github-starred/cs249r_book
2
0
Fork 0
mirror of https://github.com/harvard-edge/cs249r_book.git synced 2026-07-16 14:42:29 -05:00
Code Issues 276 Packages Projects Releases 26 Wiki Activity

feat: add multimodal figure audit automation script and README

Browse Source
This commit is contained in:
Vijay Janapa Reddi
2026-04-27 13:35:48 -04:00
parent 9ebdf77d0a
commit 42bc54275d
10 changed files with 1474 additions and 1356 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
.pre-commit-config.yaml
View File

@@ -64,10 +64,21 @@ repos:
args: ["--toml", "pyproject.toml"]
exclude: "^(_site/|_book/|htmlcov/|site/assets/games/vendor/|.*\\.js$|.*\\.pdf$)"
# --- Global: Bibliography (format + semantic) ---
# Repo-wide bibtex auto-formatter. Matches every .bib file in the repo
# (book/, tinytorch/paper/, anywhere else). The bib-lint hook below runs
# after this one so it validates the post-tidy state.
# --- Global: Bibliography (mechanical fixes + format + semantic) ---
# 1) Safe §5 field fixes (DOI prefix, title period, pages --, journal abbrevs)
# on changed .bib only — idempotent. Exits 1 if it rewrote a file
# (re-stage and commit again). Same logic as the first step of
# `book/binder bib normalize`.
# 2) bibtex-tidy — layout. 3) bib-lint (binder) — error-level schema.
- repo: local
hooks:
- id: bib-apply-mechanical
name: "Global: Apply safe §5 mechanical fixes to .bib"
entry: python3 book/tools/bib_apply_mechanical_fixes.py --pre-commit
language: system
pass_filenames: true
files: \.bib$
- repo: https://github.com/FlamingTempura/bibtex-tidy
rev: v1.14.0
hooks:
@@ -89,7 +100,7 @@ repos:
files: \.bib$
# Repo-wide bibtex semantic validator (§5 Bibliography Hygiene).
# Runs AFTER bibtex-tidy so it validates the post-tidy state. Uses the
# Runs AFTER bib-apply-mechanical + bibtex-tidy; validates post-tidy state. Uses the
# baseline allow-list at book/tools/bib_lint_baseline.json to grandfather
# pre-existing violations — only NEW violations block the commit.
# Regenerate baseline via: python3 book/tools/bib_lint.py --all --baseline
@@ -264,6 +275,13 @@ repos:
pass_filenames: false
files: ^book/quarto/contents/.*\.qmd$
- id: book-check-manual-bracket-citation
name: "Book: No manual *et al.* before bracket citations"
entry: ./book/binder check refs --scope manual-bracket
language: system
pass_filenames: false
files: ^book/quarto/contents/.*\.qmd$
- id: book-check-references
name: "Book: Check reference/citation issues"
entry: ./book/binder check refs --scope cross-refs --citations-in-code

1137
book/quarto/contents/vol1/backmatter/references.bib
View File

File diff suppressed because it is too large Load Diff

1092
book/quarto/contents/vol2/backmatter/references.bib
View File

File diff suppressed because it is too large Load Diff

70
interviews/paper/references.bib
View File

@@ -12,13 +12,13 @@
}
@inproceedings{agrawal2024sarathi,
title = {Taming Throughput-Latency Tradeoff in {LLM} Inference with {Sarathi-Serve}},
title = {Taming Throughput-Latency Tradeoff in {LLM}} Inference with {Sarathi-Serve}},
author = {
Agrawal, Amey and Kedia, Nitin and Panwar, Ashish and Mohan, Jayashree and Kwatra, Nipun and
Gulavani, Bhargav S. and Tumanov, Alexey and Ramjee, Ramachandran
},
year = {2024},
booktitle = {18th USENIX Symposium on Operating Systems Design and Implementation (OSDI)},
booktitle = {18th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI})},
publisher = {USENIX Association},
pages = {117--134},
url = {https://www.usenix.org/conference/osdi24/presentation/agrawal},
@@ -58,7 +58,7 @@
@book{anderson2001taxonomy,
title = {
A Taxonomy for Learning, Teaching, and Assessing: A Revision of Bloom's Taxonomy of Educational
A Taxonomy for Learning, Teaching, and Assessing: {A} Revision of Bloom's Taxonomy of Educational
Objectives
},
author = {
@@ -96,7 +96,7 @@
@book{biggs1982solo,
title = {
Evaluating the Quality of Learning: The {SOLO} Taxonomy (Structure of the Observed Learning
Evaluating the Quality of Learning: {The} {SOLO} Taxonomy (Structure of the Observed Learning
Outcome)
},
author = {Biggs, John B. and Collis, Kevin F.},
@@ -108,7 +108,7 @@
}
@book{bloom1956taxonomy,
title = {Taxonomy of Educational Objectives: The Classification of Educational Goals},
title = {Taxonomy of Educational Objectives: {The} Classification of Educational Goals},
author = {
Bloom, Benjamin S. and Engelhart, Max D. and Furst, Edward J. and Hill, Walker H. and
Krathwohl, David R.
@@ -171,10 +171,10 @@
}
@inproceedings{dao2022flashattention,
title = {{FlashAttention}: Fast and Memory-Efficient Exact Attention with {IO}-Awareness},
title = {{FlashAttention}}: Fast and Memory-Efficient Exact Attention with {IO}-Awareness},
author = {Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{\'e}, Christopher},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
volume = {35},
note = {
@@ -190,7 +190,7 @@
title = {{GPTQ}: Accurate Post-Training Quantization for Generative Pre-trained Transformers},
author = {Frantar, Elias and Ashkboos, Saleh and Hoefler, Torsten and Alistarh, Dan},
year = {2023},
booktitle = {International Conference on Learning Representations (ICLR)},
booktitle = {International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
url = {https://arxiv.org/abs/2210.17323},
note = {GPTQ: layer-by-layer one-shot post-training INT4 quantization for LLMs},
@@ -199,7 +199,7 @@
}
@book{gierl2013aig,
title = {Automatic Item Generation: Theory and Practice},
title = {Automatic Item Generation: {Theory} and Practice},
author = {Gierl, Mark J. and Haladyna, Thomas M.},
year = {2013},
publisher = {Routledge},
@@ -251,7 +251,7 @@
Dawn and Steinhardt, Jacob
},
year = {2021},
booktitle = {International Conference on Learning Representations (ICLR)},
booktitle = {International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
url = {https://arxiv.org/abs/2009.03300},
note = {MMLU: 57-subject benchmark for evaluating domain coverage and difficulty calibration},
@@ -261,7 +261,7 @@
}
@article{hjorland2013facet,
title = {Facet Analysis: The Logical Approach to Knowledge Organization},
title = {Facet Analysis: {The} Logical Approach to Knowledge Organization},
author = {Hj{\o}rland, Birger},
year = {2013},
journal = {Information Processing \& Management},
@@ -281,7 +281,7 @@
Mia Xu and Lee, HyoukJoong and Ngiam, Jiquan and Le, Quoc V. and Wu, Yonghui and Chen, Zhifeng
},
year = {2019},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
volume = {32},
note = {Pipeline parallelism for training models that exceed single-device memory},
@@ -310,7 +310,7 @@
Press, Ofir and Narasimhan, Karthik
},
year = {2024},
booktitle = {International Conference on Learning Representations (ICLR)},
booktitle = {International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
x-verified = {2026-04-26},
x-verified-by = {bib-web-verify},
@@ -324,7 +324,7 @@
Hao and Gonzalez, Joseph E. and Zhang, Hao and Stoica, Ion
},
year = {2023},
booktitle = {Proceedings of the ACM Symposium on Operating Systems Principles (SOSP)},
booktitle = {Proceedings of the {ACM} Symposium on Operating Systems Principles (SOSP)},
publisher = {ACM},
doi = {10.1145/3600006.3613165},
note = {vLLM: virtual memory paging for KV-cache reduces fragmentation and enables higher throughput},
@@ -348,7 +348,7 @@
title = {{MCUNet}: Tiny Deep Learning on {IoT} Devices},
author = {Lin, Ji and Chen, Wei-Ming and Lin, Yujun and Cohn, John and Gan, Chuang and Han, Song},
year = {2020},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -362,7 +362,7 @@
Wei-Chen and Xiao, Guangxuan and Dang, Xingyu and Gan, Chuang and Han, Song
},
year = {2024},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
url = {https://arxiv.org/abs/2306.00978},
note = {AWQ: salient-weight-aware INT4 weight-only quantization for LLM serving},
@@ -393,7 +393,7 @@
Zaharia, Matei
},
year = {2020},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
volume = {2},
pages = {336--349},
@@ -410,7 +410,7 @@
@article{messick1995validity,
title = {
Validity of Psychological Assessment: Validation of Inferences from Persons' Responses and
Validity of Psychological Assessment: {Validation} of Inferences from Persons' Responses and
Performances as Scientific Inquiry into Score Meaning
},
author = {Messick, Samuel},
@@ -468,7 +468,7 @@
}
@techreport{nvidia2022h100,
title = {{NVIDIA H100} Tensor Core {GPU} Architecture},
title = {{NVIDIA} H100} Tensor Core {GPU} Architecture},
author = {{NVIDIA Corporation}},
year = {2022},
url = {https://resources.nvidia.com/en-us-tensor-core/gtc22-whitepaper-hopper},
@@ -479,7 +479,7 @@
}
@techreport{nvidia2022orin,
title = {{NVIDIA Jetson AGX Orin} Series Technical Brief},
title = {{NVIDIA} Jetson AGX Orin} Series Technical Brief},
author = {{NVIDIA Corporation}},
year = {2022},
url = {https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-orin/},
@@ -490,7 +490,7 @@
}
@techreport{openai2023gpt4,
title = {GPT-4 Technical Report},
title = {{GPT}-4 Technical Report},
author = {{OpenAI}},
year = {2023},
institution = {OpenAI},
@@ -531,13 +531,13 @@
@inproceedings{rasley2020deepspeed,
title = {
{DeepSpeed}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
{DeepSpeed}}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
Parameters
},
author = {Rasley, Jeff and Rajbhandari, Samyam and Ruwase, Olatunji and He, Yuxiong},
year = {2020},
booktitle = {
Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery \& Data
Proceedings of the 26th {ACM} SIGKDD International Conference on Knowledge Discovery \& Data
Mining
},
publisher = {ACM},
@@ -567,7 +567,7 @@
Zhong, Aaron and Zhang, Peizhao and Zhou, Yuchen
},
year = {2020},
booktitle = {Proceedings of the ACM/IEEE International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the {ACM}/{IEEE} International Symposium on Computer Architecture ({ISCA})},
publisher = {IEEE},
pages = {446--459},
doi = {10.1109/ISCA45697.2020.00045},
@@ -605,7 +605,7 @@
}
@software{reddi2026mlsysim,
title = {{MLSys$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
title = {{MLSys}$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
author = {Reddi, Vijay Janapa},
year = {2026},
url = {https://mlsysbook.ai/mlsysim},
@@ -619,7 +619,7 @@
}
@article{reddi2026tinytorch,
title = {TinyTorch: Building Machine Learning Systems from First Principles},
title = {TinyTorch: {Building} Machine Learning Systems from First Principles},
author = {Reddi, Vijay Janapa},
year = {2026},
journal = {arXiv preprint arXiv:2601.19107},
@@ -632,7 +632,7 @@
}
@inproceedings{reimers2019sbert,
title = {{Sentence-BERT}: Sentence Embeddings using {Siamese BERT}-Networks},
title = {{Sentence-{BERT}}: Sentence Embeddings using {Siamese BERT}-Networks},
author = {Reimers, Nils and Gurevych, Iryna},
year = {2019},
booktitle = {
@@ -651,7 +651,7 @@
author = {Shang, Chao and Liu, Jingbo and Cheng, Jiawei and Peng, Hao and Ren, Xiang and Han, Jiawei},
year = {2018},
booktitle = {
Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery \& Data
Proceedings of the 24th {ACM} SIGKDD International Conference on Knowledge Discovery \& Data
Mining
},
publisher = {ACM},
@@ -664,7 +664,7 @@
}
@article{shoeybi2019megatron,
title = {Megatron-{LM}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
title = {{Megatron}-{LM}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
author = {
Shoeybi, Mohammad and Patwary, Mostofa and Puri, Raul and LeGresley, Patrick and Casper, Jared
and Catanzaro, Bryan
@@ -690,7 +690,7 @@
}
@book{soergel1985organizing,
title = {Organizing Information: Principles of Data Base and Retrieval Systems},
title = {Organizing Information: {Principles} of Data Base and Retrieval Systems},
author = {Soergel, Dagobert},
year = {1985},
publisher = {Academic Press},
@@ -729,7 +729,7 @@
author = {Webb, Norman L.},
year = {1997},
journal = {Research Monograph No. 6},
publisher = {National Institute for Science Education, University of Wisconsin--Madison},
publisher = {National Institute for Science Education},
note = {
Depth of Knowledge (DOK) framework: four levels of cognitive complexity for assessment
alignment
@@ -761,7 +761,7 @@
}
@article{williams2009roofline,
title = {Roofline: An Insightful Visual Performance Model for Multicore Architectures},
title = {Roofline: {An} Insightful Visual Performance Model for Multicore Architectures},
author = {Williams, Samuel and Waterman, Andrew and Patterson, David},
year = {2009},
journal = {Communications of the ACM},
@@ -782,7 +782,7 @@
author = {Yu, Gyeong-In and Jeong, Joo Seong and Kim, Geon-Woo and Kim, Soojeong and Chun, Byung-Gon},
year = {2022},
booktitle = {
Proceedings of the 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI)
Proceedings of the 16th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI})
},
publisher = {USENIX Association},
pages = {521--538},
@@ -795,14 +795,14 @@
}
@inproceedings{zheng2023judging,
title = {Judging {LLM}-as-a-Judge with {MT-Bench} and Chatbot Arena},
title = {Judging {LLM}}-as-a-Judge with {MT-Bench} and Chatbot Arena},
author = {
Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and
Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao
and Gonzalez, Joseph E. and Stoica, Ion
},
year = {2023},
booktitle = {Advances in Neural Information Processing Systems 36: Datasets and Benchmarks Track},
booktitle = {Advances in Neural Information Processing Systems 36: {Datasets} and Benchmarks Track},
publisher = {Neural Information Processing Systems Foundation},
url = {
https://proceedings.neurips.cc/paper_files/paper/2023/hash/91f18a1287b398d378ef22505bf41832-Abstract-Datasets_and_Benchmarks.html

71
mlsysim/docs/references.bib
View File

@@ -2,7 +2,7 @@
title = {Deep Learning with Differential Privacy},
author = {Abadi, Martin and Chu, Andy and Goodfellow, Ian and others},
year = {2016},
booktitle = {ACM SIGSAC Conference on Computer and Communications Security (CCS)},
booktitle = {{ACM} SIGSAC Conference on Computer and Communications Security (CCS)},
publisher = {ACM},
doi = {10.1145/2976749.2978318},
x-verified = {2026-04-08},
@@ -11,7 +11,7 @@
}
@misc{amodei2018ai,
title = {{AI} and Compute},
title = {{AI}} and Compute},
author = {Amodei, Dario and Hernandez, Danny},
year = {2018},
url = {https://openai.com/research/ai-and-compute},
@@ -22,7 +22,7 @@
}
@book{barroso2018datacenter,
title = {The Datacenter as a Computer: Designing Warehouse-Scale Machines},
title = {The Datacenter as a Computer: {Designing} Warehouse-Scale Machines},
author = {Barroso, Luiz Andr{\'e} and H{\"o}lzle, Urs and Ranganathan, Parthasarathy},
year = {2018},
publisher = {Morgan \& Claypool Publishers},
@@ -33,7 +33,7 @@
@inproceedings{calculon2023,
title = {
Calculon: a Methodology and Tool for High-Level Co-Design of Systems and Large Language Models
Calculon: {a} Methodology and Tool for High-Level Co-Design of Systems and Large Language Models
},
author = {Isaev, Mikhail and McDonald, Nic and Dennison, Larry and Vuduc, Richard},
year = {2023},
@@ -76,10 +76,10 @@
}
@inproceedings{dao2022flashattention,
title = {{FlashAttention}: Fast and Memory-Efficient Exact Attention with IO-Awareness},
title = {{FlashAttention}}: Fast and Memory-Efficient Exact Attention with IO-Awareness},
author = {Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{\'e}, Christopher},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -92,6 +92,7 @@
year = {2012},
journal = {Advances in Neural Information Processing Systems},
volume = {25},
pages = {1223--1231},
x-verified = {2026-04-09},
x-verified-by = {pass-17-bib-hygiene},
}
@@ -111,14 +112,14 @@
}
@inproceedings{eisenman2022checknrun,
title = {Check-N-Run: a Checkpointing System for Training Deep Learning Recommendation Models},
title = {Check-N-Run: {a} Checkpointing System for Training Deep Learning Recommendation Models},
author = {
Eisenman, Assaf and Matam, Kiran Kumar and Ingram, Steven and Mudigere, Dheevatsa and
Krishnamoorthi, Raghuraman and Nair, Krishnakumar and Smelyanskiy, Misha and Annavaram, Murali
},
year = {2022},
booktitle = {
Proceedings of the 19th USENIX Symposium on Networked Systems Design and Implementation (NSDI)
Proceedings of the 19th {USENIX} Symposium on Networked Systems Design and Implementation ({NSDI})
},
publisher = {USENIX Association},
x-verified = {2026-04-26},
@@ -137,7 +138,7 @@
@article{han2015deep,
title = {
Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and
Deep Compression: {Compressing} Deep Neural Networks with Pruning, Trained Quantization and
Huffman Coding
},
author = {Han, Song and Mao, Huizi and Dally, William J.},
@@ -148,7 +149,7 @@
}
@book{hennessy2019architecture,
title = {Computer Architecture: A Quantitative Approach},
title = {Computer Architecture: {A} Quantitative Approach},
author = {Hennessy, John L. and Patterson, David A.},
year = {2019},
publisher = {Morgan Kaufmann},
@@ -171,7 +172,7 @@
title = {In-Datacenter Performance Analysis of a Tensor Processing Unit},
author = {Jouppi, Norman P. and Young, Cliff and Patil, Nishant and others},
year = {2017},
booktitle = {Proceedings of the 44th Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 44th Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {ACM},
pages = {1--12},
doi = {10.1145/3079856.3080246},
@@ -195,7 +196,7 @@
title = {Efficient Memory Management for Large Language Model Serving with {PagedAttention}},
author = {Kwon, Woosuk and Li, Zhuohan and Zhuang, Siyuan and others},
year = {2023},
booktitle = {Proceedings of the 29th ACM Symposium on Operating Systems Principles (SOSP)},
booktitle = {Proceedings of the 29th {ACM} Symposium on Operating Systems Principles (SOSP)},
publisher = {ACM},
doi = {10.1145/3600006.3613165},
x-verified = {2026-04-08},
@@ -204,7 +205,7 @@
}
@article{leiserson1985fat,
title = {Fat-Trees: Universal Networks for Hardware-Efficient Supercomputing},
title = {Fat-Trees: {Universal} Networks for Hardware-Efficient Supercomputing},
author = {Leiserson, Charles E.},
year = {1985},
journal = {IEEE Transactions on Computers},
@@ -221,7 +222,7 @@
title = {Fast Inference from Transformers via Speculative Decoding},
author = {Leviathan, Yaniv and Kalman, Matan and Matias, Yossi},
year = {2023},
booktitle = {Proceedings of the 40th International Conference on Machine Learning (ICML)},
booktitle = {Proceedings of the 40th International Conference on Machine Learning ({ICML})},
publisher = {PMLR},
url = {https://arxiv.org/abs/2211.17192},
x-verified = {2026-04-08},
@@ -259,18 +260,20 @@
@book{mlsysbook2024,
title = {
Machine Learning Systems: Principles and Practices of Engineering Artificially Intelligent
Machine Learning Systems: {Principles} and Practices of Engineering Artificially Intelligent
Systems
},
author = {Reddi, Vijay Janapa},
year = {2024},
publisher = {Harvard EDGE Lab},
url = {https://mlsysbook.ai},
x-verified = {2026-04-26},
x-verified-by = {bib-web-verify},
}
@book{mlsysbook2025,
title = {
Machine Learning Systems: Principles and Practices of Engineering Artificially Intelligent
Machine Learning Systems: {Principles} and Practices of Engineering Artificially Intelligent
Systems
},
author = {Reddi, Vijay Janapa and others},
@@ -309,7 +312,7 @@
}
@inproceedings{narayanan2021efficient,
title = {Efficient Large-Scale Language Model Training on {GPU} Clusters Using {Megatron-LM}},
title = {Efficient Large-Scale Language Model Training on {GPU}} Clusters Using {Megatron-LM}},
author = {Narayanan, Deepak and Shoeybi, Mohammad and Casper, Jared and others},
year = {2021},
booktitle = {
@@ -324,7 +327,7 @@
}
@misc{nvidia2023h100,
title = {{NVIDIA H100 Tensor Core GPU} Datasheet},
title = {{NVIDIA} H100 Tensor Core {GPU}} Datasheet},
author = {{NVIDIA Corporation}},
year = {2023},
note = {Accessed: 2024-06-15},
@@ -334,13 +337,13 @@
}
@inproceedings{parashar2019timeloop,
title = {Timeloop: A Systematic Approach to {DNN} Accelerator Evaluation},
title = {Timeloop: {A} Systematic Approach to {DNN} Accelerator Evaluation},
author = {
Parashar, Angshuman and Raina, Priyanka and Shao, Yakun Sophia and Chen, Yu-Hsin and Emer, Joel
and others
},
year = {2019},
booktitle = {IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)},
booktitle = {{IEEE} International Symposium on Performance Analysis of Systems and Software (ISPASS)},
publisher = {IEEE},
doi = {10.1109/ISPASS.2019.00042},
x-verified = {2026-04-08},
@@ -349,13 +352,13 @@
}
@inproceedings{patel2024splitwise,
title = {Splitwise: Efficient Generative {LLM} Inference Using Phase Splitting},
title = {Splitwise: {Efficient} Generative {LLM}} Inference Using Phase Splitting},
author = {
Patel, Pratyush and Choukse, Esha and Zhang, Chaojie and Shah, Aashaka and Goiri, {\'I}{\~n}igo
and Maleki, Saeed and Bianchini, Ricardo
},
year = {2024},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {IEEE},
doi = {10.1109/ISCA59077.2024.00019},
x-verified = {2026-04-26},
@@ -376,7 +379,7 @@
title = {Efficiently Scaling Transformer Inference},
author = {Pope, Reiner and Douglas, Sholto and Chowdhery, Aakanksha and others},
year = {2023},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
volume = {5},
url = {
@@ -404,12 +407,12 @@
@inproceedings{rasley2020deepspeed,
title = {
{DeepSpeed}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
{DeepSpeed}}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
Parameters
},
author = {Rasley, Jeff and Rajbhandari, Samyam and Ruwase, Olatunji and He, Yuxiong},
year = {2020},
booktitle = {ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
booktitle = {{ACM} SIGKDD International Conference on Knowledge Discovery and Data Mining},
publisher = {ACM},
doi = {10.1145/3394486.3406703},
x-verified = {2026-04-08},
@@ -418,7 +421,7 @@
}
@article{shazeer2017outrageously,
title = {Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer},
title = {Outrageously Large Neural Networks: {The} Sparsely-Gated Mixture-of-Experts Layer},
author = {Shazeer, Noam and Mirhoseini, Azalia and Maziarz, Krzysztof and others},
year = {2017},
journal = {arXiv preprint arXiv:1701.06538},
@@ -427,7 +430,7 @@
}
@article{shoeybi2019megatron,
title = {{Megatron-LM}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
title = {{{Megatron}-LM}}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
author = {
Shoeybi, Mohammad and Patwary, Mostofa and Puri, Raul and LeGresley, Patrick and Casper, Jared
and Catanzaro, Bryan
@@ -439,10 +442,10 @@
}
@inproceedings{snell2025scaling,
title = {Scaling {LLM} Test-Time Compute Optimally can be More Effective than Scaling Model Parameters},
title = {Scaling {LLM}} Test-Time Compute Optimally can be More Effective than Scaling Model Parameters},
author = {Snell, Charlie and Lee, Jaehoon and Xu, Kelvin and Kumar, Aviral},
year = {2025},
booktitle = {Proceedings of the 13th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 13th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
note = {Oral presentation. arXiv:2408.03314},
x-verified = {2026-04-26},
@@ -451,7 +454,7 @@
}
@article{williams2009roofline,
title = {Roofline: An Insightful Visual Performance Model for Multicore Architectures},
title = {Roofline: {An} Insightful Visual Performance Model for Multicore Architectures},
author = {Williams, Samuel and Waterman, Andrew and Patterson, David},
year = {2009},
journal = {Communications of the ACM},
@@ -475,7 +478,7 @@
Sudarshan and Krishna, Tushar
},
year = {2023},
booktitle = {IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)},
booktitle = {{IEEE} International Symposium on Performance Analysis of Systems and Software (ISPASS)},
publisher = {IEEE},
doi = {10.1109/ISPASS57527.2023.00035},
x-verified = {2026-04-08},
@@ -484,10 +487,10 @@
}
@inproceedings{wu2019accelergy,
title = {Accelergy: An Architecture-Level Energy Estimation Methodology for Accelerator Designs},
title = {Accelergy: {An} Architecture-Level Energy Estimation Methodology for Accelerator Designs},
author = {Wu, Yannan Nellie and Emer, Joel S. and Sze, Vivienne},
year = {2019},
booktitle = {IEEE/ACM International Conference on Computer-Aided Design (ICCAD)},
booktitle = {{IEEE}/{ACM} International Conference on Computer-Aided Design (ICCAD)},
publisher = {IEEE},
doi = {10.1109/ICCAD45719.2019.8942149},
x-verified = {2026-04-08},
@@ -510,7 +513,7 @@
}
@article{zheng2024sglang,
title = {{SGLang}: Efficient Execution of Structured Language Model Programs},
title = {{SGLang}}: Efficient Execution of Structured Language Model Programs},
author = {Zheng, Lianmin and Yin, Liangsheng and Xie, Zhiqiang and others},
year = {2024},
journal = {arXiv preprint arXiv:2312.07104},

161
mlsysim/paper/references.bib
View File

@@ -2,7 +2,7 @@
title = {Deep Learning with Differential Privacy},
author = {Abadi, Martin and Chu, Andy and Goodfellow, Ian and others},
year = {2016},
booktitle = {Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security},
booktitle = {Proceedings of the 2016 {ACM} SIGSAC Conference on Computer and Communications Security},
publisher = {ACM},
pages = {308--318},
doi = {10.1145/2976749.2978318},
@@ -12,31 +12,33 @@
}
@inproceedings{agrawal2024sarathi,
title = {Taming Throughput-Latency Tradeoff in {LLM} Inference with {Sarathi-Serve}},
title = {Taming Throughput-Latency Tradeoff in {LLM}} Inference with {Sarathi-Serve}},
author = {
Agrawal, Amey and Kedia, Nitin and Panwar, Ashish and Mohan, Jayashree and Kwatra, Nipun and
Gulavani, Bhargav S. and Tumanov, Alexey and Ramjee, Ramachandran
},
year = {2024},
booktitle = {
Proceedings of the 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI)
Proceedings of the 18th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI})
},
publisher = {USENIX Association},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
publisher = {USENIX Association},
pages = {117--134},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
x-verified-source = {https://www.usenix.org/conference/osdi24/presentation/agrawal},
}
@inproceedings{agrawal2024vidur,
title = {Vidur: A Large-Scale Simulation Framework For {LLM} Inference},
title = {Vidur: {A} Large-Scale Simulation Framework For {LLM}} Inference},
author = {
Agrawal, Amey and Kedia, Nitin and Panwar, Ashish and Mohan, Jayashree and Kwatra, Nipun and
Gulavani, Bhargav S. and Tumanov, Alexey and Ramjee, Ramachandran
},
year = {2024},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
publisher = {mlsys.org},
url = {https://arxiv.org/abs/2405.05465},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
pages = {351--366},
url = {https://arxiv.org/abs/2405.05465},
x-verified = {2026-04-26},
x-verified-by = {bib-web-verify},
x-verified-source = {
@@ -45,7 +47,7 @@
}
@misc{amodei2018ai,
title = {{AI} and Compute},
title = {{AI}} and Compute},
author = {Amodei, Dario and Hernandez, Danny},
year = {2018},
url = {https://openai.com/research/ai-and-compute},
@@ -56,7 +58,7 @@
}
@article{bambhaniya2024genz,
title = {Demystifying Platform Requirements for Diverse {LLM} Inference Use Cases},
title = {Demystifying Platform Requirements for Diverse {LLM}} Inference Use Cases},
author = {Bambhaniya, Aman and Shao, Ritik and Juneja, Suhas Somashekar and others},
year = {2024},
journal = {arXiv preprint arXiv:2406.01698},
@@ -79,7 +81,7 @@
}
@book{barroso2018datacenter,
title = {The Datacenter as a Computer: Designing Warehouse-Scale Machines},
title = {The Datacenter as a Computer: {Designing} Warehouse-Scale Machines},
author = {Barroso, Luiz Andr{\'e} and H{\"o}lzle, Urs and Ranganathan, Parthasarathy},
year = {2018},
publisher = {Morgan \& Claypool},
@@ -119,7 +121,7 @@
@inproceedings{calculon2023,
title = {
Calculon: a Methodology and Tool for High-Level Co-Design of Systems and Large Language Models
Calculon: {a} Methodology and Tool for High-Level Co-Design of Systems and Large Language Models
},
author = {Isaev, Mikhail and McDonald, Nic and Dennison, Larry and Vuduc, Richard},
year = {2023},
@@ -147,7 +149,7 @@
}
@misc{cox2011xv6,
title = {xv6: A Simple, {Unix}-like Teaching Operating System},
title = {xv6: {A} Simple, {Unix}-like Teaching Operating System},
author = {Cox, Russ and Kaashoek, M. Frans and Morris, Robert},
year = {2011},
url = {https://pdos.csail.mit.edu/6.828/xv6},
@@ -172,10 +174,10 @@
}
@inproceedings{dao2022flashattention,
title = {{FlashAttention}: Fast and Memory-Efficient Exact Attention with {IO}-Awareness},
title = {{FlashAttention}}: Fast and Memory-Efficient Exact Attention with {IO}-Awareness},
author = {Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{\'e}, Christopher},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
volume = {35},
url = {https://arxiv.org/abs/2205.14135},
@@ -187,13 +189,14 @@
}
@article{dean2012large,
title = {Large Scale Distributed Deep Networks},
author = {Dean, Jeffrey and Corrado, Greg S. and Monga, Rajat and others},
year = {2012},
journal = {Advances in Neural Information Processing Systems},
volume = {25},
x-verified = {2026-04-09},
x-verified-by = {pass-17-bib-hygiene},
title = {Large Scale Distributed Deep Networks},
author = {Dean, Jeffrey and Corrado, Greg S. and Monga, Rajat and others},
year = 2012,
journal = {Advances in Neural Information Processing Systems},
volume = {25},
pages = {1223--1231},
x-verified = {2026-04-09},
x-verified-by = {pass-17-bib-hygiene},
}
@article{dean2013tail,
@@ -217,7 +220,7 @@
},
author = {{DeepSeek-AI}},
year = {2025},
booktitle = {Proceedings of the 52nd Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 52nd Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {ACM},
doi = {10.1145/3695053.3731412},
x-verified = {2026-04-08},
@@ -226,14 +229,14 @@
}
@inproceedings{eisenman2022checknrun,
title = {Check-N-Run: a Checkpointing System for Training Deep Learning Recommendation Models},
title = {Check-N-Run: {a} Checkpointing System for Training Deep Learning Recommendation Models},
author = {
Eisenman, Assaf and Matam, Kiran Kumar and Ingram, Steven and Mudigere, Dheevatsa and
Krishnamoorthi, Raghuraman and Nair, Krishnakumar and Smelyanskiy, Misha and Annavaram, Murali
},
year = {2022},
booktitle = {
Proceedings of the 19th USENIX Symposium on Networked Systems Design and Implementation (NSDI)
Proceedings of the 19th {USENIX} Symposium on Networked Systems Design and Implementation ({NSDI})
},
publisher = {USENIX Association},
x-verified = {2026-04-26},
@@ -248,7 +251,7 @@
Noman, Abdulrahman
},
year = {2024},
booktitle = {Proceedings of the 12th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 12th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -259,7 +262,7 @@
title = {{GPTQ}: Accurate Post-Training Quantization for Generative Pre-trained Transformers},
author = {Frantar, Elias and Ashkboos, Saleh and Hoefler, Torsten and Alistarh, Dan},
year = {2023},
booktitle = {Proceedings of the 11th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 11th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -282,7 +285,7 @@
Brooks, David and Wu, Carole-Jean
},
year = {2022},
booktitle = {Proceedings of the 49th Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 49th Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {ACM},
doi = {10.1145/3470496.3527408},
x-verified = {2026-04-08},
@@ -291,13 +294,13 @@
}
@inproceedings{gupta2022chasing,
title = {Chasing Carbon: The Elusive Environmental Footprint of Computing},
title = {Chasing Carbon: {The} Elusive Environmental Footprint of Computing},
author = {
Gupta, Udit and Kim, Young Geun and Lee, Sylvia and Tse, Jordan and Lee, Hsien-Hsin S and Wei,
Gu-Yeon and Brooks, David and Wu, Carole-Jean
},
year = {2022},
booktitle = {IEEE International Symposium on High-Performance Computer Architecture (HPCA)},
booktitle = {{IEEE} International Symposium on High-Performance Computer Architecture ({HPCA})},
publisher = {IEEE},
doi = {10.1109/HPCA53966.2022.00076},
x-verified = {2026-04-08},
@@ -307,12 +310,12 @@
@inproceedings{han2016deep,
title = {
Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and
Deep Compression: {Compressing} Deep Neural Networks with Pruning, Trained Quantization and
Huffman Coding
},
author = {Han, Song and Mao, Huizi and Dally, William J.},
year = {2016},
booktitle = {Proceedings of the 4th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 4th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
note = {Best Paper Award},
x-verified = {2026-04-08},
@@ -321,7 +324,7 @@
}
@book{hennessy2024architecture,
title = {Computer Architecture: A Quantitative Approach},
title = {Computer Architecture: {A} Quantitative Approach},
author = {Hennessy, John L. and Patterson, David A. and Kozyrakis, Christos},
year = {2024},
publisher = {Morgan Kaufmann},
@@ -335,7 +338,7 @@
title = {Training Compute-Optimal Large Language Models},
author = {Hoffmann, Jordan and Borgeaud, Sebastian and Mensch, Arthur and others},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
publisher = {Curran Associates Inc.},
volume = {35},
url = {https://arxiv.org/abs/2203.15556},
@@ -350,7 +353,7 @@
title = {Beyond Data and Model Parallelism for Deep Neural Networks},
author = {Jia, Zhihao and Zaharia, Matei and Aiken, Alex},
year = {2019},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
url = {https://arxiv.org/abs/1807.05358},
x-verified = {2026-04-26},
@@ -364,7 +367,7 @@
title = {In-Datacenter Performance Analysis of a Tensor Processing Unit},
author = {Jouppi, Norman P. and Young, Cliff and Patil, Nishant and others},
year = {2017},
booktitle = {Proceedings of the 44th Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 44th Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {ACM},
pages = {1--12},
doi = {10.1145/3079856.3080246},
@@ -383,7 +386,7 @@
}
@misc{kim2023llmanalysis,
title = {llm-analysis: Latency and Memory Analysis of Transformer Models},
title = {{LLM}-analysis: {Latency} and Memory Analysis of Transformer Models},
author = {Li, Cheng},
year = {2023},
note = {Accessed: 2025-01-15},
@@ -396,7 +399,7 @@
title = {Efficient Memory Management for Large Language Model Serving with {PagedAttention}},
author = {Kwon, Woosuk and Li, Zhuohan and Zhuang, Siyuan and others},
year = {2023},
booktitle = {Proceedings of the 29th ACM Symposium on Operating Systems Principles (SOSP)},
booktitle = {Proceedings of the 29th {ACM} Symposium on Operating Systems Principles (SOSP)},
publisher = {ACM},
doi = {10.1145/3600006.3613165},
x-verified = {2026-04-08},
@@ -405,7 +408,7 @@
}
@article{leiserson1985fat,
title = {Fat-Trees: Universal Networks for Hardware-Efficient Supercomputing},
title = {Fat-Trees: {Universal} Networks for Hardware-Efficient Supercomputing},
author = {Leiserson, Charles E.},
year = {1985},
journal = {IEEE Transactions on Computers},
@@ -420,7 +423,7 @@
title = {Fast Inference from Transformers via Speculative Decoding},
author = {Leviathan, Yaniv and Kalman, Matan and Matias, Yossi},
year = {2023},
booktitle = {Proceedings of the 40th International Conference on Machine Learning (ICML)},
booktitle = {Proceedings of the 40th International Conference on Machine Learning ({ICML})},
publisher = {PMLR},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -428,13 +431,13 @@
}
@inproceedings{liang2025lumos,
title = {Lumos: Efficient Performance Modeling and Estimation for Large-scale {LLM} Training},
title = {Lumos: {Efficient} Performance Modeling and Estimation for Large-scale {LLM}} Training},
author = {
Liang, Mingyu and Kassa, Hiwot Tadese and Fu, Wenyin and Coutinho, Brian and Feng, Louis and
Delimitrou, Christina
},
year = {2025},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -442,10 +445,10 @@
}
@inproceedings{lie2022cerebras,
title = {Cerebras Architecture Deep Dive: First Look Inside the {HW/SW} Co-Design for Deep Learning},
title = {Cerebras Architecture Deep Dive: {First} Look Inside the {HW/SW} Co-Design for Deep Learning},
author = {Lie, Sean},
year = {2022},
booktitle = {IEEE Hot Chips 34 Symposium},
booktitle = {{IEEE} Hot Chips 34 Symposium},
publisher = {IEEE},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -459,7 +462,7 @@
Wei-Chen and Xiao, Guangxuan and Dang, Xingyu and Gan, Chuang and Han, Song
},
year = {2024},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
url = {https://arxiv.org/abs/2306.00978},
x-verified = {2026-04-26},
@@ -493,7 +496,7 @@
}
@misc{lottick2019codecarbon,
title = {Energy Usage Reports: Environmental Awareness as Part of Algorithmic Accountability},
title = {Energy Usage Reports: {Environmental} Awareness as Part of Algorithmic Accountability},
author = {Lottick, Kadan and Susai, Silvia and Friedler, Sorelle A. and Wilson, Jonathan P.},
year = {2019},
howpublished = {arXiv preprint arXiv:1911.08354},
@@ -517,7 +520,7 @@
@book{mlsysbook2025,
title = {
Machine Learning Systems: Principles and Practices of Engineering Artificially Intelligent
Machine Learning Systems: {Principles} and Practices of Engineering Artificially Intelligent
Systems
},
author = {Reddi, Vijay Janapa and others},
@@ -542,7 +545,7 @@
}
@inproceedings{murray2021tf,
title = {tf.data: A Machine Learning Data Processing Framework},
title = {tf.data: {A} Machine Learning Data Processing Framework},
author = {Murray, Derek G. and Simsa, Jiri and Klimovic, Ana and Indyk, Ihor},
year = {2021},
booktitle = {Proceedings of the VLDB Endowment},
@@ -556,7 +559,7 @@
}
@inproceedings{narayanan2021efficient,
title = {Efficient Large-Scale Language Model Training on {GPU} Clusters Using {Megatron-LM}},
title = {Efficient Large-Scale Language Model Training on {GPU}} Clusters Using {Megatron-LM}},
author = {Narayanan, Deepak and Shoeybi, Mohammad and Casper, Jared and others},
year = {2021},
booktitle = {
@@ -571,7 +574,7 @@
}
@misc{nvidia2023h100,
title = {{NVIDIA H100 Tensor Core GPU} Datasheet},
title = {{NVIDIA} H100 Tensor Core {GPU}} Datasheet},
author = {{NVIDIA Corporation}},
year = {2023},
note = {Accessed: 2024-06-15},
@@ -581,13 +584,13 @@
}
@inproceedings{parashar2019timeloop,
title = {Timeloop: A Systematic Approach to {DNN} Accelerator Evaluation},
title = {Timeloop: {A} Systematic Approach to {DNN} Accelerator Evaluation},
author = {
Parashar, Angshuman and Raina, Priyanka and Shao, Yakun Sophia and Chen, Yu-Hsin and Emer, Joel
and others
},
year = {2019},
booktitle = {IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)},
booktitle = {{IEEE} International Symposium on Performance Analysis of Systems and Software (ISPASS)},
publisher = {IEEE},
doi = {10.1109/ISPASS.2019.00042},
x-verified = {2026-04-08},
@@ -596,13 +599,13 @@
}
@inproceedings{patel2024splitwise,
title = {Splitwise: Efficient Generative {LLM} Inference Using Phase Splitting},
title = {Splitwise: {Efficient} Generative {LLM}} Inference Using Phase Splitting},
author = {
Patel, Pratyush and Choukse, Esha and Zhang, Chaojie and Shah, Aashaka and Goiri, {\'I}{\~n}igo
and Maleki, Saeed and Bianchini, Ricardo
},
year = {2024},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {IEEE},
doi = {10.1109/ISCA59077.2024.00019},
x-verified = {2026-04-26},
@@ -611,7 +614,7 @@
}
@book{patterson2014organization,
title = {Computer Organization and Design: The Hardware/Software Interface},
title = {Computer Organization and Design: {The} Hardware/Software Interface},
author = {Patterson, David A. and Hennessy, John L.},
year = {2014},
publisher = {Morgan Kaufmann},
@@ -634,7 +637,7 @@
title = {Efficiently Scaling Transformer Inference},
author = {Pope, Reiner and Douglas, Sholto and Chowdhery, Aakanksha and others},
year = {2023},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
volume = {5},
url = {
@@ -649,7 +652,7 @@
title = {{PALEO}: A Performance Model for Deep Neural Networks},
author = {Qi, Hang and Sparks, Evan R. and Talwalkar, Ameet},
year = {2017},
booktitle = {Proceedings of the 5th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 5th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -673,12 +676,12 @@
@inproceedings{rasley2020deepspeed,
title = {
{DeepSpeed}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
{DeepSpeed}}: System Optimizations Enable Training Deep Learning Models with Over 100 Billion
Parameters
},
author = {Rasley, Jeff and Rajbhandari, Samyam and Ruwase, Olatunji and He, Yuxiong},
year = {2020},
booktitle = {ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
booktitle = {{ACM} SIGKDD International Conference on Knowledge Discovery and Data Mining},
publisher = {ACM},
doi = {10.1145/3394486.3406703},
x-verified = {2026-04-08},
@@ -687,10 +690,10 @@
}
@inproceedings{shazeer2017outrageously,
title = {Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer},
title = {Outrageously Large Neural Networks: {The} Sparsely-Gated Mixture-of-Experts Layer},
author = {Shazeer, Noam and Mirhoseini, Azalia and Maziarz, Krzysztof and others},
year = {2017},
booktitle = {Proceedings of the 5th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 5th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -698,7 +701,7 @@
}
@article{shoeybi2019megatron,
title = {{Megatron-LM}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
title = {{{Megatron}-LM}}: Training Multi-Billion Parameter Language Models Using Model Parallelism},
author = {
Shoeybi, Mohammad and Patwary, Mostofa and Puri, Raul and LeGresley, Patrick and Casper, Jared
and Catanzaro, Bryan
@@ -710,10 +713,10 @@
}
@inproceedings{snell2025scaling,
title = {Scaling {LLM} Test-Time Compute Optimally can be More Effective than Scaling Model Parameters},
title = {Scaling {LLM}} Test-Time Compute Optimally can be More Effective than Scaling Model Parameters},
author = {Snell, Charlie and Lee, Jaehoon and Xu, Kelvin and Kumar, Aviral},
year = {2025},
booktitle = {Proceedings of the 13th International Conference on Learning Representations (ICLR)},
booktitle = {Proceedings of the 13th International Conference on Learning Representations ({ICLR})},
publisher = {OpenReview.net},
note = {Oral presentation. arXiv:2408.03314},
x-verified = {2026-04-26},
@@ -745,7 +748,7 @@
}
@book{tanenbaum2006minix,
title = {Operating Systems: Design and Implementation},
title = {Operating Systems: {Design} and Implementation},
author = {Tanenbaum, Andrew S. and Woodhull, Albert S.},
year = {2006},
publisher = {Prentice Hall},
@@ -773,7 +776,7 @@
author = {Wang, Zhuo and Zheng, Weicheng and Liu, Chengwei and others},
year = {2025},
booktitle = {
Proceedings of the 22nd USENIX Symposium on Networked Systems Design and Implementation (NSDI)
Proceedings of the 22nd {USENIX} Symposium on Networked Systems Design and Implementation ({NSDI})
},
publisher = {USENIX Association},
x-verified = {2026-04-08},
@@ -782,7 +785,7 @@
}
@article{williams2009roofline,
title = {Roofline: An Insightful Visual Performance Model for Multicore Architectures},
title = {Roofline: {An} Insightful Visual Performance Model for Multicore Architectures},
author = {Williams, Samuel and Waterman, Andrew and Patterson, David},
year = {2009},
journal = {Communications of the ACM},
@@ -806,7 +809,7 @@
Sudarshan and Krishna, Tushar
},
year = {2023},
booktitle = {IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)},
booktitle = {{IEEE} International Symposium on Performance Analysis of Systems and Software (ISPASS)},
publisher = {IEEE},
doi = {10.1109/ISPASS57527.2023.00035},
x-verified = {2026-04-08},
@@ -816,7 +819,7 @@
@misc{wongpanich2025fleet,
title = {
Machine Learning Fleet Efficiency: Analyzing and Optimizing Large-Scale {Google TPU} Systems
Machine Learning Fleet Efficiency: {Analyzing} and Optimizing Large-Scale {Google} {TPU}} Systems
with {ML} Productivity Goodput
},
author = {
@@ -831,10 +834,10 @@
}
@inproceedings{wu2019accelergy,
title = {Accelergy: An Architecture-Level Energy Estimation Methodology for Accelerator Designs},
title = {Accelergy: {An} Architecture-Level Energy Estimation Methodology for Accelerator Designs},
author = {Wu, Yannan Nellie and Emer, Joel S. and Sze, Vivienne},
year = {2019},
booktitle = {IEEE/ACM International Conference on Computer-Aided Design (ICCAD)},
booktitle = {{IEEE}/{ACM} International Conference on Computer-Aided Design (ICCAD)},
publisher = {IEEE},
doi = {10.1109/ICCAD45719.2019.8942149},
x-verified = {2026-04-08},
@@ -857,10 +860,10 @@
}
@inproceedings{yu2021habitat,
title = {Habitat: A Runtime-Based Computational Performance Predictor for Deep Neural Network Training},
title = {Habitat: {A} Runtime-Based Computational Performance Predictor for Deep Neural Network Training},
author = {Yu, Geoffrey X. and Gao, Yubo and Golber, Pavel and Pekhimenko, Gennady},
year = {2021},
booktitle = {Proceedings of the 2021 USENIX Annual Technical Conference (ATC)},
booktitle = {Proceedings of the 2021 {USENIX} Annual Technical Conference (ATC)},
publisher = {USENIX Association},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -868,7 +871,7 @@
}
@article{yuan2024llmviewer,
title = {{LLM} Inference Unveiled: Survey and Roofline Model Insights},
title = {{LLM}} Inference Unveiled: Survey and Roofline Model Insights},
author = {Yuan, Zhihang and Shang, Yuzhang and Zhou, Yang and others},
year = {2024},
journal = {arXiv preprint arXiv:2402.16363},
@@ -880,7 +883,7 @@
title = {{LLMCompass}: Enabling Efficient Hardware Design for Large Language Model Inference},
author = {Zhang, Hengrui and Ning, August and Peng, Rohan and others},
year = {2024},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture (ISCA)},
booktitle = {Proceedings of the 51st Annual International Symposium on Computer Architecture ({ISCA})},
publisher = {IEEE},
doi = {10.1109/ISCA59077.2024.00082},
x-verified = {2026-04-26},
@@ -889,7 +892,7 @@
}
@article{zheng2024sglang,
title = {{SGLang}: Efficient Execution of Structured Language Model Programs},
title = {{SGLang}}: Efficient Execution of Structured Language Model Programs},
author = {Zheng, Lianmin and Yin, Liangsheng and Xie, Zhiqiang and others},
year = {2024},
journal = {arXiv preprint arXiv:2312.07104},
@@ -908,7 +911,7 @@
},
year = {2024},
booktitle = {
Proceedings of the 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI)
Proceedings of the 18th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI})
},
publisher = {USENIX Association},
x-verified = {2026-04-08},

42
periodic-table/paper/references.bib
View File

@@ -1,5 +1,5 @@
@techreport{asanovic2006landscape,
title = {The Landscape of Parallel Computing Research: A View from Berkeley},
title = {The Landscape of Parallel Computing Research: {A} View from Berkeley},
author = {Asanovic, Krste and others},
year = {2006},
number = {UCB/EECS-2006-183},
@@ -10,10 +10,10 @@
}
@inproceedings{chen2018tvm,
title = {{TVM}: An automated end-to-end optimizing compiler for deep learning},
title = {{TVM}}: An automated end-to-end optimizing compiler for deep learning},
author = {Chen, Tianqi and others},
year = {2018},
booktitle = {13th USENIX Symposium on Operating Systems Design and Implementation (OSDI 18)},
booktitle = {13th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 18)},
publisher = {USENIX Association},
pages = {578--594},
x-verified = {2026-04-08},
@@ -24,7 +24,7 @@
}
@inproceedings{dao2022flashattention,
title = {FlashAttention: Fast and memory-efficient exact attention with IO-awareness},
title = {{FlashAttention}: {Fast} and memory-efficient exact attention with IO-awareness},
author = {Dao, Tri and others},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems},
@@ -49,7 +49,7 @@
}
@article{fedus2022switch,
title = {Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity},
title = {Switch transformers: {Scaling} to trillion parameter models with simple and efficient sparsity},
author = {Fedus, William and Zoph, Barret and Shazeer, Noam},
year = {2022},
journal = {Journal of Machine Learning Research},
@@ -61,7 +61,7 @@
}
@article{gu2023mamba,
title = {Mamba: Linear-time sequence modeling with selective state spaces},
title = {Mamba: {Linear}-time sequence modeling with selective state spaces},
author = {Gu, Albert and Dao, Tri},
year = {2023},
journal = {arXiv preprint arXiv:2312.00752},
@@ -70,7 +70,7 @@
}
@book{hennessy2019architecture,
title = {Computer Architecture: A Quantitative Approach},
title = {Computer Architecture: {A} Quantitative Approach},
author = {Hennessy, John L. and Patterson, David A.},
year = {2019},
publisher = {Morgan Kaufmann},
@@ -117,7 +117,7 @@
}
@inproceedings{ivanov2021data,
title = {Data movement is all you need: A case study on optimizing transformers},
title = {Data movement is all you need: {A} case study on optimizing transformers},
author = {Ivanov, Andrei and Dryden, Nikoli and Ben-Nun, Tal and Li, Shigang and Hoefler, Torsten},
year = {2021},
booktitle = {Proceedings of Machine Learning and Systems},
@@ -172,10 +172,10 @@
}
@inproceedings{lattner2021mlir,
title = {{MLIR}: Scaling compiler infrastructure for domain specific computation},
title = {{MLIR}}: Scaling compiler infrastructure for domain specific computation},
author = {Lattner, Chris and others},
year = {2021},
booktitle = {2021 IEEE/ACM International Symposium on Code Generation and Optimization (CGO)},
booktitle = {2021 {IEEE}/{ACM} International Symposium on Code Generation and Optimization (CGO)},
publisher = {IEEE},
pages = {2--14},
doi = {10.1109/CGO51591.2021.9370308},
@@ -216,13 +216,13 @@
}
@inproceedings{narayanan2019pipedream,
title = {PipeDream: Generalized pipeline parallelism for {DNN} training},
title = {PipeDream: {Generalized} pipeline parallelism for {DNN} training},
author = {
Narayanan, Deepak and Harlap, Aaron and Phanishayee, Amar and Seshadri, Vivek and Devanur,
Nikhil R. and Ganger, Gregory R. and Gibbons, Phillip B. and Zaharia, Matei
},
year = {2019},
booktitle = {Proceedings of the 27th ACM Symposium on Operating Systems Principles},
booktitle = {Proceedings of the 27th {ACM} Symposium on Operating Systems Principles},
publisher = {ACM},
pages = {1--15},
doi = {10.1145/3341301.3359646},
@@ -231,7 +231,7 @@
}
@inproceedings{narayanan2021efficient,
title = {Efficient large-scale language model training on {GPU} clusters using {Megatron-LM}},
title = {Efficient large-scale language model training on {GPU}} clusters using {Megatron-LM}},
author = {
Narayanan, Deepak and Shoeybi, Mohammad and Casper, Jared and LeGresley, Patrick and Patwary,
Mostofa and Korthikanti, Vijay and Vainbrand, Dmitri and Kasber, Prethvi and Catanzaro, Bryan
@@ -264,7 +264,7 @@
}
@article{rafailov2024direct,
title = {Direct preference optimization: Your language model is secretly a reward model},
title = {Direct preference optimization: {Your} language model is secretly a reward model},
author = {Rafailov, Rafael and others},
year = {2024},
journal = {Advances in Neural Information Processing Systems},
@@ -273,11 +273,11 @@
}
@inproceedings{rajbhandari2020zero,
title = {ZeRO: Memory optimizations toward training trillion parameter models},
title = {ZeRO: {Memory} optimizations toward training trillion parameter models},
author = {Rajbhandari, Samyam and others},
year = {2020},
booktitle = {
SC20: International Conference for High Performance Computing, Networking, Storage and Analysis
SC20: {International} Conference for High Performance Computing, Networking, Storage and Analysis
},
publisher = {IEEE},
doi = {10.1109/SC41405.2020.00024},
@@ -298,7 +298,7 @@
}
@misc{reddi2026mlsysim,
title = {{MLSys$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
title = {{MLSys}$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
author = {Reddi, Vijay Janapa},
year = {2026},
url = {https://mlsysbook.ai/mlsysim},
@@ -318,7 +318,7 @@
}
@inproceedings{shoeybi2019megatron,
title = {Megatron-{LM}: Training multi-billion parameter language models using model parallelism},
title = {{Megatron}-{LM}: Training multi-billion parameter language models using model parallelism},
author = {
Shoeybi, Mohammad and Patwary, Mostofa and Puri, Raul and LeGresley, Patrick and Casper, Jared
and Catanzaro, Bryan
@@ -367,7 +367,7 @@
}
@article{williams2009roofline,
title = {Roofline: an insightful visual performance model for multicore architectures},
title = {Roofline: {an} insightful visual performance model for multicore architectures},
author = {Williams, Samuel and Waterman, Andrew and Patterson, David},
year = {2009},
journal = {Communications of the ACM},
@@ -379,10 +379,10 @@
}
@inproceedings{yu2022orca,
title = {Orca: A distributed serving system for {Transformer}-based generative models},
title = {Orca: {A} distributed serving system for {Transformer}-based generative models},
author = {Yu, Gyeong-In and Jeong, Joo Seong and Kim, Geon-Woo and Kim, Soojeong and Chun, Byung-Gon},
year = {2022},
booktitle = {16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22)},
booktitle = {16th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 22)},
publisher = {USENIX Association},
pages = {521--538},
x-verified = {2026-04-08},

31
scripts/README.md Normal file
View File

@@ -0,0 +1,31 @@
# Figure Audit Automation
This directory contains `figure_audit.py`, a script designed to automate the visual auditing of figures within the ML Systems textbook.
## What it does
The script orchestrates a multimodal audit of every figure across Volume 1 and Volume 2 of the textbook. It ensures that the prose, the captions (`fig-cap`), and the alt-text (`fig-alt`) precisely match the content of the fully rendered visual images.
1. **Discovery:** It scans the `book/quarto/contents/` directory to identify all `.qmd` chapters containing figures.
2. **Visual Extraction:** It resolves the corresponding published HTML URL for each chapter, parses the HTML, and downloads the exact rendered `<img src="...">` and inline `<svg>` visual assets locally.
3. **Auditing:** It dispatches parallel worker tasks via the `gemini` CLI. The CLI is given explicit instructions to load the local images visually, compare them directly against the `.qmd` source text, and evaluate them based on the `figure-audit-brief.md` rubric.
4. **Reporting:** It generates strict, granular YAML output files in `.claude/_reviews/Figure Audit/`, detailing any misalignments (e.g., the text claims $10^4$ but the chart shows $10^3$) along with surgically precise `.qmd` fix recommendations.
## How to use it
Run the script from the repository root:
```bash
python3 scripts/figure_audit.py
```
### Pre-requisites
* You must have `gemini` CLI installed and authenticated on your local machine.
* The script assumes the rendered HTML book is available at `https://harvard-edge.github.io/cs249r_book_dev/...` (used purely to scrape the final image variants).
### Applying the fixes
Once `figure_audit.py` finishes running, your `.claude/_reviews/Figure Audit/` directory will be populated with `.yml` files containing `proposed_fix` entries.
These fixes are written as precise, minimal adjustments targeting the `.qmd` source files. They can either be applied manually by a human reviewing the YAML reports, or parsed programmatically/agentically to apply the diffs across the workspace.

122
scripts/figure_audit.py Normal file
View File

@@ -0,0 +1,122 @@
import os
import glob
import subprocess
import urllib.request
import urllib.parse
from bs4 import BeautifulSoup
from concurrent.futures import ThreadPoolExecutor
# Default paths assuming the script is run from the repository root
book_dir = "book/quarto/contents"
out_dir = ".claude/_reviews/Figure Audit"
img_tmp_dir = ".claude/_reviews/Figure Audit/audit_images_tmp"
brief_path = ".claude/_plans/figure-audit-brief.md"
os.makedirs(out_dir, exist_ok=True)
os.makedirs(img_tmp_dir, exist_ok=True)
qmd_files = []
for vol in ['vol1', 'vol2']:
vol_dir = os.path.join(book_dir, vol)
for root, _, files in os.walk(vol_dir):
for f in files:
if f.endswith('.qmd') and not f.startswith('_'):
qmd_files.append(os.path.join(root, f))
chapters_with_figures = []
for qmd in qmd_files:
with open(qmd, 'r', encoding='utf-8') as f:
if 'fig-cap' in f.read():
chapters_with_figures.append(qmd)
print(f"Found {len(chapters_with_figures)} chapters with figures.")
def download_file(url, local_path):
req = urllib.request.Request(url, headers={'User-Agent': 'Mozilla/5.0'})
try:
with urllib.request.urlopen(req) as response, open(local_path, 'wb') as out_file:
data = response.read()
out_file.write(data)
return True
except Exception as e:
print(f"Failed to download {url}: {e}")
return False
def process_chapter(qmd_path):
rel_path = os.path.relpath(qmd_path, book_dir)
parts = rel_path.split(os.sep)
vol = parts[0]
if len(parts) > 2:
chap = parts[-2]
else:
chap = parts[-1].replace('.qmd', '')
out_name = f"{vol}-{chap}.yml"
out_path = os.path.join(out_dir, out_name)
if os.path.exists(out_path):
return f"Skipping {out_name}, already exists."
html_url = f"https://harvard-edge.github.io/cs249r_book_dev/{vol}/contents/{rel_path.replace('.qmd', '.html')}"
local_images = []
try:
req = urllib.request.Request(html_url, headers={'User-Agent': 'Mozilla/5.0'})
html_content = urllib.request.urlopen(req).read().decode('utf-8')
soup = BeautifulSoup(html_content, 'html.parser')
for figure in soup.find_all('figure'):
fig_id = figure.get('id', 'unknown_fig')
img_tag = figure.find('img')
svg_tag = figure.find('svg')
if img_tag and img_tag.get('src'):
src = img_tag['src']
abs_url = urllib.parse.urljoin(html_url, src)
ext = os.path.splitext(urllib.parse.urlparse(abs_url).path)[1] or '.png'
local_img_path = os.path.join(img_tmp_dir, f"{vol}_{chap}_{fig_id}{ext}")
if download_file(abs_url, local_img_path):
local_images.append(f"{fig_id}: {os.path.abspath(local_img_path)}")
elif svg_tag:
local_img_path = os.path.join(img_tmp_dir, f"{vol}_{chap}_{fig_id}.svg")
with open(local_img_path, 'w', encoding='utf-8') as f:
f.write(str(svg_tag))
local_images.append(f"{fig_id}: {os.path.abspath(local_img_path)}")
except Exception as e:
print(f"Error parsing HTML for {chap}: {e}")
images_instruction = "\\n".join(local_images)
prompt = f"""You are the Three-Artifact Figure Audit agent. Read the brief at: {os.path.abspath(brief_path)}
Your task is to audit the chapter located at: {os.path.abspath(qmd_path)}
Instructions:
1. I have downloaded the raw rendered HTML images to your local disk so you can visually audit them.
Use the `read_file` tool on the following local paths to view the visual figures:
{images_instruction}
2. Compare the image you see via `read_file` to the prose, `fig-cap`, and `fig-alt` in the source QMD file at {os.path.abspath(qmd_path)}.
3. Output your findings strictly in the YAML format specified in the brief. Write the final YAML report to: {os.path.abspath(out_path)}
CRITICAL CONSTRAINTS FOR FIXES:
- **You CANNOT change the figures/images themselves.** The images are immutable.
- Your `proposed_fix` in the YAML MUST ONLY target the prose, the caption (`fig-cap`), or the alt-text (`fig-alt`) in the source `.qmd` file.
- **DO NOT rewrite the caption or prose completely.** You MUST propose the most MINIMAL, surgical tweaks to the existing `.qmd` text so that it accurately aligns with what is actually shown in the immutable image.
Do not ask for permission. Act autonomously to complete the audit and write the file."""
print(f"Starting {out_name}...")
cmd = ["gemini", "-m", "gemini-3.1-pro-preview", "-y", "-p", prompt]
try:
result = subprocess.run(cmd, capture_output=True, text=True, check=True)
return f"Success: {out_name}"
except subprocess.CalledProcessError as e:
error_msg = e.stderr if e.stderr else e.stdout
return f"Failed: {out_name}\nError: {error_msg[-500:]}"
with ThreadPoolExecutor(max_workers=5) as executor:
results = list(executor.map(process_chapter, chapters_with_figures))
print("\\n--- Audit Run Complete ---")

76
tinytorch/paper/references.bib
View File

@@ -11,7 +11,7 @@
}
@book{aho2006compilers,
title = {Compilers: Principles, Techniques, and Tools},
title = {Compilers: {Principles}, Techniques, and Tools},
author = {Aho, Alfred V. and Lam, Monica S. and Sethi, Ravi and Ullman, Jeffrey D.},
year = {2006},
publisher = {Addison-Wesley},
@@ -36,7 +36,7 @@
}
@article{banbury2021benchmarking,
title = {Benchmarking TinyML Systems: Challenges and Direction},
title = {Benchmarking TinyML Systems: {Challenges} and Direction},
author = {
Banbury, Colby R. and Reddi, Vijay Janapa and Lam, Max and Fu, William and Fazel, Amin and
Holleman, Jeremy and Huang, Xinyuan and Hurtado, Robert and Kanter, David and Lokhmotov, Anton
@@ -74,7 +74,7 @@
}
@article{blank2019nbgrader,
title = {nbgrader: A Tool for Creating and Grading Assignments in the Jupyter Notebook},
title = {nbgrader: {A} Tool for Creating and Grading Assignments in the Jupyter Notebook},
author = {
{Project Jupyter} and Blank, Douglas and Bourgin, David and Brown, Alexander and Bussonnier,
Matthias and Frederic, Jonathan and Granger, Brian and Griffiths, Thomas and Hamrick, Jessica
@@ -93,7 +93,7 @@
}
@misc{bradbury2018jax,
title = {JAX: composable transformations of Python+NumPy programs},
title = {JAX: {composable} transformations of Python+NumPy programs},
author = {
Bradbury, James and Frostig, Roy and Hawkins, Peter and Johnson, Matthew James and Leary, Chris
and Maclaurin, Dougal and Necula, George and Paszke, Adam and VanderPlas, Jake and
@@ -131,7 +131,7 @@
}
@misc{chen2022dlsyscourse,
title = {CS 10-414/614: Deep Learning Systems},
title = {CS 10-414/614: {Deep} Learning Systems},
author = {Chen, Tianqi and Zheng, Zico},
year = {2022},
publisher = {Carnegie Mellon University},
@@ -146,7 +146,7 @@
author = {Christopher, Wayne A. and Procter, Steven J. and Anderson, Thomas E.},
year = {1993},
journal = {USENIX Winter},
booktitle = {Proceedings of the USENIX Winter 1993 Conference},
booktitle = {Proceedings of the {USENIX} Winter 1993 Conference},
publisher = {USENIX Association},
pages = {481--488},
url = {
@@ -159,7 +159,7 @@
}
@incollection{collins1989cognitive,
title = {Cognitive Apprenticeship: Teaching the Crafts of Reading, Writing, and Mathematics},
title = {Cognitive Apprenticeship: {Teaching} the Crafts of Reading, Writing, and Mathematics},
author = {Collins, Allan and Brown, John Seely and Newman, Susan E.},
year = {2018},
booktitle = {Knowing, Learning, and Instruction},
@@ -176,7 +176,7 @@
}
@article{dao2022flashattention,
title = {FlashAttention: Fast and memory-efficient exact attention with IO-awareness},
title = {{FlashAttention}: {Fast} and memory-efficient exact attention with IO-awareness},
author = {Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{\'e}, Christopher},
year = {2022},
journal = {Advances in Neural Information Processing Systems},
@@ -219,7 +219,7 @@
}
@misc{hotz2023tinygrad,
title = {tinygrad: A simple and hackable neural network library},
title = {tinygrad: {A} simple and hackable neural network library},
author = {Hotz, George},
year = {2023},
publisher = {GitHub},
@@ -231,7 +231,7 @@
}
@article{howard2020fastai,
title = {Fastai: A Layered API for Deep Learning},
title = {Fastai: {A} Layered API for Deep Learning},
author = {Howard, Jeremy and Gugger, Sylvain},
year = {2020},
journal = {Information},
@@ -264,7 +264,7 @@
}
@misc{johnson2016cs231n,
title = {CS231n: Convolutional Neural Networks for Visual Recognition},
title = {CS231n: {Convolutional} Neural Networks for Visual Recognition},
author = {Johnson, Justin and Karpathy, Andrej and Fei-Fei, Li},
year = {2016},
publisher = {Stanford University},
@@ -275,7 +275,7 @@
}
@book{kaashoek2023xv6,
title = {xv6: a simple, Unix-like teaching operating system},
title = {xv6: {a} simple, Unix-like teaching operating system},
author = {Kaashoek, M. Frans and Morris, Robert and Cox, Russ},
year = {2023},
publisher = {MIT PDOS},
@@ -288,7 +288,7 @@
@inproceedings{kannan2022astrasim,
title = {
ASTRA-sim2.0: Modeling Hierarchical Networks and Disaggregated Systems for Large-model Training
ASTRA-sim2.0: {Modeling} Hierarchical Networks and Disaggregated Systems for Large-model Training
at Scale
},
author = {
@@ -297,7 +297,7 @@
},
year = {2023},
booktitle = {
Proceedings of the 2023 IEEE International Symposium on Performance Analysis of Systems and
Proceedings of the 2023 {IEEE} International Symposium on Performance Analysis of Systems and
Software (ISPASS)
},
publisher = {IEEE},
@@ -329,7 +329,7 @@
}
@misc{karpathy2022micrograd,
title = {micrograd: A tiny scalar-valued autograd engine and neural net library},
title = {micrograd: {A} tiny scalar-valued autograd engine and neural net library},
author = {Karpathy, Andrej},
year = {2022},
publisher = {GitHub},
@@ -340,7 +340,7 @@
}
@misc{keller2025ai,
title = {{AI} \& Machine-Learning Talent Gap 2025},
title = {{AI}} \& Machine-Learning Talent Gap 2025},
author = {{Keller Executive Search}},
year = {2025},
url = {https://www.kellerexecutivesearch.com/intelligence/ai-machine-learning-talent-gap-2025/},
@@ -350,7 +350,7 @@
}
@article{kingma2014adam,
title = {Adam: A Method for Stochastic Optimization},
title = {Adam: {A} Method for Stochastic Optimization},
author = {Kingma, Diederik P. and Ba, Jimmy},
year = {2014},
journal = {arXiv preprint arXiv:1412.6980},
@@ -435,7 +435,7 @@
Yamazaki, Masafumi and Young, Cliff and Zaharia, Matei
},
year = {2020},
booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
booktitle = {Proceedings of Machine Learning and Systems ({MLSys})},
publisher = {mlsys.org},
x-verified = {2026-04-08},
x-verified-by = {pass-16-bib-sweep},
@@ -443,7 +443,7 @@
}
@book{meadows2008thinking,
title = {Thinking in Systems: A Primer},
title = {Thinking in Systems: {A} Primer},
author = {Meadows, Donella H.},
year = {2008},
publisher = {Chelsea Green Publishing},
@@ -454,12 +454,12 @@
@incollection{meyer2003threshold,
title = {
Threshold concepts and troublesome knowledge: Linkages to ways of thinking and practising
Threshold concepts and troublesome knowledge: {Linkages} to ways of thinking and practising
within the disciplines
},
author = {Meyer, Jan H. F. and Land, Ray},
year = {2003},
booktitle = {Improving Student Learning: Theory and Practice Ten Years On},
booktitle = {Improving Student Learning: {Theory} and Practice Ten Years On},
publisher = {Oxford Centre for Staff and Learning Development},
pages = {412--424},
editor = {Rust, C.},
@@ -484,7 +484,7 @@
}
@book{mlsysbook2025,
title = {Machine Learning Systems: Design and Implementation},
title = {Machine Learning Systems: {Design} and Implementation},
author = {Janapa Reddi, Vijay},
year = {2025},
publisher = {MIT Press},
@@ -496,7 +496,7 @@
}
@misc{mlsysim2025,
title = {{MLSys$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
title = {{MLSys}$\cdot$im}: First-Principles Infrastructure Modeling for Machine Learning Systems},
author = {Reddi, Vijay Janapa},
year = {2026},
note = {Companion modeling framework for the Machine Learning Systems textbook},
@@ -506,7 +506,7 @@
}
@book{papert1980mindstorms,
title = {Mindstorms: Children, Computers, and Powerful Ideas},
title = {Mindstorms: {Children}, Computers, and Powerful Ideas},
author = {Papert, Seymour},
year = {1980},
publisher = {Basic Books},
@@ -517,14 +517,14 @@
}
@inproceedings{parashar2019timeloop,
title = {Timeloop: A Systematic Approach to DNN Accelerator Evaluation},
title = {Timeloop: {A} Systematic Approach to DNN Accelerator Evaluation},
author = {
Parashar, Angshuman and Raina, Priyanka and Shao, Yakun Sophia and Chen, Yu-Hsin and Ying,
Victor A. and Mukkara, Anurag and Venkatesan, Rangharajan and Khailany, Brucek and Keckler,
Stephen W. and Emer, Joel
},
year = {2019},
booktitle = {2019 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)},
booktitle = {2019 {IEEE} International Symposium on Performance Analysis of Systems and Software (ISPASS)},
publisher = {IEEE},
pages = {304--315},
doi = {10.1109/ispass.2019.00042},
@@ -605,7 +605,7 @@
}
@misc{pytorch04release,
title = {{PyTorch} 0.4.0 Release Notes: Tensor and Variable Merge},
title = {{PyTorch}} 0.4.0 Release Notes: Tensor and Variable Merge},
author = {{PyTorch Team}},
year = {2018},
url = {https://github.com/pytorch/pytorch/releases/tag/v0.4.0},
@@ -639,7 +639,7 @@
}
@inproceedings{reddi2024mlsysbook,
title = {MLSysBook.AI: Principles and Practices of Machine Learning Systems Engineering},
title = {MLSysBook.{AI}: {Principles} and Practices of Machine Learning Systems Engineering},
author = {Reddi, Vijay Janapa},
year = {2024},
booktitle = {2024 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)},
@@ -655,7 +655,7 @@
@misc{roberthalf2024talent,
title = {
Building Future-Forward Tech Teams: New Research Reveals Severity of the Technology Skills Gap
Building Future-Forward Tech Teams: {New} Research Reveals Severity of the Technology Skills Gap
Amid Talent Shortage
},
author = {{Robert Half}},
@@ -672,7 +672,7 @@
}
@article{rosenblatt1958perceptron,
title = {The perceptron: A probabilistic model for information storage and organization in the brain},
title = {The perceptron: {A} probabilistic model for information storage and organization in the brain},
author = {Rosenblatt, F.},
year = {1958},
journal = {Psychological Review},
@@ -708,7 +708,7 @@
}
@inproceedings{samajdar2018scale,
title = {SCALE-Sim: Systolic CNN Accelerator Simulator},
title = {SCALE-Sim: {Systolic} {CNN} Accelerator Simulator},
author = {Samajdar, Ananda and Zhu, Yuhao and Whatmough, Paul and Mattina, Matthew and Krishna, Tushar},
year = {2018},
booktitle = {arXiv preprint arXiv:1811.02883},
@@ -719,7 +719,7 @@
}
@misc{schneider2020minitorch,
title = {MiniTorch: A DIY Teaching Library for Machine Learning Engineers},
title = {MiniTorch: {A} DIY Teaching Library for Machine Learning Engineers},
author = {Rush, Sasha},
year = {2020},
publisher = {Cornell Tech},
@@ -730,7 +730,7 @@
}
@techreport{sei2020aieng,
title = {AI Engineering for Defense and National Security},
title = {{AI} Engineering for Defense and National Security},
author = {{Software Engineering Institute}},
year = {2020},
url = {https://insights.sei.cmu.edu/library/ai-engineering-for-defense-and-national-security/},
@@ -767,7 +767,7 @@
}
@article{sweller1988cognitive,
title = {Cognitive Load During Problem Solving: Effects on Learning},
title = {Cognitive Load During Problem Solving: {Effects} on Learning},
author = {Sweller, John},
year = {1988},
journal = {Cognitive Science},
@@ -785,7 +785,7 @@
}
@book{tanenbaum1987minix,
title = {Operating Systems: Design and Implementation},
title = {Operating Systems: {Design} and Implementation},
author = {Tanenbaum, Andrew S.},
year = {1987},
publisher = {Prentice-Hall},
@@ -797,7 +797,7 @@
}
@misc{tensorflow20,
title = {{TensorFlow} 2.0: Easy model building with Keras and eager execution},
title = {{TensorFlow}} 2.0: Easy model building with Keras and eager execution},
author = {{TensorFlow Team}},
year = {2019},
url = {https://www.tensorflow.org/guide/effective_tf2},
@@ -839,7 +839,7 @@
}
@book{vygotsky1978mind,
title = {Mind in Society: The Development of Higher Psychological Processes},
title = {Mind in Society: {The} Development of Higher Psychological Processes},
author = {Vygotsky, Lev S.},
year = {1978},
publisher = {Harvard University Press},
@@ -851,7 +851,7 @@
}
@article{williams2009roofline,
title = {Roofline: An Insightful Visual Performance Model for Multicore Architectures},
title = {Roofline: {An} Insightful Visual Performance Model for Multicore Architectures},
author = {Williams, Samuel and Waterman, Andrew and Patterson, David},
year = {2009},
journal = {Communications of the ACM},