MétaCan
Menu
Back to cohort
Record W3211236060 · doi:10.5281/zenodo.4279923

ultralytics/yolov3: v8 - Final Darknet Compatible Release

2020· article· en· W3211236060 on OpenAlexaff
Glenn Jocher, Yonghye Kwon, guigarfr, Perry, Josh Veitch-Michaelis, Ttayu, Daniel Suess, Fatih Baltacı, Gabriel Bianconi, IlyaOvodov, Marc, Chang Lee, Dustin Kendall, Francisco Reveriano, GoogleWiki, Jason Nataprawira, Jeremy Hu, LinCoce, LukeAI, NirZarrabi, Reda Oulbacha, Piotr Skalski, Shiwei Song, Thomas Havlik, Timothy M. Shead, Xinyu Wang

Bibliographic record

VenueZenodo (CERN European Organization for Nuclear Research) · 2020
Typearticle
Languageen
FieldDecision Sciences
TopicScientific Computing and Data Management
Canadian institutionsPolytechnique Montréal
Fundersnot available
KeywordsComputer science

Abstract

fetched live from OpenAlex

This is the final release of the darknet-compatible version of the https://github.com/ultralytics/yolov3 repository. This release is backwards-compatible with darknet *.cfg files for model configuration. All pytorch (<em>.pt) and darknet (</em>.weights) models/backbones available are attached to this release in the Assets section below. Breaking Changes There are no breaking changes in this release. Bug Fixes Various Added Functionality Various Speed https://cloud.google.com/deep-learning-vm/<br> <strong>Machine type:</strong> preemptible n1-standard-8 (8 vCPUs, 30 GB memory)<br> <strong>CPU platform:</strong> Intel Skylake<br> <strong>GPUs:</strong> K80 ($0.14/hr), T4 ($0.11/hr), V100 ($0.74/hr) CUDA with Nvidia Apex FP16/32<br> <strong>HDD:</strong> 300 GB SSD <strong>Dataset:</strong> COCO train 2014 (117,263 images)<br> <strong>Model:</strong> <code>yolov3-spp.cfg</code><br> <strong>Command:</strong> <code>python3 train.py --data coco2017.data --img 416 --batch 32</code> GPU n <code>--batch-size</code> img/s epoch&lt;br&gt;time epoch&lt;br&gt;cost K80 1 32 x 2 11 175 min $0.41 T4 1&lt;br&gt;2 32 x 2&lt;br&gt;64 x 1 41&lt;br&gt;61 48 min&lt;br&gt;32 min $0.09&lt;br&gt;$0.11 V100 1&lt;br&gt;2 32 x 2&lt;br&gt;64 x 1 122&lt;br&gt;<strong>178</strong> 16 min&lt;br&gt;<strong>11 min</strong> <strong>$0.21</strong>&lt;br&gt;$0.28 2080Ti 1&lt;br&gt;2 32 x 2&lt;br&gt;64 x 1 81&lt;br&gt;140 24 min&lt;br&gt;14 min -&lt;br&gt;- mAP &lt;i&gt;&lt;/i&gt; Size COCO mAP&lt;br&gt;@0.5...0.95 COCO mAP&lt;br&gt;@0.5 YOLOv3-tiny&lt;br&gt;YOLOv3&lt;br&gt;YOLOv3-SPP&lt;br&gt;<strong>YOLOv3-SPP-ultralytics</strong> 320 14.0&lt;br&gt;28.7&lt;br&gt;30.5&lt;br&gt;<strong>37.7</strong> 29.1&lt;br&gt;51.8&lt;br&gt;52.3&lt;br&gt;<strong>56.8</strong> YOLOv3-tiny&lt;br&gt;YOLOv3&lt;br&gt;YOLOv3-SPP&lt;br&gt;<strong>YOLOv3-SPP-ultralytics</strong> 416 16.0&lt;br&gt;31.2&lt;br&gt;33.9&lt;br&gt;<strong>41.2</strong> 33.0&lt;br&gt;55.4&lt;br&gt;56.9&lt;br&gt;<strong>60.6</strong> YOLOv3-tiny&lt;br&gt;YOLOv3&lt;br&gt;YOLOv3-SPP&lt;br&gt;<strong>YOLOv3-SPP-ultralytics</strong> 512 16.6&lt;br&gt;32.7&lt;br&gt;35.6&lt;br&gt;<strong>42.6</strong> 34.9&lt;br&gt;57.7&lt;br&gt;59.5&lt;br&gt;<strong>62.4</strong> YOLOv3-tiny&lt;br&gt;YOLOv3&lt;br&gt;YOLOv3-SPP&lt;br&gt;<strong>YOLOv3-SPP-ultralytics</strong> 608 16.6&lt;br&gt;33.1&lt;br&gt;37.0&lt;br&gt;<strong>43.1</strong> 35.4&lt;br&gt;58.2&lt;br&gt;60.7&lt;br&gt;<strong>62.8</strong> TODO NA

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.003
metaresearch head score (Gemma)0.007
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesScience and technology studies, Scholarly communication, Insufficient payload (model declined to judge)
Consensus categoriesInsufficient payload (model declined to judge)
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Not applicable · Consensus signal: Not applicable
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.819
Threshold uncertainty score0.999

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0030.007
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.002
Science and technology studies0.0020.000
Scholarly communication0.0030.000
Open science0.0030.002
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0250.047

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.291
GPT teacher head0.347
Teacher spread0.056 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; both teacher heads agree on what is shown here.

Study designNot applicable
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations1
Published2020
Admission routes1
Has abstractyes

Explore more

Same venueZenodo (CERN European Organization for Nuclear Research)Same topicScientific Computing and Data ManagementFrench-language works237,207