ggml-org/whisper.cpp

2.5

Adjusted Score

2.5

Raw Score

100%

Time Factor

2026-07-11

Last Push

51.8K

Stars

C++

Language

472.4K

Lines of Code

1.1K

Files

1.1K

Pattern Hits

2026-07-14

Scan Date

0.01

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 1HIGH 15MEDIUM 31LOW 1017

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 1064 distinct pattern matches across 19 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 Block857 hits · 762 pts

Severity	File	Line	Snippet	Context
LOW	CMakeLists.txt	141	endif()	COMMENT
LOW	ci/run.sh	1	#!/bin/bash	COMMENT
LOW	include/whisper.h	1	#ifndef WHISPER_H	COMMENT
LOW	include/whisper.h	21	# ifdef WHISPER_BUILD	COMMENT
LOW	include/whisper.h	41		COMMENT
LOW	include/whisper.h	61	// }	COMMENT
LOW	include/whisper.h	241	WHISPER_API struct whisper_state * whisper_init_state(struct whisper_context * ctx);	COMMENT
LOW	include/whisper.h	361	// Examples:	COMMENT
LOW	include/parakeet.h	1	#ifndef PARAKEET_H	COMMENT
LOW	include/parakeet.h	21	# ifdef PARAKEET_BUILD	COMMENT
LOW	tests/run-tests.sh	1	#!/bin/bash	COMMENT
LOW	tests/parakeet-verification.h	1	#pragma once	COMMENT
LOW	tests/test-vad-full.cpp	1	#include "whisper.h"	COMMENT
LOW	models/convert-parakeet-to-ggml.py	1	#!/usr/bin/env python3	COMMENT
LOW	models/convert-h5-to-ggml.py	1	# Convert Hugging Face fine-tuned models to ggml format	COMMENT
LOW	models/convert-pt-to-ggml.py	1	# Convert Whisper transformer model from PyTorch to ggml format	COMMENT
LOW	models/convert-pt-to-ggml.py	21	# - mel filters	COMMENT
LOW	models/convert-pt-to-ggml.py	41	import numpy as np	COMMENT
LOW	models/convert-pt-to-ggml.py	61	# "nl": "dutch",	COMMENT
LOW	models/convert-pt-to-ggml.py	81	# "hr": "croatian",	COMMENT
LOW	models/convert-pt-to-ggml.py	101	# "is": "icelandic",	COMMENT
LOW	models/convert-pt-to-ggml.py	121	# "tg": "tajik",	COMMENT
LOW	models/convert-pt-to-ggml.py	141	# "tt": "tatar",	COMMENT
LOW	examples/ffmpeg-transcode.cpp	1	#ifdef WHISPER_COMMON_FFMPEG	COMMENT
LOW	examples/grammar-parser.h	1	// Implements a parser for an extended Backus-Naur form (BNF), producing the	COMMENT
LOW	examples/json.hpp	21	#include <algorithm> // all_of, find, for_each	COMMENT
LOW	examples/json.hpp	41	// SPDX-FileCopyrightText: 2013-2022 Niels Lohmann <https://nlohmann.me>	COMMENT
LOW	examples/json.hpp	61	#ifndef JSON_SKIP_LIBRARY_VERSION_CHECK	COMMENT
LOW	examples/json.hpp	81	#if JSON_DIAGNOSTICS	COMMENT
LOW	examples/json.hpp	141	{	COMMENT
LOW	examples/json.hpp	161	#include <algorithm> // transform	COMMENT
LOW	examples/json.hpp	181		COMMENT
LOW	examples/json.hpp	201	#include <array> // array	COMMENT
LOW	examples/json.hpp	221	// \| \| \|__ \| \| \| \| \| \| version 3.11.2	COMMENT
LOW	examples/json.hpp	321	// \| \| \|__ \| \| \| \| \| \| version 3.11.2	COMMENT
LOW	examples/json.hpp	341	#define JSON_HEDLEY_STRINGIFY_EX(x) #x	COMMENT
LOW	examples/json.hpp	361	#define JSON_HEDLEY_CONCAT3_EX(a,b,c) a##b##c	COMMENT
LOW	examples/json.hpp	381	#define JSON_HEDLEY_VERSION_DECODE_MINOR(version) (((version) % 1000000) / 1000)	COMMENT
LOW	examples/json.hpp	401	#define JSON_HEDLEY_GNUC_VERSION_CHECK(major,minor,patch) (JSON_HEDLEY_GNUC_VERSION >= JSON_HEDLEY_VERSION_ENCODE(ma	COMMENT
LOW	examples/json.hpp	421	#define JSON_HEDLEY_MSVC_VERSION_CHECK(major,minor,patch) (0)	COMMENT
LOW	examples/json.hpp	441	#endif	COMMENT
LOW	examples/json.hpp	461	#define JSON_HEDLEY_INTEL_CL_VERSION_CHECK(major,minor,patch) (0)	COMMENT
LOW	examples/json.hpp	481	#undef JSON_HEDLEY_SUNPRO_VERSION	COMMENT
LOW	examples/json.hpp	501		COMMENT
LOW	examples/json.hpp	521	#if defined(__CC_ARM) && defined(__ARMCOMPILER_VERSION)	COMMENT
LOW	examples/json.hpp	541	#elif defined(__xlC__) && defined(__xlC_ver__)	COMMENT
LOW	examples/json.hpp	561	( \	COMMENT
LOW	examples/json.hpp	581	#undef JSON_HEDLEY_TI_CL2000_VERSION	COMMENT
LOW	examples/json.hpp	601	#endif	COMMENT
LOW	examples/json.hpp	621	#endif	COMMENT
LOW	examples/json.hpp	641	#define JSON_HEDLEY_TI_CL6X_VERSION_CHECK(major,minor,patch) (0)	COMMENT
LOW	examples/json.hpp	661	#undef JSON_HEDLEY_TI_CLPRU_VERSION	COMMENT
LOW	examples/json.hpp	681	#define JSON_HEDLEY_CRAY_VERSION JSON_HEDLEY_VERSION_ENCODE(_RELEASE_MAJOR, _RELEASE_MINOR, _RELEASE_PATCHLEVEL)	COMMENT
LOW	examples/json.hpp	701	#define JSON_HEDLEY_IAR_VERSION JSON_HEDLEY_VERSION_ENCODE((__VER__ / 1000000), ((__VER__ / 1000) % 1000), (__VE	COMMENT
LOW	examples/json.hpp	721	#endif	COMMENT
LOW	examples/json.hpp	741	#endif	COMMENT
LOW	examples/json.hpp	761	#define JSON_HEDLEY_COMPCERT_VERSION_CHECK(major,minor,patch) (0)	COMMENT
LOW	examples/json.hpp	781	#undef JSON_HEDLEY_MCST_LCC_VERSION	COMMENT
LOW	examples/json.hpp	821	#if defined(JSON_HEDLEY_GCC_VERSION)	COMMENT
LOW	examples/json.hpp	841	#undef JSON_HEDLEY_GNUC_HAS_ATTRIBUTE	COMMENT
797 more matches not shown…

Redundant / Tautological Comments34 hits · 66 pts

Severity	File	Line	Snippet	Context
LOW	models/convert-silero-vad-to-ggml.py	146	# Check if this is a convolution weight tensor	COMMENT
LOW	models/download-ggml-model.sh	27	# Check if the script is inside a /bin/ directory	COMMENT
LOW	models/download-ggml-model.sh	143	# Check if 'whisper-cli' is available in the system PATH	COMMENT
LOW	models/download-coreml-model.sh	24	# Check if the script is inside a /bin/ directory	COMMENT
LOW	models/download-coreml-model.sh	88	# Check if 'whisper-cli' is available in the system PATH	COMMENT
LOW	models/download-vad-model.sh	24	# Check if the script is inside a /bin/ directory	COMMENT
LOW	models/download-vad-model.sh	93	# Check if 'whisper-cli' is available in the system PATH	COMMENT
LOW	examples/server.py	60	# Check if this is a request to a file in whisper.wasm	COMMENT
LOW	examples/python/whisper_processor.py	17	# Check if the file exists	COMMENT
LOW	scripts/bench.py	128	# Check if the sample file exists	COMMENT
LOW	scripts/apple/validate-macos.sh	685	# Check if authentication credentials are provided	COMMENT
LOW⚡	scripts/apple/validate-macos.sh	704	# Check if app was created successfully	COMMENT
LOW⚡	scripts/apple/validate-macos.sh	712	# Check if app binary exists and is executable	COMMENT
LOW⚡	scripts/apple/validate-macos.sh	720	# Check if framework was properly embedded	COMMENT
LOW⚡	scripts/apple/validate-macos.sh	728	# Check if framework binary exists	COMMENT
LOW	scripts/apple/validate-visionos.sh	679	# Check if authentication credentials are provided	COMMENT
LOW⚡	scripts/apple/validate-visionos.sh	704	# Check if validation failed because the app isn't in App Store Connect	COMMENT
LOW⚡	scripts/apple/validate-visionos.sh	713	# Check if IPA was created successfully	COMMENT
LOW⚡	scripts/apple/validate-visionos.sh	721	# Check if app binary exists and is executable	COMMENT
LOW⚡	scripts/apple/validate-visionos.sh	729	# Check if framework was properly embedded	COMMENT
LOW⚡	scripts/apple/validate-visionos.sh	737	# Check if framework binary exists	COMMENT
LOW	scripts/apple/validate-ios.sh	688	# Check if authentication credentials are provided	COMMENT
LOW⚡	scripts/apple/validate-ios.sh	713	# Check if validation failed because the app isn't in App Store Connect	COMMENT
LOW⚡	scripts/apple/validate-ios.sh	722	# Check if IPA was created successfully	COMMENT
LOW⚡	scripts/apple/validate-ios.sh	730	# Check if app binary exists and is executable	COMMENT
LOW⚡	scripts/apple/validate-ios.sh	738	# Check if framework was properly embedded	COMMENT
LOW⚡	scripts/apple/validate-ios.sh	746	# Check if framework binary exists	COMMENT
LOW	scripts/apple/validate-tvos.sh	681	# Check if authentication credentials are provided	COMMENT
LOW⚡	scripts/apple/validate-tvos.sh	706	# Check if validation failed because the app isn't in App Store Connect	COMMENT
LOW⚡	scripts/apple/validate-tvos.sh	715	# Check if IPA was created successfully	COMMENT
LOW⚡	scripts/apple/validate-tvos.sh	723	# Check if app binary exists and is executable	COMMENT
LOW⚡	scripts/apple/validate-tvos.sh	731	# Check if framework was properly embedded	COMMENT
LOW⚡	scripts/apple/validate-tvos.sh	739	# Check if framework binary exists	COMMENT
LOW⚡	ggml/src/ggml-virtgpu/regenerate_remoting.py	46	# Check if clang-format is available	COMMENT

Cross-File Repetition12 hits · 60 pts

Severity	File	Snippet	Context
HIGH	tests/earnings21/normalizers/basic.py	replace any other markers, symbols, and punctuations with a space, and drop any diacritics (category 'mn' and some manua	STRING
HIGH	tests/librispeech/normalizers/basic.py	replace any other markers, symbols, and punctuations with a space, and drop any diacritics (category 'mn' and some manua	STRING
HIGH	tests/librispeech-parakeet/normalizers/basic.py	replace any other markers, symbols, and punctuations with a space, and drop any diacritics (category 'mn' and some manua	STRING
HIGH	tests/earnings21/normalizers/basic.py	replace any other markers, symbols, punctuations with a space, keeping diacritics	STRING
HIGH	tests/librispeech/normalizers/basic.py	replace any other markers, symbols, punctuations with a space, keeping diacritics	STRING
HIGH	tests/librispeech-parakeet/normalizers/basic.py	replace any other markers, symbols, punctuations with a space, keeping diacritics	STRING
HIGH	tests/earnings21/normalizers/english.py	convert any spelled-out numbers into arabic numbers, while handling: - remove any commas - keep the suffixes such as: `1	STRING
HIGH	tests/librispeech/normalizers/english.py	convert any spelled-out numbers into arabic numbers, while handling: - remove any commas - keep the suffixes such as: `1	STRING
HIGH	tests/librispeech-parakeet/normalizers/english.py	convert any spelled-out numbers into arabic numbers, while handling: - remove any commas - keep the suffixes such as: `1	STRING
HIGH	tests/earnings21/normalizers/english.py	applies british-american spelling mappings as listed in [1]. [1] https://www.tysto.com/uk-us-spelling-list.html	STRING
HIGH	tests/librispeech/normalizers/english.py	applies british-american spelling mappings as listed in [1]. [1] https://www.tysto.com/uk-us-spelling-list.html	STRING
HIGH	tests/librispeech-parakeet/normalizers/english.py	applies british-american spelling mappings as listed in [1]. [1] https://www.tysto.com/uk-us-spelling-list.html	STRING

Structural Annotation Overuse31 hits · 55 pts

Severity	File	Line	Snippet	Context
LOW	examples/talk-llama/llama.h	1185	/// NOTE: This function does not use a jinja parser. It only support a pre-defined list of template. See more: https	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	2945	// Step 1: Look for restart sequences to limit the maximum repetition length.	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3006	// Step 2: Iterate in reverse over the last N tokens of the context, using the "Z-algorithm" (in	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3070	// Step 3: Iterate over dry_repeat_count and last_tokens, examining the maximum repeat length	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3097	// Step 4: Apply logit penalties based on the maximum repeat length for relevant tokens.	COMMENT
LOW	examples/talk-llama/models/kimi-linear.cpp	197	// Step 1: Q, K, V projections -> [d_inner, n_tokens]	COMMENT
LOW⚡	examples/talk-llama/models/kimi-linear.cpp	377	// Step 1: Q projection and reshape	COMMENT
LOW⚡	examples/talk-llama/models/kimi-linear.cpp	382	// Step 2: KV compression	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	198	// Step 1: Apply decay gate to state	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	203	// Step 2: delta = (v - S @ k) * beta	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	210	// Step 3: state += outer(delta, k)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	216	// Step 4: attn_out = S @ q * scale	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	25	// Step 1: Calculate mean along the last dimension	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	30	// Step 2: Calculate (input - mean)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	33	// Step 3: Calculate squared differences (input - mean)^2	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	37	// Step 4: Calculate variance = mean((input - mean)^2)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	41	// Step 5: Get epsilon from op_params	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	45	// Step 6: Calculate std = sqrt(variance + eps)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	49	// Step 7: Normalize: output = (input - mean) / std	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	2809	// Step 5: multiply by attn_factor	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	2826	// Step 6: repeat	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3090	// Step 1: Prepare trans tensors for F16 type conversion to F32 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3115	// Step 2: Prepare head tensors for tail splitting if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3144	// Step 3: Execute RotaryPositionEmbedding	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3191	// Step 4: Copy unrotated tail portion from source to destination	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3220	// Step 5: Cast back to F16 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3906	// Step 1: cast the src0 (Query) to fp16 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3929	// Step 2: create the acl tensors for src1 (Key), src2 (Value),	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3973	// Step 3: create the PSEShift tensor if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	4042	// Step 4: set the inputs for FusedInferAttention.	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	4103	// Step 6: post-processing — slice padded output and/or cast to f32	COMMENT

Verbosity Indicators31 hits · 55 pts

Severity	File	Line	Snippet	Context
LOW	examples/talk-llama/llama-sampler.cpp	2945	// Step 1: Look for restart sequences to limit the maximum repetition length.	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3006	// Step 2: Iterate in reverse over the last N tokens of the context, using the "Z-algorithm" (in	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3070	// Step 3: Iterate over dry_repeat_count and last_tokens, examining the maximum repeat length	COMMENT
LOW	examples/talk-llama/llama-sampler.cpp	3097	// Step 4: Apply logit penalties based on the maximum repeat length for relevant tokens.	COMMENT
LOW	examples/talk-llama/models/kimi-linear.cpp	197	// Step 1: Q, K, V projections -> [d_inner, n_tokens]	COMMENT
LOW⚡	examples/talk-llama/models/kimi-linear.cpp	377	// Step 1: Q projection and reshape	COMMENT
LOW⚡	examples/talk-llama/models/kimi-linear.cpp	382	// Step 2: KV compression	COMMENT
LOW	ggml/src/ggml-webgpu/ggml-webgpu.cpp	1291	// For set rows specifically, we need to check if src and idx are empty	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	198	// Step 1: Apply decay gate to state	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	203	// Step 2: delta = (v - S @ k) * beta	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	210	// Step 3: state += outer(delta, k)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp	216	// Step 4: attn_out = S @ q * scale	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	25	// Step 1: Calculate mean along the last dimension	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	30	// Step 2: Calculate (input - mean)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	33	// Step 3: Calculate squared differences (input - mean)^2	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	37	// Step 4: Calculate variance = mean((input - mean)^2)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	41	// Step 5: Get epsilon from op_params	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	45	// Step 6: Calculate std = sqrt(variance + eps)	COMMENT
LOW⚡	ggml/src/ggml-openvino/openvino/op/norm.cpp	49	// Step 7: Normalize: output = (input - mean) / std	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	2809	// Step 5: multiply by attn_factor	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	2826	// Step 6: repeat	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3090	// Step 1: Prepare trans tensors for F16 type conversion to F32 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3115	// Step 2: Prepare head tensors for tail splitting if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3144	// Step 3: Execute RotaryPositionEmbedding	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3191	// Step 4: Copy unrotated tail portion from source to destination	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3220	// Step 5: Cast back to F16 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3906	// Step 1: cast the src0 (Query) to fp16 if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3929	// Step 2: create the acl tensors for src1 (Key), src2 (Value),	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	3973	// Step 3: create the PSEShift tensor if needed	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	4042	// Step 4: set the inputs for FusedInferAttention.	COMMENT
LOW	ggml/src/ggml-cann/aclnn_ops.cpp	4103	// Step 6: post-processing — slice padded output and/or cast to f32	COMMENT

Self-Referential Comments11 hits · 33 pts

Severity	File	Line	Snippet	Context
MEDIUM	build-xcframework.sh	387	# Create a separate directory for dSYMs for all platforms	COMMENT
MEDIUM	build-xcframework.sh	395	# Create a copy of the binary that will be stripped	COMMENT
MEDIUM	scripts/bench.py	18	# Define the argument to accept a list	COMMENT
MEDIUM	scripts/bench.py	54	# Define the models, threads, and processor counts to benchmark	COMMENT
MEDIUM	scripts/apple/validate-macos.sh	571	# Create a simple xcscheme file to avoid xcodebuild scheme issues	COMMENT
MEDIUM	scripts/apple/validate-macos.sh	667	# Create a zip file for potential distribution	COMMENT
MEDIUM	scripts/apple/validate-visionos.sh	565	# Create a simple xcscheme file to avoid xcodebuild scheme issues	COMMENT
MEDIUM	scripts/apple/validate-ios.sh	574	# Create a simple xcscheme file to avoid xcodebuild scheme issues	COMMENT
MEDIUM	scripts/apple/validate-tvos.sh	567	# Create a simple xcscheme file to avoid xcodebuild scheme issues	COMMENT
MEDIUM	.github/workflows/release.yml	367	# Create the directory where the CUDA Toolkit will be installed	COMMENT
MEDIUM	.github/workflows/release.yml	423	# Create the directory where the CUDA Toolkit will be installed	COMMENT

Excessive Try-Catch Wrapping15 hits · 24 pts

Severity	File	Line	Snippet	Context
MEDIUM	models/convert-parakeet-to-ggml.py	333	print(f"Error: {args.model} not found")	CODE
LOW	models/convert-pt-to-ggml.py	206	except Exception:	CODE
MEDIUM	models/convert-pt-to-ggml.py	207	print("Error: failed to load PyTorch model file:" , fname_inp)	CODE
MEDIUM	models/convert-pt-to-ggml.py	238	print("Error: failed to find either tiktoken or hf_transformers tokenizer file:", tokenizer)	CODE
MEDIUM	examples/server.py	119	print(f"Error: {e}")	CODE
LOW	examples/python/test_whisper_processor.py	6	except Exception as e:	CODE
MEDIUM	examples/python/test_whisper_processor.py	7	print(f"Error: {e}")	CODE
LOW	examples/python/whisper_processor.py	48	except Exception as e:	CODE
MEDIUM	examples/python/whisper_processor.py	49	print(f"Error: {e}")	CODE
MEDIUM	…r.swiftui/whisper.swiftui.demo/UI/DownloadButton.swift	32	print("Error: \(error.localizedDescription)")	CODE
MEDIUM	…r.swiftui/whisper.swiftui.demo/UI/DownloadButton.swift	48	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…r.swiftui/whisper.swiftui.demo/UI/DownloadButton.swift	90	print("Error deleting file: \(error)")	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	69	except Exception as e:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	327	except Exception as e:	STRING
MEDIUM	ggml/src/ggml-virtgpu/regenerate_remoting.py	323	def main():	CODE

Hyper-Verbose Identifiers21 hits · 22 pts

Severity	File	Line	Snippet	Context
LOW	tests/earnings21/normalizers/basic.py	27	def remove_symbols_and_diacritics(s: str, keep=""):	CODE
LOW	tests/librispeech/normalizers/basic.py	27	def remove_symbols_and_diacritics(s: str, keep=""):	CODE
LOW	tests/librispeech-parakeet/normalizers/basic.py	27	def remove_symbols_and_diacritics(s: str, keep=""):	CODE
LOW	models/convert-whisper-to-coreml.py	47	def correct_for_bias_scale_order_inversion(state_dict, prefix, local_metadata,	CODE
LOW	examples/server.py	15	class CustomHTTPRequestHandler(http.server.SimpleHTTPRequestHandler):	CODE
LOW	…/wchess.wasm/chessboardjs-1.0.0/js/chessboard-1.0.0.js	496	function calculatePositionFromMoves (position, moves) {	CODE
LOW	…/wchess.wasm/chessboardjs-1.0.0/js/chessboard-1.0.0.js	541	function expandConfigArgumentShorthand (config) {	CODE
LOW	…/wchess.wasm/chessboardjs-1.0.0/js/chessboard-1.0.0.js	951	function animateSparePieceToSquare (piece, dest, completeFn) {	CODE
LOW	bindings/javascript/whisper.js	8	function GROWABLE_HEAP_I8(){if(wasmMemory.buffer!=HEAP8.buffer){updateMemoryViews()}return HEAP8}function GROWABLE_HEAP_	CODE
LOW	bindings/ruby/test/test_context_params.rb	60	def test_new_with_kw_args_default_values(param)	CODE
LOW	bindings/ruby/test/test_vad_params.rb	83	def test_new_with_kw_args_default_values(param)	CODE
LOW	bindings/ruby/test/test_params.rb	278	def test_new_with_kw_args_default_values(param)	CODE
LOW	bindings/ruby/test/test_parakeet_params.rb	63	def test_new_with_kw_args_default_values(param)	CODE
LOW	ggml/include/ggml-backend.h	211	typedef ggml_backend_buffer_type_t (ggml_backend_split_buffer_type_t)(int main_device, const float tensor_split	CODE
LOW⚡	ggml/src/ggml-virtgpu/regenerate_remoting.py	49	def _check_clang_format_available(self) -> bool:	CODE
LOW⚡	ggml/src/ggml-virtgpu/regenerate_remoting.py	53	def _format_file_with_clang_format(self, file_path: Path) -> bool:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	78	def generate_backend_function_name(self, group_name: str, function_name: str) -> str:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	89	def generate_frontend_function_name(self, group_name: str, function_name: str) -> str:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	125	def generate_apir_backend_header(self) -> str:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	176	def generate_backend_dispatched_header(self) -> str:	CODE
LOW	ggml/src/ggml-virtgpu/regenerate_remoting.py	222	def generate_virtgpu_forward_header(self) -> str:	STRING

AI Slop Vocabulary6 hits · 18 pts

Severity	File	Line	Snippet	Context
MEDIUM	examples/stb_vorbis.c	51	// 1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory	COMMENT
MEDIUM	ggml/src/ggml-zendnn/ggml-zendnn.cpp	600	// gather + batched matmul + scatter approach. Future versions will leverage	COMMENT
MEDIUM	ggml/src/ggml-cpu/amx/mmq.cpp	603	// pack again with 128 to fully utilize vector length	COMMENT
MEDIUM	ggml/src/ggml-cpu/amx/mmq.cpp	645	// pack again with 128 to fully utilize vector length	COMMENT
MEDIUM	ggml/src/ggml-cpu/amx/mmq.cpp	743	// pack again with 128 to fully utilize vector length	COMMENT
MEDIUM	src/whisper.cpp	8473	// guaranteed. Might get removed at some point unless a robust algorithm implementation is found	COMMENT

Unused Imports16 hits · 16 pts

Severity	File	Line	Context
LOW	tests/earnings21/normalizers/__init__.py	1	CODE
LOW	tests/earnings21/normalizers/__init__.py	2	CODE
LOW	tests/librispeech/normalizers/__init__.py	1	CODE
LOW	tests/librispeech/normalizers/__init__.py	2	CODE
LOW	tests/librispeech-parakeet/normalizers/__init__.py	1	CODE
LOW	tests/librispeech-parakeet/normalizers/__init__.py	2	CODE
LOW	models/convert-parakeet-to-ggml.py	17	CODE
LOW	models/convert-parakeet-to-ggml.py	23	CODE
LOW	models/convert-parakeet-to-ggml.py	27	CODE
LOW	models/convert-h5-to-ggml.py	18	CODE
LOW	models/convert-h5-to-ggml.py	23	CODE
LOW	models/convert-silero-vad-to-ggml.py	4	CODE
LOW	models/convert-pt-to-ggml.py	35	CODE
LOW	models/convert-pt-to-ggml.py	39	CODE
LOW	models/convert-h5-to-coreml.py	13	CODE
LOW	examples/server.py	6	CODE

Synthetic Comment Markers3 hits · 15 pts

Severity	File	Line	Snippet	Context
HIGH	CONTRIBUTING.md	18	Code that is initially generated by AI and subsequently edited will still be considered AI-generated. AI assistance is p	CODE
HIGH	ggml/src/ggml-virtgpu/regenerate_remoting.py	3	# Generated by Claude AI	STRING
HIGH	ggml/src/ggml-hexagon/ggml-hexagon.cpp	2054	// Check for F16 support only as requested	COMMENT

Decorative Section Separators4 hits · 12 pts

Severity	File	Line	Snippet	Context
MEDIUM	examples/addon.node/CMakeLists.txt	4	#==================================================================	COMMENT
MEDIUM	examples/addon.node/CMakeLists.txt	8	#==================================================================	COMMENT
MEDIUM	examples/addon.node/CMakeLists.txt	19	#==================================================================	COMMENT
MEDIUM	examples/addon.node/CMakeLists.txt	27	#==================================================================	COMMENT

TODO Padding7 hits · 10 pts

Severity	File	Line	Snippet	Context
LOW	examples/talk-llama/llama-model-saver.cpp	304	// TODO: implement split file support	COMMENT
LOW	examples/talk-llama/llama-model-saver.cpp	352	// TODO: implement LoRA support	COMMENT
LOW	ggml/src/ggml.c	3880	// TODO: implement non F32 return	COMMENT
LOW	ggml/src/ggml.c	3904	// TODO: implement non F32 return	COMMENT
LOW	ggml/src/ggml.c	5050	// TODO: implement antialias for modes other than bilinear	COMMENT
LOW	ggml/src/ggml-zdnn/ggml-zdnn.cpp	22	// TODO: implement support for quantized types	COMMENT
LOW	src/parakeet.cpp	2511	// TODO: implement beam search?	COMMENT

Hallucination Indicators1 hit · 10 pts

Severity	File	Line	Snippet	Context
CRITICAL	examples/wchess/wchess.wasm/jquery-3.7.1.min.js	2	!function(e,t){"use strict";"object"==typeof module&&"object"==typeof module.exports?module.exports=e.document?t(e,!0):f	CODE

Example Usage Blocks6 hits · 9 pts

Severity	File	Line	Snippet	Context
LOW	tests/run-tests.sh	14	# Usage:	COMMENT
LOW	models/convert-h5-to-ggml.py	3	# Usage:	COMMENT
LOW	examples/generate-karaoke.sh	4	# Usage:	COMMENT
LOW	examples/yt-wsp.sh	31	# Sample usage:	COMMENT
LOW	scripts/sync-ggml-am.sh	5	# Usage:	COMMENT
LOW	ggml/src/ggml-sycl/dpct/helper.hpp	3020	/// Example usage:	COMMENT

Deep Nesting6 hits · 6 pts

Severity	File	Line	Context
LOW	tests/earnings21/normalizers/english.py	165	CODE
LOW	tests/librispeech/normalizers/english.py	165	CODE
LOW	tests/librispeech-parakeet/normalizers/english.py	165	CODE
LOW	models/convert-parakeet-to-ggml.py	108	CODE
LOW	examples/server.py	19	CODE
LOW	ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py	68	CODE

Slop Phrases1 hit · 2 pts

Severity	File	Line	Snippet	Context
LOW	tests/run-tests.sh	92	# when adding new files, make sure to add the expected "ref.txt" file with the correct transcript	COMMENT

Fake / Example Data1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	examples/talk-llama/llama-graph.cpp	1293	LLAMA_LOG_DEBUG("%s: can_reuse = %d\n", "placeholder", cur);	CODE

Overly Generic Function Names1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	bindings/javascript/whisper.js	8	function GROWABLE_HEAP_I8(){if(wasmMemory.buffer!=HEAP8.buffer){updateMemoryViews()}return HEAP8}function GROWABLE_HEAP_	CODE

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings