Tulip, the Frontline Operations Platform, is empowering the world’s leading manufacturers to improve the productivity of their teams, the quality of their output, and the efficiency of their operations

Tulip Interfaces

ToolTalk: Benchmarking Tool-Augmented LLMs in Conversational AI

Authors: (1) Savvas Petridis, Google Research, New York, New York, USA; (2) Ben Wedin, Google Research, Cambridge, Massachusetts, USA; (3) James Wexler, Google Research, Cambridge, Massachusetts, USA; (4) Aaron Donsbach, Google Research, Seattle, Washington, USA; (5) Mahima Pushkarna, Google Research, Cambridge, Massachusetts, USA; (6) Nitesh Goyal, Google Research, New York, New York, USA; (7) Carrie J. Cai, Google Research, Mountain View, California, USA; (8) Michael Terry, Google Research, Cambridge, Massachusetts, USA. Table Of Links Abstract & Introduction Related Work Formative Study Constitution Maker Implementation User Study Findings Discussion Conclusion and References 9 CONCLUSION This paper presents ConstitutionMaker, a tool for interactively refining LLM outputs by converting users’ intuitive feedback into principles. ConstitutionMaker’s design is informed by a formative study, where we also collected and classified the types of principles users wanted to write. ConstitutionMaker incorporates three principle-elicitation features: kudos, critique, and rewrite. In a user study with 14 industry professionals, participants felt that ConstitutionMaker helped them (1) write principles that effectively guided the chatbot, (2) convert their feedback into principles more easily, and (3) write principles more efficiently, with (4) less mental demand than the baseline. This was due to ConstitutionMaker supporting their thought processes, including helping them to: identify ways to improve the bot’s responses, convert their intuition into verbal feedback, and phrase their feedback as specific principles. There are many avenues of future work, including supporting users in iterating on and clarifying their principles, organizing larger sets of principles and supporting multiple writers, and helping users test chatbots across multiple user journeys. Together, these findings inform future tools that support interactively customizing LLM outputs. REFERENCES [1] Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, Carol Chen, Catherine Olsson, Christopher Olah, Danny Hernandez, Dawn Drain, Deep Ganguli, Dustin Li, Eli Tran-Johnson, Ethan Perez, Jamie Kerr, Jared Mueller, Jeffrey Ladish, Joshua Landau, Kamal Ndousse, Kamile Lukosuite, Liane Lovitt, Michael Sellitto, Nelson Elhage, Nicholas Schiefer, Noemi Mercado, Nova DasSarma, Robert Lasenby, Robin Larson, Sam Ringer, Scott Johnston, Shauna Kravec, Sheer El Showk, Stanislav Fort, Tamera Lanham, Timothy TelleenLawton, Tom Conerly, Tom Henighan, Tristan Hume, Samuel R. Bowman, Zac Hatfield-Dodds, Ben Mann, Dario Amodei, Nicholas Joseph, Sam McCandlish, Tom Brown, and Jared Kaplan. 2022. Constitutional AI: Harmlessness from AI Feedback. arXiv:2212.08073 [cs.CL] [2] Svetlin Bostandjiev, John O’Donovan, and Tobias Höllerer. 2012. TasteWeights: A Visual Interactive Hybrid Recommender System. In Proceedings of the Sixth ACM Conference on Recommender Systems (Dublin, Ireland) (RecSys ’12). Association for Computing Machinery, New York, NY, USA, 35–42. https://doi.org/10.1145/ 2365952.2365964 [3] Stephen Brade, Bryan Wang, Mauricio Sousa, Sageev Oore, and Tovi Grossman. 2023. Promptify: Text-to-Image Generation through Interactive Prompt Exploration with Large Language Models. In Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology (UIST). https: //doi.org/10.1145/3586183.3606725 arXiv:2304.09337 [cs.HC] [4] Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language Models are Few-Shot Learners. In Advances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin (Eds.), Vol. 33. Curran Associates, Inc., 1877–1901. https://proceedings.neurips.cc/paper_files/paper/2020/file/ 1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf [5] Sébastien Bubeck, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, Harsha Nori, Hamid Palangi, Marco Tulio Ribeiro, and Yi Zhang. 2023. Sparks of Artificial General Intelligence: Early experiments with GPT-4. arXiv:2303.12712 [cs.CL] [6] Ernie Chang, Xiaoyu Shen, Hui-Syuan Yeh, and Vera Demberg. 2021. On Training Instance Selection for Few-Shot Neural Text Generation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers). Association for Computational Linguistics, Online, 8–13. https: //doi.org/10.18653/v1/2021.acl-short.2 [7] Qiaochu Chen, Xinyu Wang, Xi Ye, Greg Durrett, and Isil Dillig. 2020. MultiModal Synthesis of Regular Expressions. In Proceedings of the 41st ACM SIGPLAN Conference on Programming Language Design and Implementation (London, UK) (PLDI 2020). Association for Computing Machinery, New York, NY, USA, 487–502. https://doi.org/10.1145/3385412.3385988 [8] Katy Ilonka Gero, Vivian Liu, and Lydia Chilton. 2022. Sparks: Inspiration for Science Writing Using Language Models. In Proceedings of the 2022 ACM Designing Interactive Systems Conference (Virtual Event, Australia) (DIS ’22). Association for Computing Machinery, New York, NY, USA, 1002–1019. https: //doi.org/10.1145/3532106.3533533 [9] Sandra G. Hart and Lowell E. Staveland. 1988. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. In Human Mental Workload, Peter A. Hancock and Najmedin Meshkati (Eds.). Advances in Psychology, Vol. 52. North-Holland, 139–183. https://doi.org/10.1016/S0166- 4115(08)62386-9 [10] Jeremy Howard and Sebastian Ruder. 2018. Universal Language Model Finetuning for Text Classification. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Association for Computational Linguistics, Melbourne, Australia, 328–339. https://doi.org/ 10.18653/v1/P18-1031 [11] Jiyou Jia. 2009. CSIEC: A computer assisted English learning chatbot based on textual knowledge and reasoning. Knowledge-Based Systems 22, 4 (2009), 249–255. https://doi.org/10.1016/j.knosys.2008.09.001 Artificial Intelligence (AI) in Blended Learning. [12] Ellen Jiang, Kristen Olson, Edwin Toh, Alejandra Molina, Aaron Donsbach, Michael Terry, and Carrie J Cai. 2022. PromptMaker: Prompt-Based Prototyping with Large Language Models. In Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems (New Orleans, LA, USA) (CHI EA ’22). Association for Computing Machinery, New York, NY, USA, Article 35, 8 pages. https://doi.org/10.1145/3491101.3503564 [13] Ellen Jiang, Edwin Toh, Alejandra Molina, Aaron Donsbach, Carrie J Cai, and Michael Terry. 2021. GenLine and GenForm: Two Tools for Interacting with Generative Language Models in a Code Editor. In Adjunct Proceedings of the 34th Annual ACM Symposium on User Interface Software and Technology (Virtual Event, USA) (UIST ’21 Adjunct). Association for Computing Machinery, New York, NY, USA, 145–147. https://doi.org/10.1145/3474349.3480209 [14] Ellen Jiang, Edwin Toh, Alejandra Molina, Kristen Olson, Claire Kayacik, Aaron Donsbach, Carrie J Cai, and Michael Terry. 2022. Discovering the Syntax and Strategies of Natural Language Programming with Generative Language Models. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems (New Orleans, LA, USA) (CHI ’22). Association for Computing Machinery, New York, NY, USA, Article 386, 19 pages. https://doi.org/10.1145/3491102.3501870 [15] Daniel Kahneman. 2011. Thinking, fast and slow. macmillan. [16] Tae Soo Kim, Yoonjoo Lee, Minsuk Chang, and Juho Kim. 2023. Cells, Generators, and Lenses: Design Framework for Object-Oriented Interaction with Large Language Models. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. ACM. 1–18. [17] Johannes Kunkel, Benedikt Loepp, and Jürgen Ziegler. 2017. A 3D Item Space Visualization for Presenting and Manipulating User Preferences in Collaborative Filtering. In Proceedings of the 22nd International Conference on Intelligent User Interfaces (Limassol, Cyprus) (IUI ’17). Association for Computing Machinery, New York, NY, USA, 3–15. https://doi.org/10.1145/3025171.3025189 [18] Peter Lee, Sebastien Bubeck, and Joseph Petro. 2023. Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine. New England Journal of Medicine 388, 13 (2023), 1233–1239. https://doi.org/10.1056/NEJMsr2214184 arXiv:https://doi.org/10.1056/NEJMsr2214184 PMID: 36988602. [19] Brian Lester, Rami Al-Rfou, and Noah Constant. 2021. The Power of Scale for Parameter-Efficient Prompt Tuning. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, Online and Punta Cana, Dominican Republic, 3045–3059. https: //doi.org/10.18653/v1/2021.emnlp-main.243 [20] Michael Xieyang Liu, Advait Sarkar, Carina Negreanu, Benjamin Zorn, Jack Williams, Neil Toronto, and Andrew D. Gordon. 2023. “What It Wants Me To Say”: Bridging the Abstraction Gap Between End-User Programmers and CodeGenerating Large Language Models. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 598, 31 pages. https: //doi.org/10.1145/3544548.3580817 [21] Vivian Liu, Han Qiao, and Lydia Chilton. 2022. Opal: Multimodal Image Generation for News Illustration. In Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology (Bend, OR, USA) (UIST ’22). Association for Computing Machinery, New York, NY, USA, Article 73, 17 pages. https://doi.org/10.1145/3526113.3545621 [22] Vivian Liu, Jo Vermeulen, George Fitzmaurice, and Justin Matejka. 2023. 3DALLE: Integrating Text-to-Image AI in 3D Design Workflows. In Proceedings of the 2023 ACM Designing Interactive Systems Conference (Pittsburgh, PA, USA) (DIS ’23). Association for Computing Machinery, New York, NY, USA, 1955–1977. https://doi.org/10.1145/3563657.3596098 [23] Aditi Mishra, Utkarsh Soni, Anjana Arunkumar, Jinbin Huang, Bum Chul Kwon, and Chris Bryan. 2023. PromptAid: Prompt Exploration, Perturbation, Testing and Iteration using Visual Analytics for Large Language Models. arXiv:2304.01964 [cs.HC] [24] Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. 2023. Generative Agents: Interactive Simulacra of Human Behavior. arXiv:2304.03442 [cs.HC] [25] Savvas Petridis, Nediyana Daskalova, Sarah Mennicken, Samuel F Way, Paul Lamere, and Jennifer Thom. 2022. TastePaths: Enabling Deeper Exploration and Understanding of Personal Preferences in Recommender Systems. In 27th International Conference on Intelligent User Interfaces (Helsinki, Finland) (IUI ’22). Association for Computing Machinery, New York, NY, USA, 120–133. https: //doi.org/10.1145/3490099.3511156 [26] Savvas Petridis, Nicholas Diakopoulos, Kevin Crowston, Mark Hansen, Keren Henderson, Stan Jastrzebski, Jeffrey V Nickerson, and Lydia B Chilton. 2023. AngleKindling: Supporting Journalistic Angle Ideation with Large Language Models. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 225, 16 pages. https://doi.org/10.1145/3544548. 3580907 [27] Reid Pryzant, Dan Iter, Jerry Li, Yin Tat Lee, Chenguang Zhu, and Michael Zeng. 2023. Automatic Prompt Optimization with "Gradient Descent" and Beam Search. arXiv:2305.03495 [cs.CL] [28] Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. Language models are unsupervised multitask learners. OpenAI blog 1, 8 (2019), 9. [29] Kiran Ramesh, Surya Ravishankaran, Abhishek Joshi, and K. Chandrasekaran. 2017. A Survey of Design Techniques for Conversational Agents. In Information, Communication and Computing Technology, Saroj Kaushik, Daya Gupta, Latika Kharb, and Deepak Chahal (Eds.). Springer Singapore, Singapore, 336–350. [30] Emily Reif, Minsuk Kahng, and Savvas Petridis. 2023. Visualizing Linguistic Diversity of Text Datasets Synthesized by Large Language Models. arXiv:2305.11364 [cs.CL] [31] Laria Reynolds and Kyle McDonell. 2021. Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm. In Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems (Yokohama, Japan) (CHI EA ’21). Association for Computing Machinery, New York, NY, USA, Article 314, 7 pages. https://doi.org/10.1145/3411763.3451760 [32] Taylor Shin, Yasaman Razeghi, Robert L. Logan IV, Eric Wallace, and Sameer Singh. 2020. AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). Association for Computational Linguistics, Online, 4222–4235. https://doi.org/10.18653/v1/2020. emnlp-main.346 [33] Hendrik Strobelt, Albert Webson, Victor Sanh, Benjamin Hoover, Johanna Beyer, Hanspeter Pfister, and Alexander M. Rush. 2023. Interactive and Visual Prompt Engineering for Ad-hoc Task Adaptation with Large Language Models. IEEE Transactions on Visualization and Computer Graphics 29, 1 (2023), 1146–1156. https://doi.org/10.1109/TVCG.2022.3209479 [34] Hongjin Su, Jungo Kasai, Chen Henry Wu, Weijia Shi, Tianlu Wang, Jiayi Xin, Rui Zhang, Mari Ostendorf, Luke Zettlemoyer, Noah A. Smith, and Tao Yu. 2022. Selective Annotation Makes Language Models Better Few-Shot Learners. arXiv:2209.01975 [cs.CL] [35] Gust Verbruggen, Vu Le, and Sumit Gulwani. 2021. Semantic Programming by Example with Pre-Trained Models. Proc. ACM Program. Lang. 5, OOPSLA, Article 100 (oct 2021), 25 pages. https://doi.org/10.1145/3485477 [36] Sitong Wang, Savvas Petridis, Taeahn Kwon, Xiaojuan Ma, and Lydia B Chilton. 2023. PopBlends: Strategies for Conceptual Blending with Large Language Models. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 435, 19 pages. https://doi.org/10.1145/3544548. [37] Yunlong Wang, Shuyuan Shen, and Brian Y Lim. 2023. RePrompt: Automatic Prompt Editing to Refine AI-Generative Art Towards Precise Expressions. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 22, 29 pages. https://doi.org/10.1145/3544548.3581402 [38] Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed Chi, Quoc V Le, and Denny Zhou. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. In Advances in Neural Information Processing Systems, S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35. Curran Associates, Inc., 24824–24837. https://proceedings.neurips.cc/paper_files/paper/2022/file/ 9d5609613524ecf4f15af0f7b31abca4-Paper-Conference.pdf [39] Sherry Wu, Hua Shen, Daniel S Weld, Jeffrey Heer, and Marco Tulio Ribeiro. 2023. ScatterShot: Interactive In-Context Example Curation for Text Transformation. In Proceedings of the 28th International Conference on Intelligent User Interfaces (Sydney, NSW, Australia) (IUI ’23). Association for Computing Machinery, New York, NY, USA, 353–367. https://doi.org/10.1145/3581641.3584059 [40] Tongshuang Wu, Ellen Jiang, Aaron Donsbach, Jeff Gray, Alejandra Molina, Michael Terry, and Carrie J Cai. 2022. PromptChainer: Chaining Large Language Model Prompts through Visual Programming. In Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems (New Orleans, LA, USA) (CHI EA ’22). Association for Computing Machinery, New York, NY, USA, Article 359, 10 pages. https://doi.org/10.1145/3491101.3519729 [41] Tongshuang Wu, Michael Terry, and Carrie Jun Cai. 2022. AI Chains: Transparent and Controllable Human-AI Interaction by Chaining Large Language Model Prompts. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems (New Orleans, LA, USA) (CHI ’22). Association for Computing Machinery, New York, NY, USA, Article 385, 22 pages. https: //doi.org/10.1145/3491102.3517582 [42] Yu Wu, Wei Wu, Chen Xing, Ming Zhou, and Zhoujun Li. 2017. Sequential Matching Network: A New Architecture for Multi-turn Response Selection in RetrievalBased Chatbots. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Association for Computational Linguistics, Vancouver, Canada, 496–505. https://doi.org/10.18653/v1/P17-1046 [43] Anbang Xu, Zhe Liu, Yufan Guo, Vibha Sinha, and Rama Akkiraju. 2017. A New Chatbot for Customer Service on Social Media. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (Denver, Colorado, USA) (CHI ’17). Association for Computing Machinery, New York, NY, USA, 3506–3510. https://doi.org/10.1145/3025453.3025496 [44] Ann Yuan, Andy Coenen, Emily Reif, and Daphne Ippolito. 2022. Wordcraft: Story Writing With Large Language Models. In 27th International Conference on Intelligent User Interfaces (Helsinki, Finland) (IUI ’22). Association for Computing Machinery, New York, NY, USA, 841–852. https://doi.org/10.1145/3490099.3511105 [45] J.D. Zamfirescu-Pereira, Heather Wei, Amy Xiao, Kitty Gu, Grace Jung, Matthew G Lee, Bjoern Hartmann, and Qian Yang. 2023. Herding AI Cats: Lessons from Designing a Chatbot by Prompting GPT-3. In Proceedings of the 2023 ACM Designing Interactive Systems Conference (Pittsburgh, PA, USA) (DIS ’23). Association for Computing Machinery, New York, NY, USA, 2206–2220. https://doi.org/10.1145/3563657.3596138 [46] J.D. Zamfirescu-Pereira, Richmond Y. Wong, Bjoern Hartmann, and Qian Yang. 2023. Why Johnny Can’t Prompt: How Non-AI Experts Try (and Fail) to Design LLM Prompts. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI ’23). Association for Computing Machinery, New York, NY, USA, Article 437, 21 pages. https://doi.org/10.1145/ 3544548.3581388 [47] Tianyi Zhang, London Lowmanstone, Xinyu Wang, and Elena L. Glassman. 2020. Interactive Program Synthesis by Augmented Examples. In Proceedings of the 33rd Annual ACM Symposium on User Interface Software and Technology (Virtual Event, USA) (UIST ’20). Association for Computing Machinery, New York, NY, USA, 627–648. https://doi.org/10.1145/3379337.3415900 [48] Hao Zhou, Minlie Huang, Tianyang Zhang, Xiaoyan Zhu, and Bing Liu. 2018. Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory. arXiv:1704.01074 [cs.CL] This paper is under CC 4.0 license. available on arxiv

Interactively Critiquing LLMs by Converting Feedback into Principle— Conclusion and References

This piece was written August 30th, 2011 . Today, people don’t point to a single exchange to say that Bitcoin has “failed” or has problems; they use the price across the many exchanges to make this absurd and bogus claim. They were wrong to conflate Bitcoin with MTGox in 2011 and they are wrong to conflate Bitcoin with groups of exchanges today. Because Bitcoin is so new to most people it is hard to think about it correctly. The vast majority experience great discomfort when exposed to the idea that money doesn't need to come from the State, and become even more disturbed when they’re told that no one controls Bitcoin. These are ideas that the “normies” are going to get used to, just as they are used to chatting on WhatsApp; a very novel and 21st Century activity. An essential feature of the standard attack against Bitcoin is to point to the price charts generated by a single Bitcoin exchange and then use that as definitive proof of Bitcoin’s unsuitability for any purpose. The attack uses these single source charts to ‘prove’ that Bitcoin is a mania, like the Dutch Tulip mania or of 1637, or that Bitcoin is ‘insecure’ or any other fundamental flaw, technical, financial, philosophical or psychological you care to mention. Bollengekte Let us be perfectly clear; these Bitcoin detractors are of what Bitcoin is. They are near horizon thinkers, dullards, luddites, and the sweetest irony of it all is they are peddling their flawed ideas on a medium that directly disproves their theories. ignorant I have already debunked and quashed many of the fallacies that are routinely trotted out whenever Bitcoin is discussed by the ignorant on our blog; now I want to clear up a different fundamental mistake that all the current detractors of Bitcoin are routinely making, which is perfectly exemplified by the recent video item, which unintelligently parrots all the anti-Bitcoin nonsense as if it were being read from a centrally provided script or press release. MarketWatch The fundamental mistake these ignorant people are making is this; Bitcoin is not an investment, . it is a container and payment method When you think about Bitcoin in these terms, it becomes instantly clear that Bitcoin should not be treated like stocks or commodities. If you think of Bitcoin as only a container you use to shuttle payments to people for things on and off-line, you immediately understand that looking at stock market style charts of its value from a single exchange as a way of gauging its future potential is completely ridiculous. itself Bitcoin’s potential lies in its power to facilitate peer to peer purchases; it is not a commodity or a stock or a company, it is a , a , a that people use to make purchases between themselves. method container protocol Think about it this way; if, in 1997 you were told about a thing called ‘the Internet’, that would replace sending letters, utility bills and postcards through the mail to people all over the world at no cost, via a system that would not be run by any central authority and which was sure to utterly change the world and make people millions of dollars, you would be interested in it as an investment. Someone could (having fundamentally misunderstood what the Internet actually is) buy many domain names and then issue certificates against them, put these certificates into an exchange, and then start to sell them to investors. Charts would have been generated, and as a land rush began as the potential of the Internet became apparent to everybody, you would have seen a massive spike in the quoted prices of domain name certificates. Unique names like ‘sex.com’ could have been bought into by syndicates, who issued shares in it so that the cost of investing in ‘rare’ domain names could be spread out. You could buy shares in that domain name syndicate, and see their value rocket up. Are you beginning to get the picture? Domain names are nothing more than a method to instruct client programmes on computers to connect to a numeric address that refers to a server computer on the internet. They have no value in and of themselves; the value in a domain name rests solely in the work that programmers put into expressing the ideas of entrepreneurs who run the websites the domain points to. A three letter domain name like ‘sex.com’ is no guarantee of success on the web in and of itself; the same is true of Bitcoin. No one would have paid a billion dollars for the domain name ‘google.com’ before Google put millions of man hours and genius into their software, for example. If you want to ‘invest in the Internet’, you need to invest in a . You cannot invest directly in the Internet, which is nothing more than a series of protocols defining containers for information that have been agreed upon by individuals. When you think about Bitcoin in these terms, you start to understand why all these foolish pundits sound so ridiculous. They literally have no idea what they are talking about. company that uses the internet to provide value to people Bitcoin is a way to convey value from one person to another without a third party. Email is to postal services as Bitcoin is to money. It has no monetary value in and of itself; it has . This is why looking at a single chart from MTGox and inferring anything about Bitcoin in general, or its future, or its utility and true nature is completely absurd. This is why attempting to apply Austrian monetary theory to Bitcoin is a fool’s errand. Bitcoin is not money, any more than a leather wallet is money or an email is a letter written on your personal stationary; you would not define a wallet as money, or a domain name as money or a piece of paper with ink on it simply because someone buys and sells them as goods. a very high utility, not intrinsic value The real issue is not whether Bitcoin will ever be so widely adopted that it, “acts like a real, stable currency”. The only issue is wether or not it is widely adopted, and when the disruptive effects it will have on the current crop of online payment systems that are in thrall to the State, begin to emerge. And Bitcoin is a very disruptive technology. very Think about Bitcoin in comparison to PayPal. PayPal is essentially a centralized brick-less bank, that keeps a ledger of user’s accounts and transfers, and which charges per transaction fees. It strictly controls how much of your own money you can withdraw from them to your own bank account, how much of your own money you can spend at any one time, and PayPal are notorious for their freezing of user’s accounts, service problems and lust for compliance with the regulations of the State. For example, users of PayPal unfortunate enough to live under the yoke of the government of India have recently been informed that they will not be allowed to receive payments that exceed $500 per transaction and that they will not be able to keep any of their money in their PayPal accounts longer than one week; all money received into PayPal must be transferred to their Indian bank account within 7 days. I will take for granted your outrage at these anti-human and arbitrary restrictions. Now consider Bitcoin. Bitcoin turns every user into an operator of their own fully functional, trans continental, free of State control PayPal service. They can accept money and then transfer Bitcoins from their computer to anywhere in the world instantly, without interference from anyone. They can accept Bitcoins on their computer in exchange for goods or services in a similar manner. The key insight that mainstream thinking people are missing is that Bitcoin . Its accounts are essentially disposable and not tied to you permanently. You do not have to identify yourself to any third party in order to use it. If you adopt Bitcoin you are at liberty to use it in any way you like, with as much of your money as you like. can be exchanged for anything, not just money When you think about Bitcoin correctly, you can begin to see that its potential is as big as the advent of the internet itself, since money is half of all transactions. In the same way that email disrupted the postal service, Bitcoin will disrupt the making and receiving of payments. If you want to send a post card, you do not have to use a postman or government mail. You simply send an email. From your mobile phone. This is taken for granted, now, but it represents a tectonic shift in the way people communicate. Think about how the internet and Fraunhofer-Gesellshaft’s Perceptual Audio Coding software (that powers the MP3 file format) has changed the way music is distributed and consumed. No more buying Cassettes, Vinyl records and CDs in stores; no more middle men between the musicians and the music lovers. This is what Bitcoin is going to do in the realm of money transfer. And of course, the circle will be completed when music lovers pay tributes to their favourite musicians with Bitcoins. Very small payments will now become possible and plentiful…anyone can develop their own money transfer and content monetizing service on top of Bitcoin without having to interface with one of the main payment processing companies. This represents a massive shift and unprecedented opportunity on a global scale. There are so many possible uses of Bitcoin you could spend all day imagining its potential uses, and you might still completely miss its killer application. None of the people trying to pour cold water on Bitcoin ever mention Namecoin, which is a DNS alternative based on Bitcoin. This is probably because they are ignorant of what Bitcoin actually is, and are simply regurgitating what others have written and said about Bitcoin, rather than doing their own thinking about it. DNS, as I say above, is the system of marrying words with the numeric addresses of computers on the internet. It is how people connect to sites on the net with their browsers, allowing them to type in a name instead of a number. The DNS system is being attacked by the State as a way of taking publishers off-line. Google “ICE domain seizures” to find out what I am talking about. Namecoin has the potential to decentralise the DNS system, making it impossible for the State to seize domain names and attack publishers. This is only one possible future use of Bitcoin, and as we have seen with the appalling totalitarian police state scandals surrounding government sabotage and poisoning of the centralised DNS system, Namecoin could remove the power of the State to control this critical part of the Internet infrastructure. The potential of Bitcoin is obvious to those that are intelligent, that understand computers and software, who have some knowledge of the present state of and recent history of the internet and the problems of money transfer online. Anyone who knows what this really means is awestruck, at how everything is about to fundamentally change. gobsmacked To conclude, whenever you hear anyone attack Bitcoin, your first response should be to be skeptical of the intelligence and depth of understanding of the attacker. They will cite any or all of the following to try and dissuade you from adopting Bitcoin: Bitcoin has no backing The exchange rate is volatile (with obligatory MTGox chart) Bitcoin is a Speculative Bubble Bitcoin is used for buying drugs Bitcoin is run by amateurs ‘The MyBitcoin Fiasco’ Bitcoin is only for techies, not for the average person All of these reasons for avoiding Bitcoin are , trotted out by the unintelligent who cannot think for themselves, have weak powers of insight, are very probably computer illiterate, or who are philosophically predisposed to disliking Bitcoin because they have mistaken it for money due to other people having claimed that it is money. straw men The first and last straw men are particularly galling. The dollar is backed by nothing, and these same people insist that it is money simply because other people accept it as money, but by magic, this logic cannot simultaneously apply to Bitcoin. The Internet was once ‘only for techies’ and now everyone uses it, and the people who do not are the exception, the ‘disadvantaged’ who must be helped to get onto it. If it were not so tragic, you would think these pretexts for rejecting Bitcoin were funny. I predict that the same will be true of the mass adoption of Bitcoin as it was for the mass adoption of the Internet. In the very near future, the people who do not use Bitcoin for sending and receiving payments will be the exceptions, and the disadvantaged. I will leave it to you to extrapolate from , what the true value of Bitcoin is. that Roasted Ginan Nuts, Grilled Mackerel, Rice, Tskune, Sake, Shisomaki, 2 taxis ↴

Thinking Correctly About Bitcoin

This is the first in a series of posts identifying the sources of value in cryptocurrencies Image: Nikita Andreev Shell: 100000+ yrs Gold wearable: 3800 Gold & silver coin: 2400 Paper IOUs for gold & silver: 250 Pure fiat: 47 Cryptocurrency: avail. now- Nick Szabo The recent trend in coin offerings, including Tezos’ massive raise, has put a spotlight on . The valuation of the world’s leading crypto currency, exceeded US$ 100 billion recently. Are cryptocurrencies overvalued and due for a correction? Perhaps. cryptocurrencies Bitcoin In this series (1/3), I will attempt to identify sources of demand of cryptocurrencies and to develop a framework for looking at their value. The source of demand for cryptocurrencies is derived from their attributes as a ‘medium of exchange’ and that as a ‘store of value’. They are not yet useful as a unit of account but that could change. In this post, I will look at the value of cryptocurrencies as a medium of exchange, particularly, as an enabler of the and the implications of this on their value. internet of money The rise of the ‘internet of money’ The most fundamental use-case of cryptocurrencies is the role they are beginning to play as a medium of exchange on the internet. a. Peer to peer payments (VOIP:Talk :: Cryptocurrencies:Payments) Cryptocurrencies are creating opportunities for a massive scale-up in peer-to-peer payments. Firstly, they can reduce the cost of transactions. They can do this by removing intermediaries and developing a lower-cost consensus system to validate transactions. This is opposed to the centralized ‘trusted’ payment infrastructure which has always been necessary for transferring value. This layered interaction involving multiple intermediaries adds costs to the transfer of value. Source: Balaji Srinivasan/a16z Secondly, they can remove friction from transactions — creating opportunities for seam-less transfer of value across peers. Come to think of it, the original Paypal innovation of being able to was at its heart about making payments frictionless (and secure). email money In fact, the first ‘killer app’ for the mainstreaming of cryptocurrencies has already been international money transfers. It came to play in China when imposition of capital controls forced users to be creative and utilize crypto mediums to move money across borders. Napster? — the use of bitcoin to move money Dejavu Valuing Crypto’s value as a medium of exchange Given the above context, we can now try to put a number on the total value of cryptocurrency coming from its attribute as a medium of exchange on the internet. For this purpose, let’s look at the global money transfer volume and try to see where cryptocurrencies can come in to capture value. We see that just cross-border remittances were close to US$ 600 billion globally in 2014. Even if we assume no growth in the global remittances market and look at a fifth of global remittances being processed by a large cryptocurrency network in steady state, US$ 120 billion of value would be transmitted by the network on an annual basis. Today, the commission charged by the lowest priced intermediary is about 1%. If the cryptocurrency network is able to pocket even 0.1%-0.2%, the annual commission captured by the network would be US$ 120–240 million annually. This is just the global remittances market. Now, let’s look at the volume of global digital payments. According to Euromonitor, the total card transactions expected in 2017 are likely to exceed US$ 30 trillion. Even if 10% of these transactions can be processed on one crypto network, that would mean US$ 3 trillion of value transfer. Again, assuming 0.1%-0.2% of commission on this volume, this could easily mean US$ 3–6 billion of commission for the largest crypto-network annually. Source: Euromonitor Therefore, just looking at 20% of global remittance volume and 10% of global card transaction volume in 2017, a single cryptocurrency can reach ~ US$ 3–6 billion in commissions. Even if one believes that the cash-flow associated with this crypto-network needs to be subject to a very high discount rate (~25%) and a growth of 15%, one can arrive at a network valuation of US$ 45 billion just for a single use-case (assuming US$ 4.5 billion commission). No wonder that there is excitement about cryptocurrencies. Below, a chart by Chris Burniske, an expert in crypto-valuations, looks at the correlation between transactions on the Ethereum network and the corresponding network value. Very logically, there is a strong alignment in the growth of transactions and value. Source: Chris Burniske‏ Specialized crypto-currencies like zCash similarly might focus on money transfers for specific user groups or specific use cases and capture a portion of the value. : b. The evolving monetization model of the internet or the Machine Payable Web Cryptocurrencies as a remittance rail was the first “killer app”. However, in a , Balaji S. Srinivasan describes how the nature of monetization in the web is evolving. Today, the dominant monetization business model in the current consumer web is advertising. 88% of Google’s revenue and 97% of facebook’s revenue in 2016 came from advertising sales. Many of their APIs are public and monetization happens by exposing users (humans) to ads. recent talk However, the ad-based monetization model is not suited for application programming interfaces ( ). APIs can’t be compensated for “viewing ads”. Additionally, the current model disincentivizes commerce by adding friction (think of the onboarding and sign-up process) to(API) interactions. Cryptocurrencies have the potential to streamline commerce between APIs or indeed any agents. In a machine payable web, cryptocurrencies replace ads (or human attention) as the dominant currency of the web and allow API-to-API or indeed machine-to-machine transactions without human intervention based on contracts at a low cost. APIs What’s the volume of transactions which will happen due to API-to-API or machine-to-machine transactions? Estimating a defensible number requires another post but it will certainly involve many billions of transactions. Structurally, the architecture of such a dominant crypto-network in the space might be split with lower ticket-size transactions facilitated through a less computationally intensive process to ensure that even very small ticket transactions are encouraged. c. The Uberization of connected computational and storage assets The third use-case for cryptocurrencies as a medium of exchange is that which comes from monetization of computational and storage assets. While ‘uberization’ of assets is not a new phenomenon anymore, it is largely unexplored in the world of personal and even professional computing. Our computational and storage resources are quite underutilized. Currencies based on cryptographic proof offer a fascinating opportunity for unlocking under-utilized resources to work and earn an income out of it. This works on the ‘supply side’ and keeps a check on price by encouraging more and more users to jump on the supply side to provide mining and/or transaction processing. In the long run, one would expect that mining would be provided by a mix of super-specialized nodes and under-utilized assets. This helps build the supply side case for cryptocurrencies. A typical PC is using only a fraction of its computational resources while active Think of connected assets and specialized mining as roof-top solar and utility scale solar respectively. Utility scale solar is efficient and profitable on an ROI basis and continues to scale but roof-top solar is attractive for many users. The significance of this can’t be understated. We may be on the verge of a solar-type revolution in computing. As cryptocurrencies become more mainstream, many more truly decentralized computational and storage resources are likely to come online and compete with conventional cloud services. Hash rate of the Bitcoin network (source: blockchain.info) Given the excitement about bitcoins, it is not surprising that the hashrate (a measure of computational power in terms of calculations per second) of all cryptocurrency miners already exceeds the equivalent computational power of internet giants like Google. According to Bitcoinist, as of Jan ’17, the total computational capacity of the bitcoin network was over 43 thousand times than every supercomputer on the TOP500 list combined. Hence, we see that the use case for demand for cryptocurrencies as a medium of exchange itself is quite high and arguably explains why people are betting that individual cryptocurrencies could be worth as much as US$ 100 billion. Does that mean that the current values of all cryptocurrencies are justified? Not really. Instead, my point is to argue that the winners of the cryptocurrency play— the first currency (or currencies) to actually find product-market fit in some of the use-cases I mentioned (and others which I will cover in subsequent posts) would be big enough to be worth a lot. Maybe even GAFA big! This post was on cryptocurrencies as mediums of exchange. I will cover cryptocurrencies as foundations of decentralized applications and contracts in the next post. Thanks to Edward Russell and Shaina Trotta for reviewing the manuscript of this post. References 1. https://github.com/ethereum/wiki/wiki/White-Paper 2. https://bitcoin.org/bitcoin.pdf 3. https://stratechery.com/2017/tulips-myths-and-cryptocurrencies/ 4. http://offers.worldpayglobal.com/rs/850-JOA-856/images/GlobalPaymentsReportNov2015.pdf 5. https://medium.com/@cburniske/cryptoasset-valuations-ac83479ffca7

What’s the true value of cryptocurrencies?

What’s shaping our technology world GitHub is not just a code hosting service with version control — it’s also an enormous developer network. The sheer size of GitHub at over 30 million accounts, more than 2 million organizations, and over 96 million repositories translates into one of the world’s most valuable development networks. How do you quantify the value of this network? And is there a way to get the top repositories? Here at U°OS, we ran the GitHub network through a simplified version¹ of and produced the top 100 most valuable repositories. our reputation algorithm The result is as fascinating as it is eclectic in the way that it does feel like a good reflection of our society’s interest in the technology and where it moves. There are the big proprietary players with open source projects — Google, Apple, Microsoft, Facebook, and even Baidu. And at the same time, there’s a Chinese anti-censorship tool. There’s Bitcoin for cryptocurrency. There’s a particle detector for CERN’s Large Hadron Collider. There are gaming projects like Space Station 13 and Cataclysm: Dark Days Ahead and a gaming engine Godot. There are education projects like freeCodeCamp, Open edX, Oppia, and Code.org. There are web and mobile app building projects like WordPress, Joomla, and Flutter to publish your content on. There are databases to store your content for the web like Ceph and CockroachDB. And there’s a search engine to navigate through the content — Elasticsearch. There are also, perhaps unsurprisingly, jailbreak projects like Cydia compatibility manager for iOS and Nintendo 3DS custom firmware. And there’s a smart home system — Home Assistant. All in all, it’s really a great outlook for the technology world: we learn, build stuff to broadcast our unique voices, we use crypto, break free from proprietary software on our hardware, and in the spare time we game in our automated homes. And the big companies open-source their projects. Before I proceed with the list, a result of running the Octoverse through the reputation algorithm also produced a value score for every individual GitHub contributor. So, if you have a GitHub account and curious, you can get your score at and convert it to a Universal Portable Reputation. https://u.community/github Top 100 projects & repositories Out of over 96 million repositories Repository: Website: Google Kubernetes Container scheduling and management https://github.com/kubernetes/kubernetes https://kubernetes.io/ Repository: Website: Apache Spark A unified analytics engine for large-scale data processing https://github.com/apache/spark http://spark.apache.org/ Repository: Website: Microsoft Visual Studio Code A source-code editor https://github.com/Microsoft/vscode https://code.visualstudio.com/ Repository: Website: NixOS Package Collection A collection of packages for the Nix package manager https://github.com/NixOS/nixpkgs https://nixos.org Repository: Website: Rust Programming language https://github.com/rust-lang/rust https://www.rust-lang.org/ Repository: Website: Firehol IP Lists Blacklists for Firehol, a firewall builder https://github.com/firehol/blocklist-ipsets https://iplists.firehol.org/ Repository: Website: Red Hat OpenShift A community distribution of Kubernetes optimized for continuous application development and multi-tenant deployment https://github.com/openshift/origin https://www.openshift.com/ Repository: Website: Ansible A deployment automation platform https://github.com/ansible/ansible https://www.ansible.com/ Repository: Website: Automattic WordPress Calypso A JavaScript and API powered front-end for WordPress.com https://github.com/Automattic/wp-calypso https://developer.wordpress.com/calypso/ Repository: Website: Microsoft .NET CoreFX Foundational class libraries for .NET Core https://github.com/dotnet/corefx https://docs.microsoft.com/en-us/dotnet/core/ Repository: Website: Microsoft .NET Roslyn .NET compiler https://github.com/dotnet/roslyn https://docs.microsoft.com/en-us/dotnet/csharp/roslyn-sdk/ Repository: Website: Node.js A JavaScript runtime built on Chrome’s V8 JavaScript engine https://github.com/nodejs/node https://nodejs.org/en/ Repository: Website: TensorFlow Google’s machine learning framework https://github.com/tensorflow/tensorflow https://www.tensorflow.org/ Repository: Website: freeCodeCamp Code learning platform https://github.com/freeCodeCamp/freeCodeCamp https://www.freecodecamp.org/ Repository: Website: Space Station 13 A round-based roleplaying game https://github.com/tgstation/tgstation https://www.tgstation13.org/ Repository: Website: Apple Swift Apple’s programming language https://github.com/apple/swift https://swift.org/ Repository: Website: Elasticsearch A search engine https://github.com/elastic/elasticsearch https://www.elastic.co/products/elasticsearch Repository: Website: Moby An open framework to assemble specialized container systems https://github.com/moby/moby https://mobyproject.org/ Repository: Website: CockroachDB A cloud-native SQL database https://github.com/cockroachdb/cockroach https://www.cockroachlabs.com/ Repository: Website: Cydia Compatibility Checker A compatibility checker for Cydia — a package manager for iOS jailbroken devices https://github.com/jlippold/tweakCompatible https://jlippold.github.io/tweakCompatible/ Repository: Website: Servo A web browser engine https://github.com/servo/servo https://servo.org/ Repository: Website: Google Flutter Google’s mobile app SDK to create interfaces for iOS and Android https://github.com/flutter/flutter https://flutter.dev/ Repository: Website: macOS Homebrew Package Manager Default formulae for the missing package manager for macOS https://github.com/homebrew/homebrew-core https://brew.sh/ Repository: Website: Home Assistant Home automation software https://github.com/home-assistant/home-assistant https://www.home-assistant.io/ Repository: Website: Microsoft .NET CoreCLR Runtime for .NET Core https://github.com/dotnet/coreclr https://docs.microsoft.com/en-us/dotnet/core/ Repository: Website: CocoaPods Specifications Specifications for CocoaPods, a Cocoa dependency manager https://github.com/CocoaPods/Specs https://cocoapods.org/ An analytics and search dashboard for ElasticsearchRepository: Website: Elastic Kibana https://github.com/elastic/kibana https://www.elastic.co/products/kibana Repository: Website: Julia Language A technical computing language https://github.com/JuliaLang/julia https://julialang.org/ Repository: Website: Microsoft TypeScript A superset of JavaScript that compiles to plain JavaScript https://github.com/Microsoft/TypeScript https://www.typescriptlang.org/ Repository: Website: Joomla A content management system https://github.com/joomla/joomla-cms https://www.joomla.org/ Repository: Website: DefinitelyTyped A repository for TypeScript type definitions https://github.com/DefinitelyTyped/DefinitelyTyped http://definitelytyped.org/ Repository: Website: Homebrew Cask A CLI workflow for the administration of macOS applications distributed as binaries https://github.com/Homebrew/homebrew-cask https://brew.sh/ Repository: Website: Ceph A distributed object, block, and file storage platform https://github.com/ceph/ceph https://ceph.com/ Repository: Website: Go Programming language https://github.com/golang/go https://golang.org/ AMP HTML Builder Repository: Website: A way to build pages for Google AMP https://github.com/ampproject/amphtml https://amp.dev/ Repository: Website: Open edX An online education platform https://github.com/edx/edx-platform https://open.edx.org/ Repository: Website: Pandas A data analysis and manipulation library for Python https://github.com/pandas-dev/pandas https://pandas.pydata.org/ Repository: Website: Istio A platform to manage microservices https://github.com/istio/istio https://istio.io/ Repository: Website: ManageIQ A containers, virtual machines, networks, and storage management platform https://github.com/ManageIQ/manageiq http://manageiq.org/ Repository: Website: Godot Engine A multi-platform 2D and 3D game engine https://github.com/godotengine/godot https://godotengine.org/ Repository: Website: Gentoo Repository Mirror A Gentoo ebuild repository mirror https://github.com/gentoo/gentoo https://www.gentoo.org/ Repository: Website: Odoo A suite of web based open source business apps https://github.com/odoo/odoo https://www.odoo.com/ Repository: Website: Azure Documentation Documentation of Microsoft Azure https://github.com/MicrosoftDocs/azure-docs https://docs.microsoft.com/azure Repository: Website: Magento An eCommerce platform https://github.com/magento/magento2 https://magento.com/ Repository: Website: Saltstack Software to automate the management and configuration of any infrastructure or application at scale https://github.com/saltstack/salt https://www.saltstack.com/ Repository: Website: AdGuard Filters Ad blocking filters for AdGuard https://github.com/AdguardTeam/AdguardFilters https://adguard.com/en/welcome.html Repository: Website: Symfony A PHP framework https://github.com/symfony/symfony https://symfony.com/ Repository: Website: CMS Software for the Large Hadron Collider Particle detector software components for CERN’s Large Hadron Collider https://github.com/cms-sw/cmssw http://cms-sw.github.io/ Repository: Website: Red Hat OpenShift OpenShift installation and configuration management https://github.com/openshift/openshift-ansible https://www.openshift.com/ Repository: Website: ownCloud Personal cloud software https://github.com/owncloud/core https://owncloud.org/ Repository: Website: gRPC A remote procedure call (RPC) framework https://github.com/grpc/grpc https://grpc.io/ Repository: Website: Liferay An enterprise web platform https://github.com/brianchandotcom/liferay-portal https://www.liferay.com/ Repository: Website: CommCare HQ A mobile data collection platform https://github.com/dimagi/commcare-hq https://www.commcarehq.org/ Repository: Website: WordPress Gutenberg An editor plugin for WordPress https://github.com/WordPress/gutenberg https://wordpress.org/gutenberg/ Repository: Website: PyTorch A Python package for Tensor computation and deep neural networks https://github.com/pytorch/pytorch https://pytorch.org/ Repository: Website: Kubernetes Test Infrastructure A test-infra repository for Kubernetes https://github.com/kubernetes/test-infra https://kubernetes.io/ Repository: Website: Keybase Keybase client repository https://github.com/keybase/client https://keybase.io/ Repository: Website: Facebook React A JavaScript library for building user interfaces https://github.com/facebook/react https://reactjs.org/ Repository: Website: Code.org Code learning resource https://github.com/code-dot-org/code-dot-org https://code.org/ Repository: Website: Bitcoin Core Bitcoin client software https://github.com/bitcoin/bitcoin https://bitcoincore.org/ Repository: Website: Arm Mbed OS A platform operating system for the Internet of Things https://github.com/ARMmbed/mbed-os https://www.mbed.com Repository: Website: scikit-learn A Python module for machine learning https://github.com/scikit-learn/scikit-learn https://scikit-learn.org Repository: Website: Nextcloud A self-hosted productivity platform https://github.com/nextcloud/server https://nextcloud.com/ Repository: Website: Helm Charts A curated list of applications for Kubernetes https://github.com/helm/charts https://kubernetes.io/ Repository: Website: Terraform An infrastructure management tool https://github.com/hashicorp/terraform https://www.terraform.io/ Repository: Website: Ant Design A UI design language https://github.com/ant-design/ant-design https://ant.design/ Repository: Website: Phalcon Framework Documentation Documentation for Phalcon, a PHP framework https://github.com/phalcon/docs https://docs.phalconphp.com Repository: Website: Documentation for CMS Software for the Large Hadron Collider Documentation for CMS Software for CERN’s Large Hadron Collider https://github.com/cms-sw/cms-sw.github.io http://cms-sw.github.io/ Repository: Website: Apache Kafka Mirror A mirror for Apache Kafka, a distributed streaming platform https://github.com/apache/kafka https://kafka.apache.org/ Repository: Website: Electron A framework to write cross-platform desktop applications using JavaScript, HTML and CSS https://github.com/electron/electron https://electronjs.org/ Repository: Website: Zephyr Project A real-time operating system https://github.com/zephyrproject-rtos/zephyr https://www.zephyrproject.org/ Repository: Website: The web-platform-tests Project A cross-browser testsuite for the Web-platform stack https://github.com/web-platform-tests/wpt https://www.w3.org/ Repository: Website: Marlin Firmware Optimized firmware for RepRap 3D printers based on the Arduino platform https://github.com/MarlinFirmware/Marlin http://marlinfw.org/ Repository: Website: Apache MXNet A library for deep learning https://github.com/apache/incubator-mxnet https://mxnet.apache.org/ Repository: Website: Apache Beam A unified programming model https://github.com/apache/beam https://beam.apache.org/ Repository: Website: Fastlane A build and release automaton for iOS and Android apps https://github.com/fastlane/fastlane https://fastlane.tools/ Repository: Website: Kubernetes Website and Documentation A repository for the Kubernetes website and documentation https://github.com/kubernetes/website https://kubernetes.io Repository: Website: Ruby on Rails A web-application framework https://github.com/rails/rails https://rubyonrails.org/ Repository: Website: Zulip Team chat software https://github.com/zulip/zulip https://zulipchat.com/ Repository: Website: Laravel A web application framework https://github.com/laravel/framework https://laravel.com/ Repository: Website: Baidu PaddlePaddle Baidu’s deep learning framework https://github.com/PaddlePaddle/Paddle http://www.paddlepaddle.org/ Repository: Website: Gatsby A web application framework https://github.com/gatsbyjs/gatsby https://www.gatsbyjs.org/ Repository: Website: Rust Crate Registry Rust’s community package registry https://github.com/rust-lang/crates.io-index https://crates.io/ Repository: Website: Nintendo 3DS Custom Firmware A complete guide to 3DS custom firmware https://github.com/hacks-guide/Guide_3DS https://3ds.hacks.guide/ Repository: Website: TiDB A NewSQL database https://github.com/pingcap/tidb https://pingcap.com Repository: Website: Angular CLI CLI tool for Angular, a Google web application framework https://github.com/angular/angular-cli https://cli.angular.io/ Repository: Website: MAPS.ME Offline OpenStreetMap maps for iOS and Android https://github.com/mapsme/omim https://maps.me/ Repository: Website: Eclipse Che A cloud IDE for Eclipse https://github.com/eclipse/che http://www.eclipse.org/che/ Repository: Website: Brave Browser A browser with native BAT cryptocurrency https://github.com/brave/browser-laptop https://www.brave.com/ Repository: Website: Patchwork A repository to learn Git https://github.com/jlord/patchwork http://jlord.us/patchwork/ Repository: Website: Angular Material Component infrastructure and Material Design components for Angular, a Google web application framework https://github.com/angular/components https://material.angular.io/ Repository: Website: Python Programming language https://github.com/python/cpython https://www.python.org/ Repository: Website: Space Station 13 A round-based roleplaying game https://github.com/vgstation-coders/vgstation13 http://ss13.moe/ Repository: Website: Cataclysm: Dark Days Ahead A turn-based survival game https://github.com/CleverRaven/Cataclysm-DDA http://cataclysmdda.org/ Repository: Website: Material-UI React components that implement Google’s Material Design https://github.com/mui-org/material-ui https://material-ui.com/ Repository: Website: Ionic A Progressive Web Apps development framework https://github.com/ionic-team/ionic https://ionicframework.com/ Repository: Website: Oppia A tool for collaboratively building interactive lessons https://github.com/oppia/oppia https://www.oppia.org Repository: Website: Alluxio A virtual distributed storage system https://github.com/Alluxio/alluxio https://www.alluxio.io/ Repository: Website: None XX Net A Chinese web proxy and anti-censorship tool https://github.com/XX-net/XX-Net Repository: Website: Microsoft .NET CLI A CLI tool for .NET https://github.com/dotnet/cli https://docs.microsoft.com/en-us/dotnet/core/tools/ [1] The explanation of the calculation of the simplified version is at the . U°OS Network GitHub repository

GitHub’s Top 100 Most Valuable Repositories Out of 96 Million

This post is curated by that an issue based bounty platform for open source projects. It is the easiest way to make bounty program for OSS. IssueHunt Anyone can fund any issues on GitHub and these money will be distributed to maintainers and contributors 😃 IssueHunt help build sustainable open source community by rewarding project owners and developers. _IssueHunt 🦉 = OSS Development ⚒ + Bounty Program 💰. IssueHunt is an issue-based bounty platform for open source…_issuehunt.io IssueHunt - Boost your issue and boost your happiness 1) TensorFlow Models If you are interested in Machine learning and Deep learning, you must have heard about the TensorFlow. TensorFlow Models is the open-source repository to find many libraries and models related to deep learning. GitHub: https://github.com/tensorflow/models 2) Keras Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. It was developed with a focus on enabling fast experimentation. GitHub: https://github.com/keras-team/keras 3) Flask Flask is a lightweight WSGI web application framework. It is designed to make getting started quick and easy, with the ability to scale up to complex applications. It began as a simple wrapper around Werkzeug and Jinja and has become one of the most popular Python web application frameworks. GitHub: https://github.com/pallets/flask 4) scikit-learn scikit-learn is a Python module for machine learning built on top of SciPy and distributed under the 3-Clause BSD license. GitHub: https://github.com/scikit-learn 5) Zulip Zulip is a powerful, open source group chat application that combines the immediacy of real-time chat with the productivity benefits of threaded conversations. Zulip is used by open source projects, Fortune 500 companies, large standards bodies, and others who need a real-time chat system that allows users to easily process hundreds or thousands of messages a day. With over 300 contributors merging over 500 commits a month, Zulip is also the largest and fastest growing open source group chat project. GitHub: https://github.com/zulip/zulip 6) Django Django is a high-level Python Web framework that encourages rapid development and clean, pragmatic design. GitHub: https://github.com/django/django 7) Rebound Want to save your time during coding session when you get a bug and you want to search it on Stack Overflow? Rebound is a command-line tool that instantly fetches Stack Overflow results when you get a compiler error. This is very handy library for programmers. GitHub: https://github.com/shobrook/rebound 8) Google Images Download This is a command line python program to search keywords/key-phrases on Google Images and optionally download images to your computer. You can also invoke this script from another python file. GitHub: https://github.com/hardikvasa/google-images-download 9) YouTube-dl Youtube-dl — download videos from youtube.com or other video platforms. GitHub: https://github.com/rg3/youtube-dl 10) System Design Primer This repo is an organized collection of resources to help you learn how to build systems at scale. GitHub: https://github.com/donnemartin/system-design-primer 11) Mask R-CNN Mask R-CNN is for object detection and segmentation. This is an implementation of Mask R-CNN on Python 3, Keras, and TensorFlow. The model generates bounding boxes and segmentation masks for each instance of an object in the image. It’s based on Feature Pyramid Network (FPN) and a ResNet101 backbone. GitHub: https://github.com/matterport/Mask_RCNN 12) Face Recognition Recognize and manipulate faces from Python or from the command line with the world’s simplest face recognition library. This also provides a simple face_recognition command line tool that lets you do face recognition on a folder of images from the command line! GitHub: https://github.com/ageitgey/face_recognition 13) snallygaster Tool to scan for secret files on HTTP servers. GitHub: https://github.com/hannob/snallygaster 14) Ansible Ansible is a radically simple IT automation system. It handles configuration-management, application deployment, cloud provisioning, ad-hoc task-execution, and multinode orchestration — including trivializing things like zero-downtime rolling updates with load balancers. GitHub: https://github.com/ansible/ansible 15) Detectron Detectron is Facebook AI Research’s software system that implements state-of-the-art object detection algorithms, including Mask R-CNN. It is written in Python and powered by the Caffe2 deep learning framework. GitHub: https://github.com/facebookresearch/Detectron 16) asciinema Terminal session recorder and the best companion of asciinema.org. GitHub: https://github.com/asciinema/asciinema 17) HTTPie HTTPie is a command line HTTP client. Its goal is to make CLI interaction with web services as human-friendly as possible. It provides a simple http command that allows for sending arbitrary HTTP requests using a simple and natural syntax, and displays colorized output. HTTPie can be used for testing, debugging, and generally interacting with HTTP servers. GitHub: https://github.com/jakubroztocil/httpie 18) You-Get You-Get is a tiny command-line utility to download media contents (videos, audios, images) from the Web, in case there is no other handy way to do it. GitHub: https://github.com/soimort/you-get 19) Sentry Sentry fundamentally is a service that helps you monitor and fix crashes in realtime. The server is in Python, but it contains a full API for sending events from any language, in any application. GitHub: https://github.com/getsentry/sentry 20) Tornado Tornado is a Python web framework and asynchronous networking library, originally developed at FriendFeed. By using non-blocking network I/O, Tornado can scale to tens of thousands of open connections, making it ideal for long polling, WebSockets, and other applications that require a long-lived connection to each user. GitHub: https://github.com/tornadoweb/tornado 21) Magenta Magenta is a research project exploring the role of machine learning in the process of creating art and music. Primarily this involves developing new deep learning and reinforcement learning algorithms for generating songs, images, drawings, and other materials. But it’s also an exploration in building smart tools and interfaces that allow artists and musicians to extend their processes using these models. GitHub: https://github.com/tensorflow/magenta 22) ZeroNet Make decentralized websites using Bitcoin crypto and the BitTorrent network. GitHub: https://github.com/HelloZeroNet/ZeroNet 23) Gym OpenAI Gym is a toolkit for developing and comparing reinforcement learning algorithms. This is the gym open-source library, which gives you access to a standardized set of environments. GitHub: https://github.com/openai/gym 24) Pandas Pandas is a Python package providing fast, flexible, and expressive data structures designed to make working with “relational” or “labeled” data both easy and intuitive. It aims to be the fundamental high-level building block for doing practical, real world data analysis in Python. Additionally, it has the broader goal of becoming the most powerful and flexible open source data analysis/manipulation tool available in any language. It is already well on its way toward this goal. GitHub: https://github.com/pandas-dev/pandas 25) Luigi Luigi is a Python package that helps you build complex pipelines of batch jobs. It handles dependency resolution, workflow management, visualization, handling failures, command line integration, and much more. GitHub: https://github.com/spotify/luigi 26) spaCy spaCy is a library for advanced Natural Language Processing in Python and Cython. It’s built on the very latest research, and was designed from day one to be used in real products. spaCy comes with pre-trained statistical models and word vectors, and currently supports tokenization for 20+ languages. It features the fastest syntactic parser in the world, convolutional neural network models for tagging, parsing and named entity recognition and easy deep learning integration. GitHub: https://github.com/explosion/spaCy 27) Theano Theano is a Python library that allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently. It can use GPUs and perform efficient symbolic differentiation. GitHub: https://github.com/Theano/Theano 28) TFlearn TFlearn is a modular and transparent deep learning library built on top of Tensorflow. It was designed to provide a higher-level API to TensorFlow in order to facilitate and speed-up experimentations, while remaining fully transparent and compatible with it. GitHub: https://github.com/tflearn/tflearn 29) Kivy Kivy is an open source, cross-platform Python framework for the development of applications that make use of innovative, multi-touch user interfaces. The aim is to allow for quick and easy interaction design and rapid prototyping whilst making your code reusable and deploy-able. GitHub: https://github.com/kivy/kivy 30) Mailpile Mailpile is a modern, fast web-mail client with user-friendly encryption and privacy features. The development of Mailpile is funded by a large community of backers and all code related to the project is and will be released under an OSI approved Free Software license. GitHub: https://github.com/mailpile/Mailpile 31) Matplotlib Matplotlib is a Python 2D plotting library which produces publication-quality figures in a variety of hardcopy formats and interactive environments across platforms. Matplotlib can be used in Python scripts, the Python and IPython shell, web application servers, and various graphical user interface toolkits. GitHub: https://github.com/matplotlib/matplotlib 32) YAPF YAPF takes the code and reformats it to the best formatting that conforms to the style guide, even if the original code didn’t violate the style guide. GitHub: https://github.com/google/yapf 33) Cookiecutter A command-line utility that creates projects from cookiecutters (project templates), e.g. creating a Python package project from a Python package project template. GitHub: https://github.com/audreyr/cookiecutter 34) HTTP Prompt HTTP Prompt is an interactive command-line HTTP client featuring autocomplete and syntax highlighting, built on HTTPie and prompt_toolkit. GitHub: https://github.com/eliangcs/http-prompt 35) speedtest-cli Command line interface for testing internet bandwidth using speedtest.net. GitHub: https://github.com/sivel/speedtest-cli 📣IssueHunt This post is curated by that a crowdfunding and sourcing platform for open-source projects. IssueHunt Anyone can fund any issues on GitHub and these money will be distributed to maintainers and contributors. https://issuehunt.io/ 36) Pattern Pattern is a web mining module for Python. It has tools for Data Mining, Natural Language Processing, Machine Learning, and Network Analysis. GitHub: https://github.com/clips/pattern 37) Gooey (Beta) Turn (almost) any Python 2 or 3 Console Program into a GUI application with one line. GitHub: https://github.com/chriskiehl/Gooey 38) Wagtail CMS Wagtail is a content management system built on Django. It’s focused on user experience, and offers precise control for designers and developers. GitHub: https://github.com/wagtail/wagtail 39) Bottle Bottle is a fast, simple and lightweight WSGI micro web-framework for Python. It is distributed as a single file module and has no dependencies other than the Python Standard Library. GitHub: https://github.com/bottlepy/bottle 40) Prophet (by Facebook) Prophet is a procedure for forecasting time series data. It is based on an additive model where non-linear trends are fit with yearly and weekly seasonality, plus holidays. It works best with daily periodicity data with at least one year of historical data. Prophet is robust to missing data, shifts in the trend, and large outliers. GitHub: https://github.com/facebook/prophet 41) Falcon Falcon is a reliable, high-performance Python web framework for building large-scale app backends and microservices. It encourages the REST architectural style, and tries to do as little as possible while remaining highly effective. GitHub: https://github.com/falconry/falcon 42) Mopidy Mopidy is an extensible music server written in Python. Mopidy plays music from local disk, Spotify, SoundCloud, Google Play Music, and more. You edit the playlist from any phone, tablet, or computer using a range of MPD and web clients. GitHub: https://github.com/mopidy/mopidy 43) Hug Hug aims to make developing Python driven APIs as simple as possible, but no simpler. As a result, it drastically simplifies Python API development. GitHub: https://github.com/timothycrosley/hug 44) SymPy A Python library for symbolic mathematics. GitHub: https://github.com/sympy/sympy 45) Dash Dash is a Python framework for building analytical web applications. No JavaScript required. https://github.com/plotly/dash 46) Visdom A flexible tool for creating, organizing, and sharing visualizations of live, rich data. Supports Torch and Numpy. GitHub: https://github.com/facebookresearch/visdom 47) LUMINOTH Luminoth is an open source toolkit for . Currently, we support object detection, but we are aiming for much more. It is built in Python, using TensorFlow and Sonnet. computer vision GitHub: https://github.com/tryolabs/luminoth 48) Pygame Pygame is a cross-platfrom library designed to make it easy to write multimedia software, such as games, in Python. GitHub: https://github.com/pygame/pygame 49) Requests Requests is a Python Library that lets you send HTTP/1.1 requests, add headers, form data, multipart files, and parameters with simple Python dictionaries. It also lets you access the response data in the same way. GitHub: https://github.com/requests/requests 50) Statsmodels Statsmodels is a Python package that provides a complement to scipy for statistical computations including descriptive statistics and estimation and inference for statistical models. GitHub: https://github.com/statsmodels/statsmodels This post is curated by that a crowdfunding and sourcing platform for open-source projects. IssueHunt Anyone can fund any issues on GitHub and these money will be distributed to maintainers and contributors. https://issuehunt.io/

50 Popular Python open-source projects on GitHub in 2018

50 proyectos populares de código abierto de Python en GitHub en 2018

Authors:
(1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA;
(2) Abhinav Tuli, Activeloop, Mountain View, CA, USA;
(3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA;
(4) Fariz Rahman, Activeloop, Mountain View, CA, USA;.
(5) Hrant Topchyan, Activeloop, Mountain View, CA, USA;
(6) David Isayan, Activeloop, Mountain View, CA, USA;
(7) Mark McQuade, Activeloop, Mountain View, CA, USA;
(8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA;
(9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA;
(10) Ivo Stranic, Activeloop, Mountain View, CA, USA;
(11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro
Current Challenges
Tensor Storage Format
Deep Lake System Overview
Machine Learning Use Cases
Performance Benchmarks
Discussion and Limitations
Related Work
Conclusions, Acknowledgement, and References 2. CURRENT CHALLENGES In this section, we discuss the current and historical challenges of unstructured or complex data management. 2.1 Complex Data Types in a Databases It is generally not recommended to store binary data, such as images, directly in a database. This is because databases are not optimized for storing and serving large files and can cause performance issues. In addition, binary data does not fit well into a database’s structured format, making it difficult to query and manipulate. This can lead to slow load times for users. Databases are typically more expensive to operate and maintain than other types of storage, such as file systems or cloud storage services. Therefore, storing large amounts of binary data in a database can be more costly than other storage solutions. 2.2 Complex Data Along with Tabular Formats Increases in large-scale analytical and BI workloads motivated the development of compressed structured formats like Parquet, ORC, Avro, or transient in-memory formats like Arrow [79, 6, 20, 13]. As tabular formats gained adoption, attempts to extend those formats, such as Petastorm [18] or Feather [7] for deep learning, have emerged. To the best of our knowledge, these formats have yet to gain wide adoption. This approach primarily benefits from native integrations with Modern Data Stack (MDS). However, as discussed previously, upstream tools require fundamental modifications to adapt to deep learning applications. 2.3 Object Storage for Deep Learning The current cloud-native choice for storing large unstructured datasets is object storage such as AWS S3 [1], Google Cloud Storage (GCS) [3], or MinIO [17]. Object storage does offer three main benefits over distributed network file systems. They are (a) cost-efficient, (b) scalable, and (c) serve as a format-agnostic repository. However, cloud storages are not without drawbacks. Firstly, they introduce significant latency overhead, especially when iterating over many small files such as text or JSON. Next, unstructured data ingestion without metadata control can produce "data swamps". Furthermore, object storage has built-in version control; it is rarely used in data science workflows. Lastly, data on object storage gets copied to a virtual machine before training, thus resulting in storage overhead and additional costs. 2.4 Second Generation of Data Lakes The second-generation data lakes led by Delta, Iceberg, Hudi [27, 15, 10] extend object storage by managing tabular format files with the following primary properties. (1) Update operations: inserting or deleting a row on top of a tabular format file. (2) Streaming: downstream data ingestion with ACID properties and upstream integration with query engine exposing SQL interface. (3) Schema evolution: evolving columnar structure while preserving backward compatibility. (4) Time travel and audit log trailing: preserving historical state with rollback property where queries can be reproducible. Also, support for row-level control on data lineage. (5) Layout optimization: Built-in feature to optimize file sizes and data compaction with custom ordering support. Significantly speeds up querying. However, second-generation data lakes are still bound by the limitations of the inherent data formats to be used in deep learning, as previously discussed in section 2.2. Hence in this paper, we extend the second generation of data lake capabilities for deep learning use cases by rethinking the format and upstream features, including querying, visualization, and native integration to deep learning frameworks to complete the ML lifecycle as shown in Fig. 2. This paper is available on arxiv under CC 4.0 license. Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Authors: Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro Current Challenges Tensor Storage Format Deep Lake System Overview Machine Learning Use Cases Performance Benchmarks Discussion and Limitations Related Work Conclusions, Acknowledgement, and References Abstract and Intro Abstract and Intro Current Challenges Current Challenges Tensor Storage Format Tensor Storage Format Deep Lake System Overview Deep Lake System Overview Machine Learning Use Cases Machine Learning Use Cases Performance Benchmarks Performance Benchmarks Discussion and Limitations Discussion and Limitations Related Work Related Work Conclusions, Acknowledgement, and References Conclusions, Acknowledgement, and References 2. CURRENT CHALLENGES In this section, we discuss the current and historical challenges of unstructured or complex data management. 2.1 Complex Data Types in a Databases It is generally not recommended to store binary data, such as images, directly in a database. This is because databases are not optimized for storing and serving large files and can cause performance issues. In addition, binary data does not fit well into a database’s structured format, making it difficult to query and manipulate. This can lead to slow load times for users. Databases are typically more expensive to operate and maintain than other types of storage, such as file systems or cloud storage services. Therefore, storing large amounts of binary data in a database can be more costly than other storage solutions. 2.2 Complex Data Along with Tabular Formats Increases in large-scale analytical and BI workloads motivated the development of compressed structured formats like Parquet, ORC, Avro, or transient in-memory formats like Arrow [79, 6, 20, 13]. As tabular formats gained adoption, attempts to extend those formats, such as Petastorm [18] or Feather [7] for deep learning, have emerged. To the best of our knowledge, these formats have yet to gain wide adoption. This approach primarily benefits from native integrations with Modern Data Stack (MDS). However, as discussed previously, upstream tools require fundamental modifications to adapt to deep learning applications. 2.3 Object Storage for Deep Learning The current cloud-native choice for storing large unstructured datasets is object storage such as AWS S3 [1], Google Cloud Storage (GCS) [3], or MinIO [17]. Object storage does offer three main benefits over distributed network file systems. They are (a) cost-efficient, (b) scalable, and (c) serve as a format-agnostic repository. However, cloud storages are not without drawbacks. Firstly, they introduce significant latency overhead, especially when iterating over many small files such as text or JSON. Next, unstructured data ingestion without metadata control can produce "data swamps". Furthermore, object storage has built-in version control; it is rarely used in data science workflows. Lastly, data on object storage gets copied to a virtual machine before training, thus resulting in storage overhead and additional costs. 2.4 Second Generation of Data Lakes The second-generation data lakes led by Delta, Iceberg, Hudi [27, 15, 10] extend object storage by managing tabular format files with the following primary properties. (1) Update operations: inserting or deleting a row on top of a tabular format file. Update operations: (2) Streaming : downstream data ingestion with ACID properties and upstream integration with query engine exposing SQL interface. Streaming (3) Schema evolution: evolving columnar structure while preserving backward compatibility. Schema evolution: (4) Time travel and audit log trailing: preserving historical state with rollback property where queries can be reproducible. Also, support for row-level control on data lineage. Time travel and audit log trailing: (5) Layout optimization: Built-in feature to optimize file sizes and data compaction with custom ordering support. Significantly speeds up querying. Layout optimization: However, second-generation data lakes are still bound by the limitations of the inherent data formats to be used in deep learning, as previously discussed in section 2.2. Hence in this paper, we extend the second generation of data lake capabilities for deep learning use cases by rethinking the format and upstream features, including querying, visualization, and native integration to deep learning frameworks to complete the ML lifecycle as shown in Fig. 2. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Deep Lake, a Lakehouse for Deep Learning: Current Challenges

গভীর লেক, গভীর শিক্ষার জন্য একটি লেকহাউস: বর্তমান চ্যালেঞ্জ

Deep Lake, ein Lakehouse für Deep Learning: Aktuelle Herausforderungen

Deep Lake, un lago para el aprendizaje profundo: desafíos actuales

Deep Lake, un Lakehouse pour le Deep Learning : les défis actuels

डीप लेक, डीप लर्निंग के लिए एक लेकहाउस: वर्तमान चुनौतियाँ

ディープラーニングのためのレイクハウス「Deep Lake」：現在の課題

딥 러닝을 위한 레이크하우스인 Deep Lake: 현재의 과제

Deep Lake, um Lakehouse para Deep Learning: desafios atuais

Deep Lake, домик у озера для глубокого обучения: текущие проблемы

Deep Lake, Derin Öğrenme için Bir Göl Evi: Güncel Zorluklar

Deep Lake, Lakehouse cho Deep Learning: Những thách thức hiện tại

Deep Lake，深度学习的 Lakehouse：当前的挑战

Authors:
(1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA;
(2) Abhinav Tuli, Activeloop, Mountain View, CA, USA;
(3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA;
(4) Fariz Rahman, Activeloop, Mountain View, CA, USA;.
(5) Hrant Topchyan, Activeloop, Mountain View, CA, USA;
(6) David Isayan, Activeloop, Mountain View, CA, USA;
(7) Mark McQuade, Activeloop, Mountain View, CA, USA;
(8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA;
(9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA;
(10) Ivo Stranic, Activeloop, Mountain View, CA, USA;
(11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro
Current Challenges
Tensor Storage Format
Deep Lake System Overview
Machine Learning Use Cases
Performance Benchmarks
Discussion and Limitations
Related Work
Conclusions, Acknowledgement, and References 5. MACHINE LEARNING USE CASES In this section, we review the applications of Deep Lake. A typical scenario in a Deep Learning application starts with (1) A raw set of files that is collected on an object storage bucket. It might include images, videos, and other types of multimedia data in their native formats such as JPEG, PNG or MP4. (2) Any associated metadata and labels stored on a relational database. Optionally, they could be stored on the same bucket along with the raw data in a normalized tabular form such as CSV, JSON, or Parquet format. As shown in Fig. 4, an empty Deep Lake dataset is created. Then, empty tensors are defined for storing both raw data as well as metadata. The number of tensors could be arbitrary. A basic example of an image classification task would have two tensors, • images tensor with htype of 𝑖𝑚𝑎𝑔𝑒 and sample compression of JPEG • labels tensor with htype of 𝑐𝑙𝑎𝑠𝑠_𝑙𝑎𝑏𝑒𝑙 and chunk compression of LZ4. After declaring tensors, the data can be appended to the dataset. If a raw image compression matches the tensor sample compression, the binary is directly copied into a chunk without additional decoding. Label data is extracted from a SQL query or CSV table into a categorical integer and appended into labels tensor. labels tensor chunks are stored using LZ4 compression. All Deep Lake data is stored in the bucket and is self-contained. After storage, the data can be accessed in a NumPy interface or as a streamable deep learning dataloader. Then, the model running on a compute machine iterates over the stream of image tensors, and stores the output of the model in a new tensor called predictions. Furthermore, we discuss below how one can train, version control, query, and inspect the quality of a Deep Lake dataset. 5.1 Deep Learning Model Training Deep learning models are trained at multiple levels in an organization, ranging from exploratory training occurring on personal computers to large-scale training that occurs on distributed machines involving many GPUs. The time and effort required to bring the data from long-term storage to the training client are often comparable to the training itself. Deep Lake solves this problem by enabling rapid streaming of data without bottlenecking the downstream training process, thus avoiding the cost and time required to duplicate data on local storage. 5.2 Data Lineage and Version Control Deep learning data constantly evolve as new data is added and existing data is quality controlled. Analytical and training workloads occur in parallel while the data is changing. Hence, knowing which data version was used by a given workload is critical to understand the relationship between the data and model performance. Deep Lake enables deep learning practitioners to understand which version of their data was used in any analytical workload and to time travel across these versions if an audit is required. Since all data is mutable, it can be edited to meet compliance-related privacy requirements. Like Git for code, Deep Lake also introduces the concept of data branches, allowing experimentation and editing of data without affecting colleagues’ work. 5.3 Data Querying and Analytics Training of deep learning models rarely occurs on all data collected by an organization for a particular application. Training datasets are often constructed by filtering the raw data based on conditions increasing model performance, which often includes data balancing, eliminating redundant data, or selecting data that contains specific features. Deep Lake provides the tools to query and analyze data so that deep learning engineers can create datasets yielding the highest accuracy models. 5.4 Data Inspection and Quality Control Though unsupervised learning is becoming more applicable in realworld use cases, most deep learning applications still rely on supervised learning. Any supervised learning system is only as good as the quality of its data, often achieved by manual and exhaustive inspection of the data. Since this process is time-consuming, it is critical to provide the humans in the loop with tools to examine vast amounts of data very quickly. Deep Lake allows inspecting deep learning datasets of any size from the browser without any setup time or need to download data. Furthermore, the tools can be extended for comparing model results with ground truth. Combined with querying and version control, this can be applied to the iterative improvement of data to achieve the best possible model. This paper is available on arxiv under CC 4.0 license. Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Authors: Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro Current Challenges Tensor Storage Format Deep Lake System Overview Machine Learning Use Cases Performance Benchmarks Discussion and Limitations Related Work Conclusions, Acknowledgement, and References Abstract and Intro Abstract and Intro Current Challenges Current Challenges Tensor Storage Format Tensor Storage Format Deep Lake System Overview Deep Lake System Overview Machine Learning Use Cases Machine Learning Use Cases Performance Benchmarks Performance Benchmarks Discussion and Limitations Discussion and Limitations Related Work Related Work Conclusions, Acknowledgement, and References Conclusions, Acknowledgement, and References 5. MACHINE LEARNING USE CASES In this section, we review the applications of Deep Lake. A typical scenario in a Deep Learning application starts with (1) A raw set of files that is collected on an object storage bucket. It might include images, videos, and other types of multimedia data in their native formats such as JPEG, PNG or MP4. (2) Any associated metadata and labels stored on a relational database. Optionally, they could be stored on the same bucket along with the raw data in a normalized tabular form such as CSV, JSON, or Parquet format. As shown in Fig. 4, an empty Deep Lake dataset is created. Then, empty tensors are defined for storing both raw data as well as metadata. The number of tensors could be arbitrary. A basic example of an image classification task would have two tensors, • images tensor with htype of 𝑖𝑚𝑎𝑔𝑒 and sample compression of JPEG • labels tensor with htype of 𝑐𝑙𝑎𝑠𝑠_𝑙𝑎𝑏𝑒𝑙 and chunk compression of LZ4. After declaring tensors, the data can be appended to the dataset. If a raw image compression matches the tensor sample compression, the binary is directly copied into a chunk without additional decoding. Label data is extracted from a SQL query or CSV table into a categorical integer and appended into labels tensor. labels tensor chunks are stored using LZ4 compression. All Deep Lake data is stored in the bucket and is self-contained. After storage, the data can be accessed in a NumPy interface or as a streamable deep learning dataloader. Then, the model running on a compute machine iterates over the stream of image tensors, and stores the output of the model in a new tensor called predictions. Furthermore, we discuss below how one can train, version control, query, and inspect the quality of a Deep Lake dataset. 5.1 Deep Learning Model Training Deep learning models are trained at multiple levels in an organization, ranging from exploratory training occurring on personal computers to large-scale training that occurs on distributed machines involving many GPUs. The time and effort required to bring the data from long-term storage to the training client are often comparable to the training itself. Deep Lake solves this problem by enabling rapid streaming of data without bottlenecking the downstream training process, thus avoiding the cost and time required to duplicate data on local storage. 5.2 Data Lineage and Version Control Deep learning data constantly evolve as new data is added and existing data is quality controlled. Analytical and training workloads occur in parallel while the data is changing. Hence, knowing which data version was used by a given workload is critical to understand the relationship between the data and model performance. Deep Lake enables deep learning practitioners to understand which version of their data was used in any analytical workload and to time travel across these versions if an audit is required. Since all data is mutable, it can be edited to meet compliance-related privacy requirements. Like Git for code, Deep Lake also introduces the concept of data branches, allowing experimentation and editing of data without affecting colleagues’ work. 5.3 Data Querying and Analytics Training of deep learning models rarely occurs on all data collected by an organization for a particular application. Training datasets are often constructed by filtering the raw data based on conditions increasing model performance, which often includes data balancing, eliminating redundant data, or selecting data that contains specific features. Deep Lake provides the tools to query and analyze data so that deep learning engineers can create datasets yielding the highest accuracy models. 5.4 Data Inspection and Quality Control Though unsupervised learning is becoming more applicable in realworld use cases, most deep learning applications still rely on supervised learning. Any supervised learning system is only as good as the quality of its data, often achieved by manual and exhaustive inspection of the data. Since this process is time-consuming, it is critical to provide the humans in the loop with tools to examine vast amounts of data very quickly. Deep Lake allows inspecting deep learning datasets of any size from the browser without any setup time or need to download data. Furthermore, the tools can be extended for comparing model results with ground truth. Combined with querying and version control, this can be applied to the iterative improvement of data to achieve the best possible model. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Deep Lake, a Lakehouse for Deep Learning: Machine Learning Use Cases

ডিপ লেক, গভীর শিক্ষার জন্য একটি লেকহাউস: মেশিন লার্নিং ব্যবহারের ক্ষেত্রে

Deep Lake, ein Lakehouse für Deep Learning: Anwendungsfälle für maschinelles Lernen

Deep Lake, una casa de lago para el aprendizaje profundo: casos de uso del aprendizaje automático

Deep Lake, un Lakehouse pour le Deep Learning : cas d'utilisation du Machine Learning

डीप लेक, डीप लर्निंग के लिए एक लेकहाउस: मशीन लर्निंग उपयोग के मामले

ディープラーニングのためのレイクハウス、Deep Lake: 機械学習のユースケース

딥 러닝을 위한 레이크하우스인 Deep Lake: 머신 러닝 사용 사례

Deep Lake, um lago para aprendizado profundo: casos de uso de aprendizado de máquina

Deep Lake, дом для глубокого обучения: примеры использования машинного обучения

Deep Lake, Derin Öğrenme için Bir Göl Evi: Makine Öğrenimi Kullanım Örnekleri

Deep Lake, Lakehouse cho Deep Learning: Các trường hợp sử dụng Machine Learning

Deep Lake，深度学习的 Lakehouse：机器学习用例

TULIP INTERFACES

tulip.co

-- employees

Since n.d.

COMPANY RANKING

TULIP INTERFACES

EVERGREEN INDEX #11235

Representing this company? Click here to claim and customize this page!

Tulip Interfaces WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!

TULIP INTERFACES

tulip.co

-- employees

Since n.d.

COMPANY RANKING

TULIP INTERFACES

EVERGREEN INDEX #11235

RELATED COMPANIESRANK

STORIES ABOUT TULIP INTERFACES

Representing this company? Click here to claim and customize this page!

Tulip Interfaces WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!