |
| LOW | diffsynth/pipelines/minimax_h3_audio_video.py | 700 | CODE | |
| LOW | diffsynth/pipelines/wan_video.py | 1455 | CODE | |
| LOW | diffsynth/pipelines/ace_step.py | 486 | CODE | |
| LOW | diffsynth/utils/controlnet/annotator.py | 10 | CODE | |
| LOW | …ynth/utils/state_dict_converters/ltx25_text_encoder.py | 1 | CODE | |
| LOW | …synth/utils/state_dict_converters/ltx2_text_encoder.py | 1 | CODE | |
| LOW | …synth/utils/state_dict_converters/ltx2_text_encoder.py | 24 | CODE | |
| LOW | …ffsynth/utils/state_dict_converters/flux_controlnet.py | 4 | CODE | |
| LOW | …ls/state_dict_converters/qwen_image_21_text_encoder.py | 1 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/image_metrics.py | 12 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/image_metrics.py | 79 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/image_metrics.py | 93 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/image_metrics.py | 107 | CODE | |
| LOW | …utils/state_dict_converters/minimax_h3_text_encoder.py | 30 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/flux_vae.py | 267 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/flux_vae.py | 326 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/flux_dit.py | 4 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/flux_dit.py | 95 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/ltx2_video_vae.py | 1 | CODE | |
| LOW | diffsynth/utils/state_dict_converters/ltx2_video_vae.py | 16 | CODE | |
| LOW | …th/utils/state_dict_converters/stable_diffusion_vae.py | 1 | CODE | |
| LOW | diffsynth/utils/lora/minimax_h3.py | 40 | CODE | |
| LOW | diffsynth/utils/lora/sdxl.py | 766 | CODE | |
| LOW | diffsynth/utils/lora/flux.py | 84 | CODE | |
| LOW | diffsynth/utils/lora/flux.py | 113 | CODE | |
| LOW | diffsynth/utils/data/minimax_h3.py | 79 | CODE | |
| LOW | diffsynth/models/longcat_video_dit.py | 697 | CODE | |
| LOW | diffsynth/models/yue2_vae.py | 395 | CODE | |
| LOW | diffsynth/models/stable_diffusion_unet.py | 63 | CODE | |
| LOW | diffsynth/models/hpsv3.py | 109 | CODE | |
| LOW | diffsynth/models/mova_dual_tower_bridge.py | 136 | CODE | |
| LOW | diffsynth/models/flux2_vae.py | 78 | CODE | |
| LOW | diffsynth/models/flux2_vae.py | 489 | CODE | |
| LOW | diffsynth/models/flux2_vae.py | 693 | CODE | |
| LOW | diffsynth/models/flux2_vae.py | 743 | CODE | |
| LOW | diffsynth/models/qwen_image_image2lora.py | 119 | CODE | |
| LOW | diffsynth/models/minimax_h3_video_vae.py | 469 | CODE | |
| LOW | diffsynth/models/minimax_h3_video_vae.py | 584 | CODE | |
| LOW | diffsynth/models/z_image_image2lora.py | 103 | CODE | |
| LOW | diffsynth/models/z_image_image2lora.py | 179 | CODE | |
| LOW | diffsynth/models/mova_audio_vae.py | 590 | CODE | |
| 78 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | docs/zh/Developer_Guide/Integrating_Your_Model.md | 5 | ## Step 1: 集成模型结构代码 | COMMENT |
| LOW | docs/zh/Developer_Guide/Integrating_Your_Model.md | 80 | ## Step 2: 模型文件格式转换 | COMMENT |
| LOW | docs/zh/Developer_Guide/Integrating_Your_Model.md | 105 | ## Step 3: 编写模型 Config | COMMENT |
| LOW | docs/zh/Developer_Guide/Integrating_Your_Model.md | 150 | ## Step 4: 检验模型是否能被识别和加载 | COMMENT |
| LOW | docs/zh/Developer_Guide/Integrating_Your_Model.md | 184 | ## Step 5: 编写模型显存管理方案 | COMMENT |
| LOW | docs/en/Pipeline_Usage/Quantization.md | 344 | ### Step 1: Save the Quantized Weights | COMMENT |
| LOW | docs/en/Pipeline_Usage/Quantization.md | 373 | ### Step 2: Register the Hash in the Model Config | COMMENT |
| LOW | docs/en/Pipeline_Usage/Quantization.md | 399 | ### Step 3: Load for Inference | COMMENT |
| LOW⚡ | docs/en/Training/Split_Training.md | 145 | # Step 1: initialize X_0, the set of directly related units | COMMENT |
| LOW⚡ | docs/en/Training/Split_Training.md | 150 | # Step 2: build the data dependency edges E and the updating chain c(p) of each parameter | COMMENT |
| LOW⚡ | docs/en/Training/Split_Training.md | 154 | # Step 3: fixed-point iteration X_{k+1} = X_k ∪ F(X_k) ∪ U(X_k) | COMMENT |
| LOW⚡ | docs/en/Training/Split_Training.md | 162 | # Step 4: output the bipartition (V1, V2) | COMMENT |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 22 | ## Step 1: Write the Quantized Linear | COMMENT |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 100 | ## Step 2: Write the Backend | COMMENT |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 154 | ## Step 3: Write the Backend Config | COMMENT |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 165 | ## Step 4: Register the Quantization Method | COMMENT |
| LOW | …en/Developer_Guide/Integrating_Quantization_Backend.md | 198 | ## Step 5: Self-Check | COMMENT |
| LOW | docs/en/Developer_Guide/Integrating_Your_Model.md | 5 | ## Step 1: Integrate Model Architecture Code | COMMENT |
| LOW | docs/en/Developer_Guide/Integrating_Your_Model.md | 80 | ## Step 2: Model File Format Conversion | COMMENT |
| LOW | docs/en/Developer_Guide/Integrating_Your_Model.md | 105 | ## Step 3: Writing Model Config | COMMENT |
| LOW | docs/en/Developer_Guide/Integrating_Your_Model.md | 150 | ## Step 4: Verifying Whether the Model Can Be Recognized and Loaded | COMMENT |
| LOW | docs/en/Developer_Guide/Integrating_Your_Model.md | 184 | ## Step 5: Writing Model VRAM Management Scheme | COMMENT |
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1237 | # Step 1: Crop first to avoid exceeding target dimensions | COMMENT |
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1242 | # Step 2: Calculate padding needed for time and frequency dimensions | COMMENT |
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1246 | # Step 3: Apply padding if needed | COMMENT |
| LOW | diffsynth/models/ltx2_audio_vae.py | 1258 | # Step 4: Final safety crop to ensure exact target shape | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 219 | # Step 1: Self-attention with adaptive layer norm (AdaLN) | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 230 | # Step 2: Cross-attention (if enabled) for conditioning on encoder outputs | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 241 | # Step 3: Feed-forward (MLP) with adaptive layer norm | COMMENT |
| LOW⚡ | diffsynth/models/boogu_image_dit.py | 2296 | # Step 3: image self-attention. | COMMENT |
| LOW⚡ | diffsynth/models/boogu_image_dit.py | 2304 | # Step 4: residual updates. | COMMENT |
| LOW | diffsynth/models/boogu_image_dit.py | 2250 | # Step 1: modulation for both streams. | COMMENT |
| LOW | diffsynth/models/boogu_image_dit.py | 2267 | # Step 2: joint attention on [instruct + img]. | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | docs/zh/conf.py | 1 | # Configuration file for the Sphinx documentation builder. | COMMENT |
| LOW | docs/zh/conf.py | 81 | # | COMMENT |
| LOW | docs/zh/conf.py | 101 | # html_theme_path = [sphinx_book_theme.get_html_theme_path()] | COMMENT |
| LOW | docs/en/conf.py | 1 | # Configuration file for the Sphinx documentation builder. | COMMENT |
| LOW | docs/en/conf.py | 81 | # | COMMENT |
| LOW | docs/en/conf.py | 101 | # html_theme_path = [sphinx_book_theme.get_html_theme_path()] | COMMENT |
| LOW | diffsynth/version.py | 1 | # Make sure to modify __release_datetime__ to release time when making official release. | COMMENT |
| LOW | diffsynth/models/stable_diffusion_unet.py | 1 | # Copyright 2025 The HuggingFace Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/flux2_vae.py | 1 | # Copyright 2025 The HuggingFace Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/ace_step_conditioner.py | 1 | # Copyright 2025 The ACESTEO Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/ace_step_vae.py | 1 | # Copyright 2025 The ACESTEO Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 1 | # Copyright 2025 The ACESTEO Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/hidream_o1_image_dit.py | 1 | # Copyright 2025 The Qwen Team and The HuggingFace Inc. team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/ace_step_tokenizer.py | 1 | # Copyright 2025 The ACESTEO Team. All rights reserved. | COMMENT |
| LOW | diffsynth/models/stable_diffusion_vae.py | 1 | # Copyright 2025 The HuggingFace Team. All rights reserved. | COMMENT |
| LOW | …es/qwen_image_21/model_training/lora/Qwen-Image-2.1.sh | 21 | # modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "qwen_image/Qwen-Image-Edit-2511/*" | COMMENT |
| LOW | …n_image_21/model_training/lora/Qwen-Image-2.1-Turbo.sh | 21 | # Edit | COMMENT |
| LOW | …s/ltx2/model_inference_low_vram/LTX-2-T2AV-TwoStage.py | 41 | # use the following modelconfig if you want to initialize model from offical checkpoints from "Lightricks/LTX-2" | COMMENT |
| LOW | examples/ltx2/model_inference/LTX-2-T2AV-TwoStage.py | 41 | # pipe = LTX2AudioVideoPipeline.from_pretrained( | COMMENT |
| LOW | examples/ltx2/model_training/lora/LTX-2-T2AV-noaudio.sh | 1 | modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "ltx2/LTX-2-T2AV-noaudio/*" --local_d | COMMENT |
| LOW | examples/ltx2/model_training/lora/LTX-2-T2AV-splited.sh | 1 | modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "ltx2/LTX-2-T2AV-splited/*" --local_d | COMMENT |
| LOW | …ples/flux2/model_training/full/FLUX.2-klein-base-9B.sh | 21 | COMMENT | |
| LOW | examples/flux2/model_training/full/FLUX.2-klein-4B.sh | 21 | # accelerate launch examples/flux2/model_training/train.py \ | COMMENT |
| LOW | examples/flux2/model_training/full/FLUX.2-klein-9B.sh | 21 | COMMENT | |
| LOW | …s/flux2/model_training/full/TreeAdapter-KleinBase4B.sh | 1 | # TreeAdapter-KleinBase4B is a multi-LoRA system that supports inference only. | COMMENT |
| LOW | …ples/flux2/model_training/full/FLUX.2-klein-base-4B.sh | 21 | # accelerate launch examples/flux2/model_training/train.py \ | COMMENT |
| LOW | …l_training/special/npu_training/FLUX.2-klein-9B-NPU.sh | 21 | # Edit | COMMENT |
| LOW | …ples/flux2/model_training/lora/FLUX.2-klein-base-9B.sh | 21 | # modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "qwen_image/Qwen-Image-Edit-2511/*" | COMMENT |
| LOW | examples/flux2/model_training/lora/FLUX.2-klein-4B.sh | 21 | # modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "qwen_image/Qwen-Image-Edit-2511/*" | COMMENT |
| LOW | examples/flux2/model_training/lora/FLUX.2-klein-9B.sh | 21 | # modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "qwen_image/Qwen-Image-Edit-2511/*" | COMMENT |
| LOW | …ples/flux2/model_training/lora/FLUX.2-klein-base-4B.sh | 21 | # modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "qwen_image/Qwen-Image-Edit-2511/*" | COMMENT |
| LOW | …image/model_inference_low_vram/Qwen-Image-Edit-2509.py | 41 | # edit_image = [Image.open("image.jpg")] | COMMENT |
| LOW | …l_inference_low_vram/Qwen-Image-Edit-2511-Lightning.py | 61 | # edit_image = [Image.open("image.jpg")] | COMMENT |
| LOW | …s/qwen_image/model_inference/FireRed-Image-Edit-1.1.py | 41 | # edit_image = [Image.open("image.jpg")] | COMMENT |
| LOW | …s/qwen_image/model_inference/FireRed-Image-Edit-1.0.py | 41 | # edit_image = [Image.open("image.jpg")] | COMMENT |
| LOW | …les/qwen_image/model_inference/Qwen-Image-Edit-2511.py | 41 | # Please use a list to input `edit_image`, even if the input contains only one image. | COMMENT |
| LOW | …training/full/Qwen-Image-Blockwise-ControlNet-Canny.sh | 21 | # and then start training with a high learning rate (1e-3). | COMMENT |
| LOW | …aining/full/Qwen-Image-Blockwise-ControlNet-Inpaint.sh | 21 | # and then start training with a high learning rate (1e-3). | COMMENT |
| LOW | …training/full/Qwen-Image-Blockwise-ControlNet-Depth.sh | 21 | # and then start training with a high learning rate (1e-3). | COMMENT |
| LOW | …_video/model_inference/lingbot-video-dense-1.3b_t2v.py | 61 | # from examples.lingbot_video.model_training.scripts.prompt_rewriter import rewrite_prompt | COMMENT |
| LOW | …m_o1_image/model_training/full/HiDream-O1-Image-Dev.sh | 21 | # --dataset_base_path data/diffsynth_example_dataset/qwen_image/Qwen-Image-Edit-2511 \ | COMMENT |
| LOW | …dream_o1_image/model_training/full/HiDream-O1-Image.sh | 21 | # --dataset_base_path data/diffsynth_example_dataset/qwen_image/Qwen-Image-Edit-2511 \ | COMMENT |
| LOW | …m_o1_image/model_training/lora/HiDream-O1-Image-Dev.sh | 21 | COMMENT | |
| LOW | …dream_o1_image/model_training/lora/HiDream-O1-Image.sh | 21 | COMMENT | |
| LOW | …mples/z_image/model_training/full/Z-Image-Omni-Base.sh | 21 | # accelerate launch --config_file examples/z_image/model_training/full/accelerate_config.yaml examples/z_image/model_tra | COMMENT |
| LOW | …mage/model_training/validate_lora/Z-Image-Omni-Base.py | 21 | image.save("image.jpg") | COMMENT |
| LOW | …mage/model_training/validate_full/Z-Image-Omni-Base.py | 21 | image = pipe(prompt=prompt, seed=42, rand_device="cuda", num_inference_steps=40, cfg_scale=4) | COMMENT |
| LOW | …mples/z_image/model_training/lora/Z-Image-Omni-Base.sh | 21 | # Image(s) to image training | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | diffsynth/models/ltx2_audio_vae.py | 1290 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | diffsynth/models/ltx2_audio_vae.py | 1293 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 58 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 60 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 122 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 124 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 309 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 311 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 354 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 356 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 405 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 407 | # ══════════════════════════════════════════════════════════════════════════════ | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 497 | # ── AR forward (standard causal LM, KV cached) ─────────────────── | COMMENT |
| MEDIUM | diffsynth/models/yue2_mot.py | 608 | # ── NAR velocity (flow matching, no KV cache) ──────────────────── | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | diffsynth/core/quant/config.py | 31 | except Exception as error: | CODE |
| LOW | diffsynth/core/quant/base.py | 154 | except Exception as error: | CODE |
| LOW | diffsynth/core/quant/base.py | 235 | except Exception as error: | CODE |
| LOW | diffsynth/core/quant/backends/comfy_kitchen.py | 183 | except Exception as error: | CODE |
| MEDIUM | diffsynth/core/quant/backends/comfy_kitchen.py | 180 | def _parse_marker(marker_tensor, layer_name): | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 10 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 18 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 24 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 30 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 36 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 42 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 49 | except Exception: | CODE |
| LOW⚡ | diffsynth/core/attention/attention.py | 57 | except Exception: | CODE |
| LOW | diffsynth/core/offload_training/memory_buffer.py | 135 | except Exception: | CODE |
| LOW | diffsynth/core/data/operators.py | 23 | except Exception as e: | CODE |
| LOW | diffsynth/core/data/operators.py | 310 | except Exception as e: | CODE |
| LOW | diffsynth/core/data/operators.py | 339 | except Exception as e: | CODE |
| LOW | diffsynth/pipelines/ace_step.py | 449 | except Exception as e: | CODE |
| LOW | diffsynth/utils/data/__init__.py | 209 | except Exception as e: | CODE |
| LOW | diffsynth/models/flux2_vae.py | 808 | except Exception as e: | STRING |
| LOW | diffsynth/models/unified_reward_edit.py | 269 | except Exception: | CODE |
| LOW | diffsynth/diffusion/runner.py | 29 | except Exception as e: | CODE |
| LOW | diffsynth/diffusion/runner.py | 49 | except Exception as e: | CODE |
| LOW | …ingbot_video/model_training/scripts/prompt_rewriter.py | 98 | except Exception: | CODE |
| LOW | …ingbot_video/model_training/scripts/prompt_rewriter.py | 104 | except Exception: | CODE |
| LOW⚡ | examples/dev_tools/webui.py | 39 | except Exception as e: | CODE |
| MEDIUM⚡ | examples/dev_tools/webui.py | 41 | print(f"Error {error_value}:\n{error_message}") | CODE |
| MEDIUM | examples/dev_tools/webui.py | 36 | def catch_error(error_value): | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | diffsynth/models/qwen_image_21_vae.py | 401 | # Create the components | STRING |
| MEDIUM | diffsynth/models/flux_vae.py | 11 | # Create a mask with shape (height, width). | COMMENT |
| MEDIUM | diffsynth/models/qwen_image_vae.py | 318 | # Create the components | STRING |
| MEDIUM | diffsynth/models/ltx2_video_vae.py | 133 | # Create a tensor containing the size of a single patch: | COMMENT |
| MEDIUM | diffsynth/models/ltx2_video_vae.py | 216 | # Define the size of the channels after the first convolution | COMMENT |
| MEDIUM | …les/flux/model_inference_low_vram/FLUX.1-dev-EliGen.py | 19 | # Create a blank image for overlays | COMMENT |
| MEDIUM | examples/flux/model_inference/FLUX.1-dev-EliGen.py | 9 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …en_image/model_inference_low_vram/Qwen-Image-EliGen.py | 9 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …image/model_inference_low_vram/Qwen-Image-EliGen-V2.py | 8 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …e/model_inference_low_vram/Qwen-Image-EliGen-Poster.py | 9 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …amples/qwen_image/model_inference/Qwen-Image-EliGen.py | 9 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …les/qwen_image/model_inference/Qwen-Image-EliGen-V2.py | 8 | # Create a blank image for overlays | COMMENT |
| MEDIUM | …qwen_image/model_inference/Qwen-Image-EliGen-Poster.py | 9 | # Create a blank image for overlays | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | diffsynth/core/npu_patch/npu_fused_operator.py | 5 | except: | CODE |
| HIGH | diffsynth/models/unified_reward_edit.py | 269 | except Exception: | CODE |
| HIGH | diffsynth/models/ltx25_diffusion_video_vae.py | 4989 | except ValueError: | CODE |
| HIGH | diffsynth/models/qwen_image_bench.py | 318 | except json.JSONDecodeError: | CODE |
| HIGH | diffsynth/models/qwen_image_bench.py | 324 | except json.JSONDecodeError: | CODE |
| HIGH | …ingbot_video/model_training/scripts/prompt_rewriter.py | 98 | except Exception: | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1237 | # Step 1: Crop first to avoid exceeding target dimensions | COMMENT |
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1242 | # Step 2: Calculate padding needed for time and frequency dimensions | COMMENT |
| LOW⚡ | diffsynth/models/ltx2_audio_vae.py | 1246 | # Step 3: Apply padding if needed | COMMENT |
| LOW | diffsynth/models/ltx2_audio_vae.py | 1258 | # Step 4: Final safety crop to ensure exact target shape | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 219 | # Step 1: Self-attention with adaptive layer norm (AdaLN) | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 230 | # Step 2: Cross-attention (if enabled) for conditioning on encoder outputs | COMMENT |
| LOW | diffsynth/models/ace_step_dit.py | 241 | # Step 3: Feed-forward (MLP) with adaptive layer norm | COMMENT |
| LOW⚡ | diffsynth/models/boogu_image_dit.py | 2296 | # Step 3: image self-attention. | COMMENT |
| LOW⚡ | diffsynth/models/boogu_image_dit.py | 2304 | # Step 4: residual updates. | COMMENT |
| LOW | diffsynth/models/boogu_image_dit.py | 2250 | # Step 1: modulation for both streams. | COMMENT |
| LOW | diffsynth/models/boogu_image_dit.py | 2267 | # Step 2: joint attention on [instruct + img]. | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| CRITICAL | diffsynth/models/qwen_image_21_text_encoder.py | 96 | self.model.model.language_model.norm.register_forward_hook(hook_fn) | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | diffsynth/models/nexus_gen_ar_model.py | 118 | Calculate the 3D rope index based on image and video's temporal, height and width in LLM. Explanation: | STRING |
| HIGH | diffsynth/models/anima_dit.py | 944 | Prepares an embedded sequence tensor by applying positional embeddings and handling padding masks. Arg | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | diffsynth/metrics/__init__.py | 17 | __all__ = [ | CODE |
| LOW⚡ | diffsynth/models/ltx2_dit.py | 1619 | def set_gradient_checkpointing(self, enable: bool) -> None: | CODE |
| LOW | diffsynth/models/flux2_vae.py | 726 | def set_use_npu_flash_attention(self, use_npu_flash_attention: bool) -> None: | STRING |
| LOW | diffsynth/models/flux2_vae.py | 889 | def set_attention_slice(self, slice_size: int) -> None: | STRING |
| LOW | diffsynth/models/flux2_vae.py | 917 | def set_processor(self, processor: "AttnProcessor") -> None: | STRING |
| LOW | diffsynth/models/minimax_h3_video_vae.py | 621 | __all__ = ["MiniMaxH3VideoVAE"] | CODE |
| LOW | diffsynth/models/ltx25_diffusion_video_vae.py | 3271 | __all__ = [ | CODE |
| LOW | diffsynth/models/nexus_gen_ar_model.py | 932 | __all__ = ["Qwen2_5_VLForConditionalGeneration", "Qwen2_5_VLModel", "Qwen2_5_VLPreTrainedModel"] | STRING |
| LOW | diffsynth/models/nexus_gen_ar_model.py | 1143 | __all__ = ["Qwen2_5_VLProcessor"] | STRING |
| LOW | diffsynth/models/boogu_image_dit.py | 180 | def set_processor(self, processor) -> None: | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | diffsynth/pipelines/joyai_image.py | 200 | CODE | |
| MEDIUM | diffsynth/models/z_image_dit.py | 991 | CODE | |
| MEDIUM | diffsynth/models/clip.py | 58 | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | examples/ltx2/model_inference/LTX-2.3-I2AV-TwoStage.py | 69 | # This example uses the first and last frames for demonstration. However, you can use any frames by setting input_images | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | diffsynth/pipelines/flux2_image.py | 442 | # Check if Qwen3 text encoder is available | COMMENT |
| LOW | diffsynth/models/ltx2_video_vae.py | 2082 | # Check if current temporal slice overlaps with previous temporal slice | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | diffsynth/models/flux2_vae.py | 975 | # here we simply pass along all tensors to the selected processor class | STRING |