{"id":14958106,"url":"https://github.com/joisino/speedbook","last_synced_at":"2025-10-24T14:30:45.243Z","repository":{"id":253729119,"uuid":"844344774","full_name":"joisino/speedbook","owner":"joisino","description":"書籍『深層ニューラルネットワークの高速化』のサポートサイトです。","archived":false,"fork":false,"pushed_at":"2024-09-09T09:02:01.000Z","size":492,"stargazers_count":50,"open_issues_count":0,"forks_count":1,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-01-31T01:58:46.459Z","etag":null,"topics":["deep-learning","deep-neural-networks","distillation","efficiency","neural-networks","pruning","pytorch","quantization"],"latest_commit_sha":null,"homepage":"https://www.amazon.co.jp/dp/4297143097","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/joisino.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-08-19T04:05:51.000Z","updated_at":"2025-01-28T09:26:49.000Z","dependencies_parsed_at":"2024-08-19T05:26:19.246Z","dependency_job_id":"f989f388-af20-4c5f-91e0-f861b4d77ff2","html_url":"https://github.com/joisino/speedbook","commit_stats":{"total_commits":6,"total_committers":2,"mean_commits":3.0,"dds":0.5,"last_synced_commit":"ddebb1e51b2b28be4eebe29712b34e9649d1d5d9"},"previous_names":["joisino/speedbook"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/joisino%2Fspeedbook","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/joisino%2Fspeedbook/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/joisino%2Fspeedbook/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/joisino%2Fspeedbook/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/joisino","download_url":"https://codeload.github.com/joisino/speedbook/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":237982075,"owners_count":19397193,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","deep-neural-networks","distillation","efficiency","neural-networks","pruning","pytorch","quantization"],"created_at":"2024-09-24T13:16:14.231Z","updated_at":"2025-10-24T14:30:45.228Z","avatar_url":"https://github.com/joisino.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"cover.jpg\" width=50%\u003e\n\u003c/p\u003e\n\n書籍『[深層ニューラルネットワークの高速化](https://www.amazon.co.jp/dp/4297143097)』のサポートページです。\n\n# ノートブック\n\n以下のノートブックは MIT ライセンスのもと配布されております。商用・非商用問わず自由に利用することができます。\n\n### 環境構築\n\n[Poetry](https://python-poetry.org/) をインストールした後、このディレクトリ上で\n\n```\n$ poetry install\n$ poetry run jupyter lab\n```\n\nを実行して Jupyter Lab を起動してください。\n\n### 一覧\n\n|箇所|説明|ファイル|Colab|\n|:----|:----|:----|:----:|\n|コード 2.1|GPU でのコンパイル|[compile.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/compile.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/compile.ipynb)|\n|コード 2.2|CPU でのコンパイル|[compile_cpu.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/compile_cpu.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/compile_cpu.ipynb)|\n|コード 2.3|メモリ配列の最適化 (channel last)|[channel_last.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/channel_last.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/channel_last.ipynb)|\n|コード 3.2|FP16 の利用|[fp16.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/fp16.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/fp16.ipynb)|\n|コード 3.4|CPU での BF16 の利用|[bf16_cpu.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/bf16_cpu.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/bf16_cpu.ipynb)|\n||GPU での BF16 の利用|[bf16.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/bf16.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/bf16.ipynb)|\n|コード 3.8|自動混合精度計算|[amp.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/amp.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/amp.ipynb)|\n|コード 3.9|行列の量子化|[quantize_matrix.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/quantize_matrix.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/quantize_matrix.ipynb)|\n|コード 3.10|INT8 量子化|[quantization.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/quantization.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/quantization.ipynb)|\n|コード 3.15, 3.16|ビットカウント|[bitcount.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/bitcount.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/bitcount.ipynb)|\n|コード 4.2|枝刈り|[pruning.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/pruning.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/pruning.ipynb)|\n|コード 4.3|枝刈りの速度計測|[pruning_benchmark.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/pruning_benchmark.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/pruning_benchmark.ipynb)|\n|コード 5.1|ResNext101 から ResNet18 への蒸留|[distillation.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/distillation.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/distillation.ipynb)|\n|図 6.3|顔写真データの低ランク性|[lowrankface.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/lowrankface.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/lowrankface.ipynb)|\n|コード 6.1|畳み込み層の低ランク近似|[lowrank.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/lowrank.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/lowrank.ipynb)|\n|コード 6.2|ランダム特徴量|[random_features.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/random_features.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/random_features.ipynb)|\n|コード 6.3|正値直交ランダム特徴量による高速注意 (FAVOR+)|[favor.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/favor.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/favor.ipynb)|\n|コード 8.1|プロファイリング|[profile.ipynb](https://github.com/joisino/speedbook/blob/main/notebooks/profile.ipynb)|[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/joisino/speedbook/blob/main/notebooks/profile.ipynb)|\n\n※ノートブック作成にあたり再実行したので書籍の結果とわずかに異なる場合があります。ご了承ください。\n\n# 正誤表\n\n本書の正誤情報は[正誤表](https://github.com/joisino/speedbook/blob/main/errata.md)にて公開しています。\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjoisino%2Fspeedbook","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjoisino%2Fspeedbook","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjoisino%2Fspeedbook/lists"}