huggingface/candle

2.7

Adjusted Score

2.7

Raw Score

100%

Time Factor

2026-07-06

Last Push

20.7K

Stars

Rust

Language

248.2K

Lines of Code

928

Files

499

Pattern Hits

2026-07-14

Scan Date

0.00

HC Hit Rate

What These Metrics Mean

Adjusted Score: Primary synthetic code indicator. Raw score normalised per 1,000 lines of code and multiplied by the temporal discount factor. This is the definitive comparative metric — use it to rank repositories by AI authorship density.
Raw Score: The unmodified sum of all severity-weighted, context-multiplied pattern match scores before temporal discounting. Reflects the absolute signal strength independent of when the repository was last active.
Time Factor: The temporal discount multiplier (0–100%) applied to the raw score. Repositories last updated before ChatGPT's launch (Nov 2022) receive a 5% factor. Full signal is only assigned to repositories active in the post-adoption era (Jan 2024+).
Pattern Hits: Total count of individual pattern matches across all files and categories. A high hit count with a low score may indicate a very large codebase with isolated AI snippets; a low count with a high score indicates dense, concentrated AI signatures.
HC Hit Rate: High+Critical pattern hits per file, averaged across the repository. This orthogonal signal catches repositories where a few files are densely packed with high-severity AI tells — a strong indicator even when the normalised score appears moderate due to codebase size.
Lines of Code / Files: Total lines and files analysed. The scanner examines 94 file extensions. These denominators are used to normalise the score, enabling fair comparison between repositories of vastly different sizes.

Score History

This chart maps the temporal evolution of the adjusted synthetic code score across successive scan runs. An upward trajectory indicates ongoing incorporation of AI-generated code or expanding LLM-assisted scaffolding; a stable or declining trajectory may reflect active human refactoring, code removal, or the adoption of stricter authorship policies. The dashed secondary line (right axis) independently tracks total raw pattern hit count, which can diverge from the normalised score when codebase size changes significantly between scans.

Severity Breakdown

Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.

CRITICAL 0HIGH 1MEDIUM 63LOW 435

Directory Score Breakdown

This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.

Pattern Findings

The scanner identified 499 distinct pattern matches across 15 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. 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 diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.

Over-Commented Block315 hits · 306 pts

Severity	File	Line	Snippet	Context
LOW	candle-transformers/src/utils.rs	1	//! Shared utilities: repeat_kv, repeat_penalty, causal mask.	COMMENT
LOW	candle-transformers/src/models/starcoder2.rs	1	//! StarCoder model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/quantized_mixformer.rs	1	//! Module containing quantized MixFormer model implementation.	COMMENT
LOW	candle-transformers/src/models/parler_tts.rs	1	//! Parler Model implementation for parler_tts text-to-speech synthesis	COMMENT
LOW	candle-transformers/src/models/eva2.rs	1	//! EVA-2 inference implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_moondream.rs	1	//! Implementation of a quantized Moondream vision language model.	COMMENT
LOW	candle-transformers/src/models/dinov2.rs	1	//! Implementation of the DINOv2 models from Meta Research.	COMMENT
LOW	candle-transformers/src/models/dinov2.rs	21	//!	COMMENT
LOW	candle-transformers/src/models/csm.rs	1	//! Implementation of the Conversational Speech Model (CSM) from Sesame	COMMENT
LOW	…e-transformers/src/models/quantized_recurrent_gemma.rs	1	//! Recurrent Gemma model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/quantized_phi.rs	1	//! Phi2 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/mobilenetv4.rs	1	//! # MobileNet-v4	COMMENT
LOW	candle-transformers/src/models/rwkv_v7.rs	1	//! RWKV v7 "Goose" (x070) model implementation.	COMMENT
LOW	candle-transformers/src/models/paligemma.rs	1	//! Multimodal multi-purpose model combining Gemma-based language model with SigLIP image understanding	COMMENT
LOW	candle-transformers/src/models/t5.rs	1	//! T5 model implementation.	COMMENT
LOW	candle-transformers/src/models/t5.rs	21	//! ```bash	COMMENT
LOW	candle-transformers/src/models/t5.rs	41	//! --prompt "<2de> How are you, my friend?" \	COMMENT
LOW	candle-transformers/src/models/phi3.rs	1	//! Microsoft Phi-3 model implementation	COMMENT
LOW	candle-transformers/src/models/bigcode.rs	1	//! BigCode implementation in Rust based on the GPT-BigCode model.	COMMENT
LOW	candle-transformers/src/models/rwkv_v6.rs	1	//! RWKV v6 model implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_blip.rs	1	//! BLIP model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/quantized_rwkv_v5.rs	1	//! RWKV v5 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/vgg.rs	1	//! VGG-16 model implementation.	COMMENT
LOW	candle-transformers/src/models/recurrent_gemma.rs	1	//! Recurrent Gemma model implementation	COMMENT
LOW	candle-transformers/src/models/quantized_mpt.rs	1	//! Quantized MPT model implementation.	COMMENT
LOW	candle-transformers/src/models/stable_lm.rs	1	//! StableLM model implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_stable_lm.rs	1	//! Module for quantized StableLM implementation.	COMMENT
LOW	candle-transformers/src/models/mixtral.rs	1	//! Mixtral Model, a sparse mixture of expert model based on the Mistral architecture	COMMENT
LOW	candle-transformers/src/models/rwkv_v5.rs	1	//! RWKV v5 model implementation.	COMMENT
LOW	candle-transformers/src/models/rwkv_v5.rs	21	//! ```bash	COMMENT
LOW	candle-transformers/src/models/olmo.rs	1	//! OLMo (Open Language Model) implementation	COMMENT
LOW	candle-transformers/src/models/qwen2.rs	1	//! Qwen2 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/mod.rs	1	//! Candle implementations for various deep learning models	COMMENT
LOW	candle-transformers/src/models/dinov2reg4.rs	1	//! Implementation of the DINOv2 revision (4 regularization)	COMMENT
LOW	candle-transformers/src/models/dinov2reg4.rs	21	//!	COMMENT
LOW	candle-transformers/src/models/quantized_rwkv_v6.rs	1	//! RWKV v6 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/phi.rs	1	//! Microsoft Phi model implementation	COMMENT
LOW	candle-transformers/src/models/convnext.rs	1	//! ConvNeXt implementation.	COMMENT
LOW	candle-transformers/src/models/segformer.rs	1	//! Segformer model implementation for semantic segmentation and image classification.	COMMENT
LOW	candle-transformers/src/models/quantized_gemma3.rs	1	//! Gemma 3 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/qwen2_moe.rs	1	//! Qwen2 model implementation with Mixture of Experts support.	COMMENT
LOW	candle-transformers/src/models/quantized_qwen2.rs	1	//! Qwen2 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/quantized_t5.rs	1	//! T5 model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/quantized_blip_text.rs	1	//! Quantized BLIP text module implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_mistral.rs	1	//! Mistral model implementation with quantization support.	COMMENT
LOW	candle-transformers/src/models/bert.rs	1	//! BERT (Bidirectional Encoder Representations from Transformers)	COMMENT
LOW	candle-transformers/src/models/efficientvit.rs	1	//! EfficientViT (MSRA) inference implementation based on timm.	COMMENT
LOW	candle-transformers/src/models/efficientvit.rs	21	//! --example efficientvit \	COMMENT
LOW	candle-transformers/src/models/quantized_llama2_c.rs	1	//! Quantized Llama2 model implementation.	COMMENT
LOW	candle-transformers/src/models/mobileclip.rs	1	//! Mobile CLIP model, combining a lightweight vision encoder with a text encoder	COMMENT
LOW	candle-transformers/src/models/quantized_llama.rs	1	//! Quantized llama model implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_llama.rs	521	span,	COMMENT
LOW	candle-transformers/src/models/quantized_llama.rs	621		COMMENT
LOW	candle-transformers/src/models/trocr.rs	1	//! TrOCR model implementation.	COMMENT
LOW	candle-transformers/src/models/nomic_bert.rs	1	//! # NomicBERT	COMMENT
LOW	candle-transformers/src/models/mpt.rs	1	//! Module implementing the MPT (Multi-Purpose Transformer) model	COMMENT
LOW	candle-transformers/src/models/yi.rs	1	//! Yi model implementation.	COMMENT
LOW	candle-transformers/src/models/quantized_metavoice.rs	1	//! Quantized MetaVoice model implementation.	COMMENT
LOW	candle-transformers/src/models/stella_en_v5.rs	1	//! Stella v5 model implementation.	COMMENT
LOW	candle-transformers/src/models/stella_en_v5.rs	501	// state = other_layernorm(xs)	COMMENT
255 more matches not shown…

Decorative Section Separators51 hits · 153 pts

Severity	File	Line	Snippet	Context
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	23	// ─── Config ──────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	77	// ─── State ───────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	155	// ─── Tokenizer ───────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	159	// ─── Helpers ─────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	191	// ─── TimeMix (Attention) ─────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	458	// ─── ChannelMix (FFN) ────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	549	// ─── DeaAttention (v7b only) ─────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	724	// ─── Block ───────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	831	// ─── Model ───────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/quantized_llama.rs	604	// ── Mask shape tests ──────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/quantized_llama.rs	622	// ── Mask value tests ──────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/quantized_llama.rs	672	// ── Mask broadcast compatibility test ─────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	5	// ── Text config defaults ────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	50	// ── Rope parameters ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	68	// ── Gemma4TextConfig ────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	147	// ── Vision config defaults ──────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	189	// ── Gemma4VisionConfig ──────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	236	// ── Audio config defaults ───────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	302	// ── Gemma4AudioConfig ───────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	374	// ── Top-level config defaults ───────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/config.rs	386	// ── Gemma4Config ────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	11	// ── RmsNorm (Gemma-style) ───────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	52	// ── 2D Vision Rotary Embedding ──────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	114	// ── PatchEmbedder ───────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	185	// ── VisionAttention ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	275	// ── VisionMlp ───────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	310	// ── VisionEncoderLayer ──────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	368	// ── VisionPooler ────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/vision.rs	437	// ── VisionTower (public) ────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/mod.rs	95	// ── Vision embedding injection ──────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/mod.rs	118	// ── Audio embedding injection ───────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	11	// ── RmsNorm (standard, no +1 offset for audio) ─────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	41	// ── LayerNorm (for SSCP conv blocks) ────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	67	// ── SSCP Conv Blocks ────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	170	// ── SubSampleConvProjection ─────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	225	// ── Relative Position Embedding ─────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	367	// ── Conformer Attention ─────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	645	// ── Conformer FeedForward ───────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	699	// ── Conformer LightConv1d ───────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	772	// ── ConformerBlock ──────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/audio.rs	808	// ── AudioModel (public) ─────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	12	// ── RmsNorm (Gemma-style with +1 offset) ────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	53	// ── RotaryEmbedding (standard, for sliding layers) ──────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	100	// ── ProportionalRotaryEmbedding (for global/full layers) ────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	153	// ── MLP ─────────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	192	// ── Flash attention ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	210	// ── KvCache ─────────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	218	// ── Attention ───────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	386	// ── DecoderLayer ────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	477	// ── Causal mask ─────────────────────────────────────────────────────────────	COMMENT
MEDIUM	candle-transformers/src/models/gemma4/text.rs	515	// ── TextModel ───────────────────────────────────────────────────────────────	COMMENT

Hyper-Verbose Identifiers42 hits · 46 pts

Severity	File	Line	Snippet	Context
LOW	candle-wasm-examples/quant-qwen3/serve.py	85	def find_or_download_tokenizer():	CODE
LOW	candle-pyo3/stub.py	127	def process_additional_function(fn):	CODE
LOW⚡	candle-pyo3/tests/native/test_tensor.py	8	def test_tensor_can_be_constructed():	CODE
LOW⚡	candle-pyo3/tests/native/test_tensor.py	13	def test_tensor_can_be_constructed_from_list():	CODE
LOW⚡	candle-pyo3/tests/native/test_tensor.py	18	def test_tensor_can_be_constructed_from_list_of_lists():	CODE
LOW⚡	candle-pyo3/tests/native/test_tensor.py	23	def test_tensor_can_be_quantized():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	43	def test_tensor_can_be_indexed():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	51	def test_tensor_can_be_sliced():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	87	def test_tensor_supports_equality_operations_with_scalars():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	109	def test_tensor_supports_equality_operations_with_tensors():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	133	def test_tensor_equality_operations_can_broadcast():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	145	def test_tensor_can_be_hashed():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	153	def test_tensor_can_be_expanded_with_none():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	168	def test_tensor_can_be_index_via_tensor():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	179	def test_tensor_can_be_index_via_list():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	190	def test_tensor_can_be_cast_via_to():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	212	def test_tensor_can_be_moved_via_to():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	234	def test_tensor_can_be_moved_and_cast_via_to():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	272	def test_tensor_can_be_subtracted():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	290	def test_tensor_can_be_multiplied():	CODE
LOW	candle-pyo3/tests/native/test_tensor.py	308	def test_tensor_can_be_divided():	CODE
LOW	candle-pyo3/tests/native/test_utils.py	10	def test_can_roundtrip_safetensors():	CODE
LOW	candle-pyo3/tests/native/test_shape.py	6	def test_absolute_shapes_are_valid():	CODE
LOW	candle-pyo3/tests/native/test_shape.py	17	def test_relative_shapes_are_valid():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	9	def test_module_can_be_constructed():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	18	def test_module_registers_tensors():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	30	def test_module_registers_submodules():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	45	def test_module_can_dump_statedict():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	61	def test_module_can_load_statedict():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	77	def test_module_throws_on_shape_mismatch():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	92	def test_module_throws_on_missing_key():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	108	def test_module_can_load_quantized_tensors():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	124	def test_module_dequantizes_tensors_automatically():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	139	def test_module_can_be_moved_to_cuda():	CODE
LOW	candle-pyo3/tests/bindings/test_module.py	151	def test_module_can_be_moved_from_cuda_to_cpu():	CODE
LOW⚡	candle-pyo3/tests/bindings/test_linear.py	6	def test_linear_layer_can_be_constructed():	CODE
LOW⚡	candle-pyo3/tests/bindings/test_linear.py	11	def test_linear_layer_can_forward_a_singular_input():	CODE
LOW⚡	candle-pyo3/tests/bindings/test_linear.py	18	def test_linear_layer_can_forward_a_batched_input():	CODE
LOW⚡	candle-pyo3/tests/bindings/test_linear.py	25	def test_quantized_linear_layer_can_forward_a_singular_input():	CODE
LOW⚡	candle-pyo3/tests/bindings/test_linear.py	33	def test_quantized_linear_layer_can_forward_a_batched_input():	CODE
LOW	candle-pyo3/tests/bindings/test_testing.py	8	def test_assert_equal_asserts_correctly(dtype: candle.DType):	CODE
LOW	candle-pyo3/tests/bindings/test_testing.py	18	def test_assert_almost_equal_asserts_correctly(dtype: candle.DType):	CODE

Verbosity Indicators18 hits · 38 pts

Severity	File	Line	Snippet	Context
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	766	// Step 1: Squeeze F dimension: (B, C, 1, H, W) -> (B, C, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	769	// Step 2: Reshape H into (H_tokens, pH): (B, C, H, W) -> (B, C, H_t, pH, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	772	// Step 3: Reshape W into (W_tokens, pW): (B, C, H_t, pH, W) -> (B, C, H_t, pH, W_t, pW)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	775	// Step 4: Permute to match Python: (C, H_t, pH, W_t, pW) -> (H_t, W_t, pH, pW, C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	780	// Step 5: Reshape to patches: (B, H_t, W_t, pH, pW, C) -> (B, H_tW_t, pHpW*C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	823	// Step 1: Reshape to (B, H_t, W_t, pH, pW, C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	826	// Step 2: Permute to match Python: (H_t, W_t, pH, pW, C) -> (C, H_t, pH, W_t, pW)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	831	// Step 3: Reshape to combine H and W: (B, C, H_t, pH, W_t, pW) -> (B, C, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	834	// Step 4: Add back F dimension: (B, C, H, W) -> (B, C, 1, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/paddleocr_vl/mod.rs	915	// Step 1: Get text embeddings	COMMENT
LOW⚡	candle-transformers/src/models/paddleocr_vl/mod.rs	923	// Step 2: Encode images	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	934	// Step 3: Merge vision embeddings into text embeddings	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	956	// Step 4: Compute M-RoPE position IDs	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	979	// Step 5: Forward through text model with export	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	1014	// Step 0: Prefill with image	COMMENT
LOW⚡	candle-core/tests/pth.py	20	# Step 1: Create a 3D NumPy array with Fortran order using a range of numbers	COMMENT
LOW⚡	candle-core/tests/pth.py	28	# Step 2: Convert the NumPy array to a PyTorch tensor	COMMENT
LOW⚡	candle-core/tests/pth.py	34	# Step 3: Save the PyTorch tensor to a .pth file	COMMENT

Structural Annotation Overuse18 hits · 38 pts

Severity	File	Line	Snippet	Context
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	766	// Step 1: Squeeze F dimension: (B, C, 1, H, W) -> (B, C, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	769	// Step 2: Reshape H into (H_tokens, pH): (B, C, H, W) -> (B, C, H_t, pH, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	772	// Step 3: Reshape W into (W_tokens, pW): (B, C, H_t, pH, W) -> (B, C, H_t, pH, W_t, pW)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	775	// Step 4: Permute to match Python: (C, H_t, pH, W_t, pW) -> (H_t, W_t, pH, pW, C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	780	// Step 5: Reshape to patches: (B, H_t, W_t, pH, pW, C) -> (B, H_tW_t, pHpW*C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	823	// Step 1: Reshape to (B, H_t, W_t, pH, pW, C)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	826	// Step 2: Permute to match Python: (H_t, W_t, pH, pW, C) -> (C, H_t, pH, W_t, pW)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	831	// Step 3: Reshape to combine H and W: (B, C, H_t, pH, W_t, pW) -> (B, C, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/z_image/transformer.rs	834	// Step 4: Add back F dimension: (B, C, H, W) -> (B, C, 1, H, W)	COMMENT
LOW⚡	candle-transformers/src/models/paddleocr_vl/mod.rs	915	// Step 1: Get text embeddings	COMMENT
LOW⚡	candle-transformers/src/models/paddleocr_vl/mod.rs	923	// Step 2: Encode images	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	934	// Step 3: Merge vision embeddings into text embeddings	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	956	// Step 4: Compute M-RoPE position IDs	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	979	// Step 5: Forward through text model with export	COMMENT
LOW	candle-transformers/src/models/paddleocr_vl/mod.rs	1014	// Step 0: Prefill with image	COMMENT
LOW⚡	candle-core/tests/pth.py	20	# Step 1: Create a 3D NumPy array with Fortran order using a range of numbers	COMMENT
LOW⚡	candle-core/tests/pth.py	28	# Step 2: Convert the NumPy array to a PyTorch tensor	COMMENT
LOW⚡	candle-core/tests/pth.py	34	# Step 3: Save the PyTorch tensor to a .pth file	COMMENT

Unused Imports26 hits · 26 pts

Severity	File	Line	Context
LOW	candle-wasm-examples/quant-qwen3/serve.py	10	CODE
LOW	candle-pyo3/e5.py	8	CODE
LOW	candle-pyo3/py_src/candle/__init__.py	4	CODE
LOW	candle-pyo3/py_src/candle/__init__.py	49	CODE
LOW	candle-pyo3/py_src/candle/nn/linear.py	1	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	1	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	3	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	4	CODE
LOW	candle-pyo3/py_src/candle/nn/__init__.py	5	CODE
LOW	candle-pyo3/py_src/candle/nn/container.py	3	CODE
LOW	candle-pyo3/py_src/candle/nn/sparse.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/sparse.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/sparse.py	2	CODE
LOW	candle-pyo3/py_src/candle/nn/normalization.py	4	CODE
LOW	candle-pyo3/py_src/candle/nn/normalization.py	4	CODE
LOW	candle-pyo3/py_src/candle/models/llama.py	3	CODE
LOW	candle-pyo3/tests/native/test_utils.py	2	CODE
LOW	candle-pyo3/tests/native/test_utils.py	2	CODE
LOW	candle-pyo3/tests/native/test_shape.py	1	CODE
LOW	candle-pyo3/tests/bindings/test_linear.py	2	CODE
LOW	candle-examples/examples/flux/t5_tokenizer.py	1	CODE
LOW	candle-examples/examples/resnet/export_models.py	2	CODE
LOW	candle-examples/examples/resnet/export_models.py	3	CODE

Excessive Try-Catch Wrapping8 hits · 14 pts

Severity	File	Line	Snippet	Context
MEDIUM	candle-wasm-examples/quant-qwen3/serve.py	12	print("Error: Required packages not installed", file=sys.stderr)	CODE
LOW⚡	candle-wasm-examples/quant-qwen3/serve.py	50	except Exception as e:	CODE
MEDIUM⚡	candle-wasm-examples/quant-qwen3/serve.py	51	print(f"Error downloading {filename}: {e}", file=sys.stderr)	CODE
MEDIUM⚡	candle-wasm-examples/quant-qwen3/serve.py	60	print(f"Error: Custom path does not exist: {custom_path}", file=sys.stderr)	CODE
LOW	candle-wasm-examples/quant-qwen3/serve.py	142	except Exception as e:	CODE
MEDIUM	candle-wasm-examples/quant-qwen3/serve.py	133	def send_file(self, filepath, content_type):	CODE
LOW	candle-pyo3/py_src/candle/nn/module.py	462	except Exception as ex:	STRING
LOW	…es/examples/marian-mt/python/convert_slow_tokenizer.py	43	except Exception:	CODE

AI Slop Vocabulary4 hits · 12 pts

Severity	File	Line	Snippet	Context
MEDIUM	candle-transformers/src/models/rwkv_v7.rs	66	/// This is more robust than computing from a formula, as different	COMMENT
MEDIUM	candle-transformers/src/models/rwkv_v5.rs	72	// Certainly a weird convention but taken from modeling_rwkv5.py	COMMENT
MEDIUM	candle-transformers/src/models/segment_anything/mod.rs	4	//! in an image based on prompts like points or boxes. //! This model provides a robust and fast image segmentation pipe	COMMENT
MEDIUM	candle-nn/src/loss.rs	78	/// A robust loss function that combines `MAE` and `MSE` losses:	COMMENT

Self-Referential Comments3 hits · 10 pts

Severity	File	Line	Snippet	Context
MEDIUM	candle-pyo3/tests/native/test_tensor.py	134	# Create a decoder attention mask as a test case	COMMENT
MEDIUM⚡	candle-core/tests/pth.py	17	# Create a tensor with fortran contiguous memory layout	COMMENT
MEDIUM	…ples/examples/reinforcement-learning/atari_wrappers.py	303	# Create the environment.	COMMENT

Deep Nesting8 hits · 8 pts

Severity	File	Line	Context
LOW	candle-pyo3/stub.py	87	CODE
LOW	candle-pyo3/py_src/candle/nn/container.py	136	CODE
LOW	candle-pyo3/py_src/candle/nn/container.py	153	CODE
LOW	candle-pyo3/py_src/candle/nn/module.py	97	CODE
LOW	candle-pyo3/py_src/candle/nn/module.py	391	CODE
LOW	candle-pyo3/py_src/candle/nn/module.py	592	CODE
LOW	candle-pyo3/py_src/candle/nn/module.py	659	CODE
LOW	…ples/examples/reinforcement-learning/atari_wrappers.py	232	CODE

Synthetic Comment Markers1 hit · 8 pts

Severity	File	Line	Snippet	Context
HIGH	candle-examples/examples/quantized-lfm2/main.rs	41	/// GGUF file to load, typically a .gguf file generated by llama.cpp.	COMMENT

Modern AI Meta-Vocabulary1 hit · 3 pts

Severity	File	Line	Snippet	Context
MEDIUM	candle-examples/examples/stella-en-v5/README.md	1	# candle-stella-en-v5: Implementation of [stella_en_1.5B_v5](https://huggingface.co/dunzhang/stella_en_1.5B_v5) embeddin	COMMENT

Fake / Example Data2 hits · 2 pts

Severity	File	Line	Snippet	Context
LOW	candle-transformers/src/models/voxtral/model.rs	901	let dummy_token = Tensor::new(&[1u32], &config.device)?;	CODE
LOW	candle-transformers/src/models/voxtral/model.rs	902	let dummy_embed = self.language_model.embed(&dummy_token)?;	CODE

Modern Structural Boilerplate1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	candle-pyo3/py_src/candle/nn/container.py	19	__all__ = ["Sequential", "ModuleList", "ModuleDict"]	CODE

Redundant / Tautological Comments1 hit · 0 pts

Severity	File	Line	Snippet	Context
LOW	candle-pyo3/stub.py	37	# Check if we have a return type annotation in the docstring	STRING

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings