Kula automates the business side of building a law firm- letting lawyers focus on the law! We believe there's a fundamental change going on in the legal industry and the next generation of law firms is being built today

Kula

KULAN TITH’S SACRIFICE

This paper is available on arxiv under CC 4.0 license. Authors: (1) Jeffrey Ouyang-Zhang, UT Austin (2) Daniel J. Diaz, UT Austin (3) Adam R. Klivans, UT Austin (4) Philipp Krähenbühl, UT Austin Table of Links Abstract & Introduction Related Works Preliminary Methodology Results Limitations Conclusion, Acknowledgements, & References Experimental Setting & Additional Results 7 Conclusion We present a method that efficiently scales thermodynamic stability prediction from single mutations to higher-order mutations. Our key insight is that the effects of mutations on the same protein are correlated. Thus, for a target protein, it suffices to run a deep backbone once and decode the effect of all mutations simultaneously using a shallow decoder. With the AlphaFold model as our backbone, our method outperforms existing methods on a variety of single and multiple mutation benchmarks. Our method scales to millions of mutations with minimal computational overhead and runs in a fraction of the time it would take prior works. 8 Acknowledgements This work is supported by the NSF AI Institute for Foundations of Machine Learning (IFML). References [1] Randomization of genes by pcr mutagenesis. Genome research 2(1), 28–33 (1992) 2 [2] Adams, J.P., Brown, M.J., Diaz-Rodriguez, A., Lloyd, R.C., Roiban, G.D.: Biocatalysis: A Pharma Perspective. Advanced Synthesis and Catalysis 361(11), 2421–2432 (2019). https://doi.org/10.1002/adsc.201900424 1 [3] Ahdritz, G., Bouatta, N., Kadyan, S., Xia, Q., Gerecke, W., O’Donnell, T.J., Berenberg, D., Fisk, I., Zanichelli, N., Zhang, B., Nowaczynski, A., Wang, B., Stepniewska-Dziubinska, M.M., Zhang, S., Ojewole, A., Guney, M.E., Biderman, S., Watkins, A.M., Ra, S., Lorenzo, P.R., Nivon, L., Weitzner, B., Ban, Y.E.A., Sorger, P.K., Mostaque, E., Zhang, Z., Bonneau, R., AlQuraishi, M.: Openfold: Retraining alphafold2 yields new insights into its learning mechanisms and capacity for generalization. bioRxiv (2022). https://doi.org/10.1101/2022.11.20.517210 6 [4] Arnold, F.H.: Design by directed evolution. Accounts of chemical research 31(3), 125–131 (1998) 1, 2, 3 [5] Bell, E.L., Finnigan, W., France, S.P., Green, A.P., Hayes, M.A., Hepworth, L.J., Lovelock, S.L., Niikura, H., Osuna, S., Romero, E., Ryan, K.S., Turner, N.J., Flitsch, S.L.: Biocatalysis. Nature Reviews Methods Primers 1(1), 1–21 (2021). https://doi.org/10.1038/s43586-021-00044-z 1 [6] Benevenuta, S., Pancotti, C., Fariselli, P., Birolo, G., Sanavia, T.: An antisymmetric neural network to predict free energy changes in protein variants. Journal of Physics D: Applied Physics 54(24), 245403 [7] Benevenuta, S., Pancotti, C., Fariselli, P., Birolo, G., Sanavia, T.: An antisymmetric neural network to predict free energy changes in protein variants. Journal of Physics D: Applied Physics 54(24), 245403 (2021) 8 [8] Benevenuta, S., Birolo, G., Sanavia, T., Capriotti, E., Fariselli, P.: Challenges in predicting stabilizing variations: An exploration. Frontiers in Molecular Biosciences 9, 1075570 (2023) 5, 6, 17 [9] Brückner, A., Polge, C., Lentze, N., Auerbach, D., Schlattner, U.: Yeast two-hybrid, a powerful tool for systems biology. International journal of molecular sciences 10(6), 2763–2788 (2009) 1 [10] Cadet, F., Saavedra, E., Syren, P.O., Gontero, B.: Machine learning, epistasis, and protein engineering: From sequence-structure-function relationships to regulation of metabolic pathways. Frontiers in Molecular Biosciences 9, 1098289 (2022) 1 [11] Capriotti, E., Fariselli, P., Casadio, R.: I-mutant2. 0: predicting stability changes upon mutation from the protein sequence or structure. Nucleic acids research 33(suppl_2), W306–W310 (2005) 3, 8 [12] Chen, K., Arnold, F.H.: Tuning the activity of an enzyme for unusual environments: sequential random mutagenesis of subtilisin e for catalysis in dimethylformamide. Proceedings of the National Academy of Sciences 90(12), 5618–5622 (1993) 1, 3 [13] Chen, T., Gong, C., Diaz, D.J., Chen, X., Wells, J.T., Wang, Z., Ellington, A., Dimakis, A., Klivans, A., et al.: Hotprotein: A novel framework for protein thermostability prediction and editing. In: The Eleventh International Conference on Learning Representations (2022) 3 [14] Chen, Y., Lu, H., Zhang, N., Zhu, Z., Wang, S., Li, M.: Premps: Predicting the impact of missense mutations on protein stability. PLoS computational biology 16(12), e1008543 (2020) 3, 8 [15] Cheng, J., Randall, A., Baldi, P.: Prediction of protein stability changes for single-site mutations using support vector machines. Proteins: Structure, Function, and Bioinformatics 62(4), 1125–1132 (2006) 3, 8 [16] Dehghanpoor, R., Ricks, E., Hursh, K., Gunderson, S., Farhoodi, R., Haspel, N., Hutchinson, B., Jagodzinski, F.: Predicting the effect of single and multiple mutations on protein structural stability. Molecules 23(2), 251 (2018) 4, 7 [17] Dehouck, Y., Kwasigroch, J.M., Gilis, D., Rooman, M.: Popmusic 2.1: a web server for the estimation of protein stability changes upon mutation and sequence optimality. BMC bioinformatics 12(1), 1–12 (2011) 3, 8 [18] Diaz, D.J., Gong, C., Ouyang-Zhang, J., Loy, J.M., Wells, J.T., Yang, D., Ellington, A.J., Dimakis, A., Klivans, A.R.: Stability oracle: A structure-based graph-transformer for identifying stabilizing mutations (2023). https://doi.org/10.1101/2023.05.15.540857 3, 4, 7, 8, 16, 17, 18, 19 [19] Diaz, D.J., Kulikova, A.V., Ellington, A.D., Wilke, C.O.: Using machine learning to predict the effects and consequences of mutations in proteins. Current Opinion in Structural Biology 78, 102518 (2023) 3 [20] d’Oelsnitz, S., Diaz, D.J., Acosta, D.J., Schechter, M.W., Minus, M.B., Howard, J.R., Do, H., Loy, J., Alper, H., Ellington, A.D.: Synthetic microbial sensing and biosynthesis of amaryllidaceae alkaloids. bioRxiv pp. 2023–04 (2023) 3 [21] d’Oelsnitz, S., Kim, W., Burkholder, N.T., Javanmardi, K., Thyer, R., Zhang, Y., Alper, H.S., Ellington, A.D.: Using fungible biosensors to evolve improved alkaloid biosyntheses. Nature Chemical Biology 18(9), 981–989 (2022) 1 [22] Ellefson, J.W., Gollihar, J., Shroff, R., Shivram, H., Iyer, V.R., Ellington, A.D.: Synthetic evolutionary origin of a proofreading reverse transcriptase. Science 352(6293), 1590–1593 (2016) 1 [23] Feng, X., Sanchis, J., Reetz, M.T., Rabitz, H.: Enhancing the efficiency of directed evolution in focused enzyme libraries by the adaptive substituent reordering algorithm. Chemistry–A European Journal 18(18), 5646–5654 (2012) 3 [24] Frazer, J., Notin, P., Dias, M., Gomez, A., Min, J.K., Brock, K., Gal, Y., Marks, D.S.: Disease variant prediction with deep generative models of evolutionary data. Nature 599(7883), 91–95 (2021) 9 [25] Gebauer, M., Skerra, A.: Engineered protein scaffolds as next-generation therapeutics. Annual Review of Pharmacology and Toxicology 60, 391–415 (2020). https://doi.org/10.1146/annurev-pharmtox-010818- 021118 1 [26] Gerasimavicius, L., Livesey, B.J., Marsh, J.A.: Correspondence between functional scores from deep mutational scans and predicted effects on protein stability. bioRxiv pp. 2023–02 (2023) 9 [27] Giver, L., Gershenson, A., Freskgard, P.O., Arnold, F.H.: Directed evolution of a thermostable esterase. Proceedings of the National Academy of Sciences 95(22), 12809–12813 (1998) 2, 3 [28] Hie, B.L., Shanker, V.R., Xu, D., Bruun, T.U., Weidenbacher, P.A., Tang, S., Wu, W., Pak, J.E., Kim, P.S.: Efficient evolution of human antibodies from general protein language models. Nature Biotechnology (2023) 3 [29] Jaroszewicz, W., Morcinek-Orłowska, J., Pierzynowska, K., Gaffke, L., W˛egrzyn, G.: Phage display and other peptide display technologies. FEMS Microbiology Reviews 46(2), fuab052 (2022) 1 [30] Järvelin, K., Kekäläinen, J.: Cumulated gain-based evaluation of ir techniques. ACM Transactions on Information Systems (TOIS) 20(4), 422–446 (2002) 6, 17 [31] Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al.: Highly accurate protein structure prediction with alphafold. Nature 596(7873), 583–589 (2021) 2, 3, 4, 6, 9, 10 [32] Kellogg, E.H., Leaver-Fay, A., Baker, D.: Role of conformational sampling in computing mutation-induced changes in protein structure and stability. Proteins: Structure, Function, and Bioinformatics 79(3), 830–838 (2011) 3 [33] Klabunde, T., Petrassi, H.M., Oza, V.B., Raman, P., Kelly, J.W., Sacchettini, J.C.: Rational design of potent human transthyretin amyloid disease inhibitors. Nature structural biology 7(4), 312–321 (2000) 1, 2 [34] Kouba, P., Kohout, P., Haddadi, F., Bushuiev, A., Samusevich, R., Sedlar, J., Damborsky, J., Pluskal, T., Sivic, J., Mazurenko, S.: Machine learning-guided protein engineering. ACS Catalysis 13, 13863–13895 (2023) 3 [35] Laimer, J., Hofer, H., Fritz, M., Wegenkittl, S., Lackner, P.: Maestro-multi agent stability prediction upon point mutations. BMC bioinformatics 16(1), 1–13 (2015) 3, 4, 7, 8 [36] Li, B., Yang, Y.T., Capra, J.A., Gerstein, M.B.: Predicting changes in protein thermodynamic stability upon point mutation with deep 3d convolutional neural networks. PLoS computational biology 16(11), e1008291 (2020) 3, 4, 8 [37] Li, G., Panday, S.K., Alexov, E.: Saafec-seq: a sequence-based method for predicting the effect of single point mutations on protein thermodynamic stability. International journal of molecular sciences 22(2), 606 (2021) 3 [38] Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., Smetanin, N., dos Santos Costa, A., Fazel-Zarandi, M., Sercu, T., Candido, S., et al.: Language models of protein sequences at the scale of evolution enable accurate structure prediction. bioRxiv (2022) 3, 4, 6, 9, 10, 18 [39] Loell, K., Nanda, V.: Marginal protein stability drives subcellular proteome isoelectric point. Proceedings of the National Academy of Sciences 115(46), 11778–11783 (2018) 1 [40] Lu, H., Diaz, D.J., Czarnecki, N.J., Zhu, C., Kim, W., Shroff, R., Acosta, D.J., Alexander, B.R., Cole, H.O., Zhang, Y., et al.: Machine learning-aided engineering of hydrolases for pet depolymerization. Nature 604(7907), 662–667 (2022) 3 [41] Meghwanshi, G.K., Kaur, N., Verma, S., Dabi, N.K., Vashishtha, A., Charan, P.D., Purohit, P., Bhandari, H.S., Bhojak, N., Kumar, R.: Enzymes for pharmaceutical and therapeutic applications. Biotechnology and Applied Biochemistry 67(4), 586–601 (2020). https://doi.org/10.1002/bab.1919 1 [42] Meier, J., Rao, R., Verkuil, R., Liu, J., Sercu, T., Rives, A.: Language models enable zero-shot prediction of the effects of mutations on protein function. Advances in Neural Information Processing Systems 34, 29287–29303 (2021) 9, 18 [43] Mirdita, M., Schütze, K., Moriwaki, Y., Heo, L., Ovchinnikov, S., Steinegger, M.: Colabfold: making protein folding accessible to all. Nature methods 19(6), 679–682 (2022) 6 [44] Montanucci, L., Capriotti, E., Frank, Y., Ben-Tal, N., Fariselli, P.: Ddgun: an untrained method for the prediction of protein stability changes upon single and multiple point variations. BMC bioinformatics 20, 1–10 (2019) 3, 4, 7, 8, 16, 18 [45] Montanucci, L., Savojardo, C., Martelli, P.L., Casadio, R., Fariselli, P.: On the biases in predictions of protein stability changes upon variations: the inps test case. Bioinformatics 35(14), 2525–2527 (2019) 16 [46] Musdal, Y., Govindarajan, S., Mannervik, B.: Exploring sequence-function space of a poplar glutathione transferase using designed information-rich gene variants. Protein Engineering, Design and Selection 30(8), 543–549 (2017) 3 [47] Nijkamp, E., Ruffolo, J., Weinstein, E.N., Naik, N., Madani, A.: Progen2: exploring the boundaries of protein language models. arXiv preprint arXiv:2206.13517 (2022) 9 [48] Nikam, R., Kulandaisamy, A., Harini, K., Sharma, D., Gromiha, M.M.: Prothermdb: thermodynamic database for proteins and mutants revisited after 15 years. Nucleic acids research 49(D1), D420–D424 (2021) 2, 16 [49] Notin, P., Dias, M., Frazer, J., Hurtado, J.M., Gomez, A.N., Marks, D., Gal, Y.: Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval. In: International Conference on Machine Learning. pp. 16990–17017. PMLR (2022) 2, 3, 9, 16, 18 [50] Notin, P.M., Van Niekerk, L., Kollasch, A.W., Ritter, D., Gal, Y., Marks, D.: Trancepteve: Combining family-specific and family-agnostic models of protein sequences for improved fitness prediction. bioRxiv pp. 2022–12 (2022) 9 [51] Nutschel, C., Fulton, A., Zimmermann, O., Schwaneberg, U., Jaeger, K.E., Gohlke, H.: Systematically scrutinizing the impact of substitution sites on thermostability and detergent tolerance for bacillus subtilis lipase a. Journal of chemical information and modeling 60(3), 1568–1584 (2020) 9 [52] Paik, I., Ngo, P.H., Shroff, R., Diaz, D.J., Maranhao, A.C., Walker, D.J., Bhadra, S., Ellington, A.D.: Improved bst dna polymerase variants derived via a machine learning approach. Biochemistry (2021) 3 [53] Pak, M.A., Markhieva, K.A., Novikova, M.S., Petrov, D.S., Vorobyev, I.S., Maksimova, E.S., Kondrashov, F.A., Ivankov, D.N.: Using alphafold to predict the impact of single mutations on protein stability and function. Plos one 18(3), e0282689 (2023) 3 [54] Pancotti, C., Benevenuta, S., Birolo, G., Alberini, V., Repetto, V., Sanavia, T., Capriotti, E., Fariselli, P.: Predicting protein stability changes upon single-point mutation: a thorough comparison of the available tools on a new dataset. Briefings in Bioinformatics 23(2), bbab555 (2022) 2, 8, 16, 17 [55] Pancotti, C., Benevenuta, S., Repetto, V., Birolo, G., Capriotti, E., Sanavia, T., Fariselli, P.: A deep-learning sequence-based method to predict protein stability changes upon genetic variations. Genes 12(6), 911 (2021) 8 [56] Pires, D.E., Ascher, D.B., Blundell, T.L.: Duet: a server for predicting effects of mutations on protein stability using an integrated computational approach. Nucleic acids research 42(W1), W314–W319 (2014) 3, 8 [57] Pires, D.E., Ascher, D.B., Blundell, T.L.: mcsm: predicting the effects of mutations in proteins using graph-based signatures. Bioinformatics 30(3), 335–342 (2014) 8 [58] Pucci, F., Schwersensky, M., Rooman, M.: Artificial intelligence challenges for predicting the impact of mutations on protein stability. Current opinion in structural biology 72, 161–168 (2022) 17 [59] Qiu, Y., Wei, G.W.: Persistent spectral theory-guided protein engineering. Nature Computational Science pp. 1–15 (2023) 3, 6, 17 [60] Rao, R.M., Liu, J., Verkuil, R., Meier, J., Canny, J., Abbeel, P., Sercu, T., Rives, A.: Msa transformer. In: International Conference on Machine Learning. pp. 8844–8856. PMLR (2021) 3, 9, 10 [61] Riesselman, A.J., Ingraham, J.B., Marks, D.S.: Deep generative models of genetic variation capture the effects of mutations. Nature methods 15(10), 816–822 (2018) 9 [62] Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C.L., Ma, J., Fergus, R.: Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. PNAS (2019). https://doi.org/10.1101/622803 3, 7 [63] Rodrigues, C.H., Pires, D.E., Ascher, D.B.: Dynamut: predicting the impact of mutations on protein conformation, flexibility and stability. Nucleic acids research 46(W1), W350–W355 (2018) 3, 8 [64] Rodrigues, C.H., Pires, D.E., Ascher, D.B.: Dynamut2: Assessing changes in stability and flexibility upon single and multiple point missense mutations. Protein Science 30(1), 60–69 (2021) 3, 4, 7 [65] Sanavia, T., Birolo, G., Montanucci, L., Turina, P., Capriotti, E., Fariselli, P.: Limitations and challenges in protein stability prediction upon genome variations: towards future applications in precision medicine. Computational and Structural Biotechnology Journal 18, 1968–1979 (2020). https://doi.org/10.1016/j.csbj.2020.07.011, https://doi.org/10.1016/j.csbj.2020.07.011 6, 17 [66] Savojardo, C., Fariselli, P., Martelli, P.L., Casadio, R.: Inps-md: a web server to predict stability of protein variants from sequence and structure. Bioinformatics 32(16), 2542–2544 (2016) 3, 8 [67] Schymkowitz, J., Borg, J., Stricher, F., Nys, R., Rousseau, F., Serrano, L.: The foldx web server: an online force field. Nucleic acids research 33(suppl_2), W382–W388 (2005) 3, 7, 8, 18 [68] Shroff, R., Cole, A.W., Diaz, D.J., Morrow, B.R., Donnell, I., Annapareddy, A., Gollihar, J., Ellington, A.D., Thyer, R.: Discovery of novel gain-of-function mutations guided by structure-based deep learning. ACS synthetic biology 9(11), 2927–2935 (2020) 3 [69] Starr, T.N., Thornton, J.W.: Epistasis in protein evolution. Protein science 25(7), 1204–1218 (2016) 1 [70] Steinegger, M., Soding, J.: Mmseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nature biotechnology 35(11), 1026–1028 (2017) 6 [71] Stemmer, W.P.: Rapid evolution of a protein in vitro by dna shuffling. Nature 370(6488), 389–391 (1994) 2, 3 [72] Tsuboyama, K., Dauparas, J., Chen, J., Laine, E., Mohseni Behbahani, Y., Weinstein, J.J., Mangan, N.M., Ovchinnikov, S., Rocklin, G.J.: Mega-scale experimental analysis of protein folding stability in biology and protein design. bioRxiv pp. 2022–12 (2022) 1, 2, 6, 16 [73] Tung, J.W., Heydari, K., Tirouvanziam, R., Sahaf, B., Parks, D.R., Herzenberg, L.A., Herzenberg, L.A.: Modern flow cytometry: a practical approach. Clinics in laboratory medicine 27(3), 453–468 (2007) 1 [74] Umerenkov, D., Shashkova, T.I., Strashnov, P.V., Nikolaev, F., Sindeeva, M., Ivanisenko, N.V., Kardymon, O.L.: Prostata: Protein stability assessment using transformers. bioRxiv pp. 2022–12 (2022) 3, 4, 6, 7, 8, 18, 19 [75] Wittmann, B.J., Johnston, K.E., Wu, Z., Arnold, F.H.: Advances in machine learning for directed evolution. Current opinion in structural biology 69, 11–18 (2021) 3 [76] Wittmann, B.J., Yue, Y., Arnold, F.H.: Informed training set design enables efficient machine learningassisted directed protein evolution. Cell Systems 12(11), 1026–1045 (2021) 3 [77] Worth, C.L., Preissner, R., Blundell, T.L.: Sdm—a server for predicting effects of mutations on protein stability and malfunction. Nucleic acids research 39(suppl_2), W215–W222 (2011) 3, 8 [78] Wu, S., Snajdrova, R., Moore, J.C., Baldenius, K., Bornscheuer, U.T.: Biocatalysis: Enzymatic Synthesis for Industrial Applications. Angewandte Chemie - International Edition 60(1), 88–119 (2021). https://doi.org/10.1002/anie.202006648 1 [79] Yeung, N., Lin, Y.W., Gao, Y.G., Zhao, X., Russell, B.S., Lei, L., Miner, K.D., Robinson, H., Lu, Y.: Rational design of a structural and functional nitric oxide reductase. Nature 462(7276), 1079–1082 (2009) 1, 2 [80] Zhou, B., Lv, O., Yi, K., Xiong, X., Tan, P., Hong, L., Wang, Y.G.: Accurate and definite mutational effect prediction with lightweight equivariant graph neural networks. arXiv preprint arXiv:2304.08299 (2023) 3 [81] Zhou, Y., Pan, Q., Pires, D.E., Rodrigues, C.H., Ascher, D.B.: Ddmut: predicting effects of mutations on protein stability using deep learning. Nucleic Acids Research p. gkad472 (2023) 3

Predicting a Protein’s Stability
under a Million Mutations: Conclusion, Acknowledgement & References

Introduction Would you like to see Mona Lisa smile like a witch? Or would you like the girl with the pearl earring to wink and smile? Google has just dropped a video generation model named Lumiere [1] that's capable of doing all of it for you. Though it's primarily a text-to-video model, it's able to do much more than that. Given a reference image with a prompt, it can stylize your videos by copying the style of the reference image to your videos. You can even edit your videos with just a single prompt. The model is Lumiaire. It's even able to animate objects within a user-specified region in an image, a technique called Cinemagraphs. When it comes to inpainting, Lumiere is even able to reason about a completely missing object such as a cake in this example. It all boils down to a diffusion model with a novel space-time U-Net architecture [3]. It's a customization of the U-Net architecture to solve the problem of temporal consistency which is quite prevalent in video generation models. Visual Explanation A visual explanation of the Lumiere paper, the model architecture, and the results are available. https://youtu.be/QcYa6xxsMTU?si=IBpUqjPtj8lJDHo0&embedable=true So, What Is Temporal Consistency? We all know that videos are a sequence of images. So, let's take a sequence of images shown in the top row in the above figure. If we narrow down just one row in the image, which is indicated by the green line going from left to right, we need to see a smooth transition in the pixel values between images in the sequence. If the transition is smooth, then we won’t see a jumping effect while we are watching the video. For example, if we take Stable Video Diffusion [2], and see the video of an astronaut walking on the moon (above), we can see that his hands just disappear between frames. In other words, the temporal consistency between frames is lacking. This kind of temporal inconsistency between time and the intensity in the x direction can be plotted as the X-T Slice as highlighted in the figure above. And if there is temporal inconsistency, it is highlighted in the plot of the X-T slice. Lumiere addresses this problem by introducing a space-time diffusion model and a modified U-Net architecture present in the diffusion model. Pipeline of a Text to Video Before looking into the details, let's start with the typical pipeline of a Text-to-video generation model. These pipelines sample every 5th frame as the keyframes from the input video sequence and train a base model that can generate these keyframes at a resolution as low as 128 by 128 at just 3 frames per second. Then temporal Super Resolution is used to increase the frame rate by predicting the intermediate frames. So the frame rate now becomes 16 frames per second. The spatial resolution of these frames is then increased to say 1024 by 1024 by a spatial superresolution network (SSR) which finally leads to our generated video. The base model used in this framework is usually a diffusion model which in turn has a U-Net inside it. The Proposed Pipeline of Lumiere The proposed pipeline of Lumiere on the other hand processes all the frames at once without dropping frames. To cope with the computational cost of processing all the frames, the architecture of the base diffusion model is modified to come up with the space-time UNet architecture or STUNet. As STUNet deals with all the input frames, the need for Temporal Super Resolution or TSR is eliminated. So, the pipeline still has the spatial super-resolution or SSR. But the novelty is the introduction of MultiDiffusion. From U-Net to STUNet Let's quickly review the U-Net before looking into the space-time U-Net. The input to the U-Net is a 3 3-dimensional image with Width W, Height H, and channels RGB. After every double convolution stage of the U-Net, we apply max pooling to downsample or reduce the spatial dimension of the features. This spatial dimension reduction step is indicated by the red arrows. Similarly, during the decoder stage, there are up convolutions to increase or upsample the resolution back to the size of the input. When it comes to videos, we have an additional dimension in the input which is time. so space-time U-Net proposes to downsample and up-sample the video not only in the spatial dimension but also in the dimension of time T. This is the main idea of temporal resizing and is the main contribution of this Lumiere paper. For resizing, they use 3D pooling instead of 2D pooling as the input now has an additional dimension. Like me, you may be surprised by the simplicity of the idea. The authors themselves have mentioned in the paper: Surprisingly, this design choice has been overlooked by previous T2V models, which follows the convention to include only spatial down and up-sampling operations in the architecture, and maintain a fixed temporal resolution across the network. Implementation Let's get to some of the nuances of the implementation. They use factorized convolution introduced in this paper called Video Diffusion Models. The idea is to change each 2D convolution into a space-only 3D convolution, for instance, by changing each 3x3 convolution into a 1x3x3 convolution. For the attention, after each spatial attention block, we insert a temporal attention block that performs attention over the first axis and treats the spatial axes as batch axes. With those two changes, the factorized convolution blocks are added to the pre-trained model, and only the additional layers are trained with the pre-trained layer weights fixed. The second novelty of the paper is the MultiDiffusion introduced during the spatial superresolution. If you take the video generation models before lumiere, the spatial superresolution model takes a sequence of frames. However, the sequences were not overlapping. For example, the first 8 frames and the next 8 frames taken as input by the SSR module are separate without any overlap. But when it comes to Lumiere, the first 8 frames and the second 8 frames have an overlap of two frames. By doing this, the spatial super-resolution model seems to achieve smooth transitions between the temporal segments. This is what is referred to as multiDiffusion in the paper. Applications Coupling the two proposed techniques along with the absence of a cascaded diffusion model architecture which is prevalent in prior architectures such as imagen video, leads to quite a few varied applications. For example: The model can convert text to videos with prompts such as “an astronaut walking on planet Mars making a detour around his base” or “a dog driving a car wearing funny sunglasses.” It can convert images to videos along with a text prompt such as “A girl winking and smiling.” It can stylize generation with a reference image and a text prompt such as “a bear dancing.” Talking of Cinemagraphs, it can animate regions selected by the user such as fire or steam. It can even edit dresses that people are wearing with just a single prompt. Evaluation To evaluate the model quantitatively, the model was run through a user study where the users compared the results of the proposed model against some state-of-the-art models such as Pika, ZeroScope, or stable video diffusion. The results indicate that the users preferred the Lumiere model both in the case of text to video and image to video. Conclusion So, to conclude, other than all the promotional video stunts, such as the smiling Monalisa, the contribution of this paper is fairly simple. Put in one line, the paper introduces a downsampling of the temporal channel. This coupled with MultiDiffusion, which is nothing but having overlapped frames fed into the superresolution model, generates high-fidelity videos that are temporally consistent. What I would have liked to see is some ablation studies in the paper showing the results with and without the multiDiffusion process. That brings us to the end of this article. The next time someone talks to you about Lumiere, you know what to say in one line. I hope that provided some insight into the Lumiere model. I will see you in my next, until then, take care… References [1] Omer Bar-Tal, Hila Chefer, Omer Tov, Charles Herrmann, Roni Paiss, Shiran Zada, Ariel Ephrat, Junhwa Hur, Yuanzhen Li, Tomer Michaeli, Oliver Wang, Deqing Sun, Tali Dekel, Inbar Mosseri, (2024), arXiv preprint. A Space-Time Diffusion Model for Video Generation [2] Andreas Blattmann, Tim Dockhorn, Sumith Kulal, Daniel Mendelevitch, Maciej Kilian, Dominik Lorenz, Yam Levi, Zion English, Vikram Voleti, Adam Letts, Varun Jampani, Robin Rombach, (2023), arXiv preprint. Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets [3] Olaf Ronneberger, Philipp Fischer and Thomas Brox, (2015), International Conference on Medical Image Computing and Computer-Assisted Intervention. U-Net: Convolutional Networks for Biomedical Image Segmentation Also published here

Google Unveils Its Most Promising Text-to-Video Model Yet: Lumiere

গুগল এখনও পর্যন্ত তার সবচেয়ে প্রতিশ্রুতিশীল পাঠ্য-টু-ভিডিও মডেল উন্মোচন করেছে: লুমিয়ের

Google stellt sein bisher vielversprechendstes Text-to-Video-Modell vor: Lumiere

Google presenta su modelo de texto a video más prometedor hasta el momento: Lumiere

Google dévoile son modèle de conversion texte-vidéo le plus prometteur à ce jour : Lumiere

Google ने अपना अब तक का सबसे आशाजनक टेक्स्ट-टू-वीडियो मॉडल लॉन्च किया: लुमिएरे

Google、これまでで最も有望なテキストからビデオへのモデル「Lumiere」を発表

Google은 지금까지 가장 유망한 텍스트-비디오 모델인 Lumiere를 공개했습니다.

Google revela seu modelo de conversão de texto em vídeo mais promissor: Lumiere

Google представляет свою самую многообещающую модель преобразования текста в видео: Lumiere

Google Şimdiye Kadarki En Umut Verici Metinden Videoya Modelini Açıkladı: Lumiere

Google công bố mô hình chuyển văn bản thành video hứa hẹn nhất: Lumiere

谷歌推出迄今为止最有前途的文本转视频模型：Lumiere

Generate and Pray: Using SALLMS to Evaluate the Security: Conclusion & References

This paper is available on arxiv under CC 4.0 license. Authors: (1) Mohammed Latif Siddiq, Department of Computer Science and Engineering, University of Notre Dame, Notre Dame; (2) Joanna C. S. Santos, Department of Computer Science and Engineering, University of Notre Dame, Notre Dame. Table of Links Abstract & Introduction Background and Motivation Our Framework: SALLM Experiments Results Limitations and Threats to the Validity Related Work Conclusion & References 3 Our Framework: SALLM Fig. 2 shows an overview of our framework and how it was created. Our framework consists of three major components: a dataset of prompts, an evaluation environment to execute the code, configurable assessment techniques, and novel evaluation metrics. Each of these components are further described in the next subsections. 3.1 Dataset of Prompts To create an effective security benchmarking framework, we first needed a high-quality dataset of prompts. Although there are two datasets available (LLMSecEval and SecurityEval) [63, 67] they have many problems. First, one of them (LLMSecEval [67]) is a dataset of natural language prompts, which is a format that not all code LLMs support. Second, SecurityEval has several prompts that do not execute and lack test cases to verify both its functional correctness and the presence of vulnerabilities in the generated code. Therefore, we aimed to create a manually curated and high-quality dataset of prompts to fulfill our needs. The creation of the framework’s dataset of prompts involved two steps. We first retrieved code snippets and texts from different sources. Then, we manually crafted a prompt from the retrieved code snippets. In the following sections, we presented the approach to collecting and crafting the prompts for our framework. 3.1.1 Code Snippets Collection Our goal was to create a prompt dataset that reflects the real-life security-centric needs of software developers. To build this dataset, we mined code snippets from the following sources: [1] is a popular question-answering website among developers. Users describe their problems, and others try to solve them via discussion. We retrieved the top most popular questions with an accepted answer containing the word “unsafe” or “vulnerable”, and that is tagged as a Python-related question. From these 500 questions, we applied a set of inclusion and exclusion criteria. The inclusion criteria were: the question has to ( ) explicitly ask “how to do X” in Python; ( ) include code in its body; ( ) have an accepted answer that includes code. We excluded questions that were ( ) open-ended and asking for best practices/guidelines for a specific problem in Python; ( ) related to finding a specific API/module for a given task; ( ) related to errors due to environment configuration (e.g., missing dependency library); ( ) related to configuring libraries/API; ( ) syntax-specific types of questions. By applying the criteria above to these 500 questions, we obtained a total of 13 code snippets. - StackOverflow 500 1 2 3 1 2 3 4 5 - The [43] is a community effort to create a list of vulnerability types (weaknesses). Each weakness may also include demonstrative examples, which are code snippets written in different programming languages (e.g., C, PHP, Java, Python, etc.) containing a vulnerability that an attacker can exploit. We retrieved the list of all CWEs and extracted all demonstrative examples written in Python. As a result, we retrieved a total of code snippet. As not all CWEs have examples in Python, we also created examples ourselves based on the CWE descriptions. We created a total of coding snippets. Common Weakness Enumeration (CWE) 1 35 - [26] is a static analysis tool that detects vulnerabilities by making queries over a source code graph representation. This tool’s documentation includes vulnerable examples in different programming languages. Thus, we retrieved a total of vulnerable Python samples from CodeQL’s documentation. CodeQL 35 - [57] is a set of pre-defined patterns used by the SonarQube tool to analyze and assess the quality of a code. These rules cover a wide range of coding standards, best practices, and vulnerabilities. Thus, we retrieved a total of Python examples provided in the documentation for the Python-related vulnerability rules. Sonar Rules 9 For each collected sample from these sources, we extract their (i.e., the raw text/code collected from the source), and title, content source URL. 3.1.2 Prompts Creation After collecting the samples, we went through them manually and created a well-structured prompt. Each prompt is a function/method signature that describes a security-relevant coding task, i.e., a problem in which there are one or more possible solutions that are functionally correct but insecure. The prompt also includes the required relevant module imports. For each prompt, we assign a unique identifier and manually classify it with a CWE-ID. For each prompt, we also create an example of an insecure solution, i.e., a functionally correct solution, but that has a vulnerability. This way, our dataset is not only a collection of prompts but also includes executable vulnerable programs. Listing 1 shows an example of a prompt in our dataset. This prompt instructs the model to use the GitHub search REST API to obtain the profile information for a given username. The first 15 lines (highlighted) include the necessary context and a docstring describing the task to complete. The rest of the code is a possible insecure solution for this prompt. As observed, this task has the risk of a model generating a code prone to server-side request forgery attacks (CWE918). 3.2 Code Generation Our framework provides as input to an LLM the prompts in its dataset. For each prompt, our framework requests the LLM to generate k solutions to the prompt (where k can be specified). Each generated code is saved in a Python script file. As prior studies have shown, LLMs can generate code with simple compilation errors (e.g., missing the end curly bracket for a code block) [15, 61, 64]. Hence, our framework includes a static filtering phase responsible for automatically fixing syntax errors through three rules and removing generated code snippets that are not executable (even after attempting to fix it). (a) (b) The rules used to repair compilation errors automatically work as follows: • models, such as ChatGPT, can include explanations (i.e., natural language text) before and/or after the generated code and then enclose the code within backticks (i.e., ```code```). Thus, the first heuristic removes the text written in natural language and only keeps the generated code in the first block of text delimited by three backticks. H1: Code Block Extraction Conversation-style In the generated code by an LLM, the initial prompt can be absent. This absence will result in syntax errors as the required function/class signature and imported libraries are missing. Thus, we prepend the generated code with the original prompt. • H2: Prompt Addition This heuristic removes any extra code after having the following patterns (including these patterns): '\ndef', '\nif', '\n@app', "\n'''", '\nclass'. These patterns usually indicate that the code generation model is done generating code. • H3: Extra Code Removal 3.3 Systematic Model Assessment To systematically evaluate the security of the code produced by a model, the infrastructure has two major components: a set and a assessment technique security checker. 3.3.1 Assessment Techniques Our framework has an evaluation environment with the runtime configuration to execute and verify a generated code security. This environment is composed of docker images for the prompts [24]. Each prompt is released in a standalone docker image with all the required dependencies to run the code. During this evaluation process, the generated code is placed into the evaluation environment and executed in a sandbox to prevent unsafe behavior. Next, our framework evaluates the security of the code generated by LLMs using two assessment techniques: dynamic-based assessment, and static-based assessment. Since each prompt in our dataset has a CWE-ID and example of an insecure solution, the expected functional and insecure output of a function from a given input is known. If a model generates an insecure code, the behavior will differ from the expected one. Deviations from the expected (secure) behavior can be used to check if the source code is susceptible to vulnerabilities. Thus, this assessment includes the development of test cases that include assertions for the expected functional and security properties. Specifically, we write a unit test for each of the prompts in our dataset using the unittest module. Each unit test class has two test methods; one verifies the functional behavior of the generated code, whereas the other checks the security behavior of the program. Dynamic-Based Assessment To clarify, the code in Listing 2 shows the test case created for the prompt shown in Listing 1. This test class has two methods (test_security and test_functionality). The first one checks whether the solution is functionally correct by making two HTTP GET requests to the flask application; one request passes as input a username that exists (octocat), and the other one requests with a non-existent username (jn2Elxop0). Then, the code asserts whether the generated successfully retrieves the metadata for these users. The second test method checks whether the generated code is prone to server-side request forgery attacks. It is important to highlight that when we generate the code, we save it in a file named A_cweID_promptID.py. This way, the test case can import the generated code for the test (as shown in line 5). Programs may use built-in or external libraries/modules/functions (henceforth, simply “APIs”) that are inherently unsafe. Since these unsafe APIs are used in the wild, they are likely to be part of the training data used by LLMs. Thus, there is a risk that the model may use these unsafe APIs in the generated code. Static-Based Assessment For example, the source code shown in Listing 3 uses the md5 hash function. This weak hash reasonably allows an adversary to determine the original input through pre-image attacks. Though this is a weak hash function and vulnerable to security attacks, it still exists due to support for backward compatibility. This is an example of a source code with CWE-328: Use of Weak Hash [12]. These API patterns can be detected using the static analysis of the source code. Our framework uses CodeQL [26] for unsafe API matching. CodeQL is a static code analysis engine designed to automatically check for vulnerabilities in a project by executing QL queries against a database generated from the source code. CodeQL can be used to match the function of the function call. For example, the QL query shown in Listing 4 is taken from the CodeQL repository, which can match a method name and check if it is called. Another thing is that several vulnerability types (i.e., injection vulnerabilities) are caused by untrusted data flows [39, 72]. These weaknesses are traditionally detectable through taint analysis, which is a technique that tracks flows of sources of potentially untrusted (tainted) data (e.g., parameters in HTTP requests) to sensitive program areas (sinks) [59]. Taint analysis can be performed at compile time (static) or runtime (dynamic). For example, the code in listing 5 contains an OS Command Injection (CWE-78) [13]. This function uses os.system but does not check the input, which may come from an untrusted source and potentially leads to OS injection. In these cases, our framework uses CodeQL to perform static analysis to track the taint variables and check if they reach a sink method (e.g., os.system). To illustrate, listing 6 represents a taint tracking code where the user input is given using a network call and dumps the untrusted data in a file. We used this taint tracking system from CodeQL to measure whether the generated code is vulnerable. 3.3.2 Security Checker Code generation models produce multiple potential solutions (i.e., code snippets) for a given prompt. Models are commonly evaluated using the pass@k metric [10, 32]. This metric aims to evaluate the probability that at least one out of k generated samples are functionally correct. To evaluate the pass@k, we generate n samples per prompt (n ≥ k), count the number of samples c that are functionally correct (c ≤ n), and calculate the unbiased estimator from Kulal et al. [32]: (3) The secure@k metric measures the probability that all code snippets out of k samples are secure (where s is the number of secure samples generated). That is, the prompt is considered secure if all of the generated code in the top-k passes our assessment techniques. To clarify, consider that we have 10 prompts, and a model generates 10 outputs for each problem described in a prompt. If our assessment technique finds that out of 10 outputs, 6 prompts have all of the generated code passing the assessment techniques, then the secure@k score will be 60%. The vulnerable@k metric measures the probability that at least one code snippets out of k samples are vulnerable (where v is the number of vulnerable generated samples). We consider a prompt to be vulnerable if any of the top-k generated samples has a vulnerability detected by our assessment techniques. For this metric, the model is better if the vulnerable@k score is lower. Since calculating Kulal et al. [32] estimator directly results in large numbers and numerical instability [35], to compute the metrics, we used a numerically stable implementation from Chen et al. [10]. This numerically stable implementation simplifies the expression and evaluates the product term-byterm. ▶ Estimating the pass@k, secure@k, and vulnerable@k:

Generate and Pray: Using SALLMS to Evaluate the Security of LLM Generated Code: Our Framework: SALLM

Twenty Thousand Leagues under the Sea by Jules Verne, is part of the HackerNoon Books Series. You can jump to any chapter in this book . A SHIFTING REEF here CHAPTER I: A SHIFTING REEF The year 1866 was signalised by a remarkable incident, a mysterious and puzzling phenomenon, which doubtless no one has yet forgotten. Not to mention rumours which agitated the maritime population and excited the public mind, even in the interior of continents, seafaring men were particularly excited. Merchants, common sailors, captains of vessels, skippers, both of Europe and America, naval officers of all countries, and the Governments of several states on the two continents, were deeply interested in the matter. For some time past, vessels had been met by “an enormous thing,” a long object, spindle-shaped, occasionally phosphorescent, and infinitely larger and more rapid in its movements than a whale. The facts relating to this apparition (entered in various log-books) agreed in most respects as to the shape of the object or creature in question, the untiring rapidity of its movements, its surprising power of locomotion, and the peculiar life with which it seemed endowed. If it was a cetacean, it surpassed in size all those hitherto classified in science. Taking into consideration the mean of observations made at divers times,—rejecting the timid estimate of those who assigned to this object a length of two hundred feet, equally with the exaggerated opinions which set it down as a mile in width and three in length,—we might fairly conclude that this mysterious being surpassed greatly all dimensions admitted by the ichthyologists of the day, if it existed at all. And that it exist was an undeniable fact; and, with that tendency which disposes the human mind in favour of the marvellous, we can understand the excitement produced in the entire world by this supernatural apparition. As to classing it in the list of fables, the idea was out of the question. did On the 20th of July, 1866, the steamer , of the Calcutta and Burnach Steam Navigation Company, had met this moving mass five miles off the east coast of Australia. Captain Baker thought at first that he was in the presence of an unknown sandbank; he even prepared to determine its exact position, when two columns of water, projected by the inexplicable object, shot with a hissing noise a hundred and fifty feet up into the air. Now, unless the sandbank had been submitted to the intermittent eruption of a geyser, the had to do neither more nor less than with an aquatic mammal, unknown till then, which threw up from its blow-holes columns of water mixed with air and vapour. Governor Higginson Governor Higginson Similar facts were observed on the 23rd of July in the same year, in the Pacific Ocean, by the , of the West India and Pacific Steam Navigation Company. But this extraordinary cetaceous creature could transport itself from one place to another with surprising velocity; as, in an interval of three days, the and the had observed it at two different points of the chart, separated by a distance of more than seven hundred nautical leagues. Columbus Governor Higginson Columbus Fifteen days later, two thousand miles farther off, the , of the Compagnie-Nationale, and the , of the Royal Mail Steamship Company, sailing to windward in that portion of the Atlantic lying between the United States and Europe, respectively signalled the monster to each other in 42° 15′ N. lat. and 60° 35′ W. long. In these simultaneous observations they thought themselves justified in estimating the minimum length of the mammal at more than three hundred and fifty feet, as the and were of smaller dimensions than it, though they measured three hundred feet over all. Helvetia Shannon Shannon Helvetia Now the largest whales, those which frequent those parts of the sea round the Aleutian, Kulammak, and Umgullich islands, have never exceeded the length of sixty yards, if they attain that. These reports arriving one after the other, with fresh observations made on board the transatlantic ship , a collision which occurred between the of the Inman line and the monster, a directed by the officers of the French frigate , a very accurate survey made by the staff of Commodore Fitz-James on board the , greatly influenced public opinion. Light-thinking people jested upon the phenomenon, but grave practical countries, such as England, America, and Germany, treated the matter more seriously. Pereire Etna procès verbal Normandie Lord Clyde In every place of great resort the monster was the fashion. They sang of it in the cafés, ridiculed it in the papers, and represented it on the stage. All kinds of stories were circulated regarding it. There appeared in the papers caricatures of every gigantic and imaginary creature, from the white whale, the terrible “Moby Dick” of hyperborean regions, to the immense kraken whose tentacles could entangle a ship of five hundred tons, and hurry it into the abyss of the ocean. The legends of ancient times were even resuscitated, and the opinions of Aristotle and Pliny revived, who admitted the existence of these monsters, as well as the Norwegian tales of Bishop Pontoppidan, the accounts of Paul Heggede, and, last of all, the reports of Mr. Harrington (whose good faith no one could suspect), who affirmed that, being on board the , in 1857, he had seen this enormous serpent, which had never until that time frequented any other seas but those of the ancient “ .” Castillan Constitutionnel Then burst forth the interminable controversy between the credulous and the incredulous in the societies of savants and the scientific journals. “The question of the monster” inflamed all minds. Editors of scientific journals, quarrelling with believers in the supernatural, spilled seas of ink during this memorable campaign, some even drawing blood; for, from the sea-serpent they came to direct personalities. For six months war was waged with various fortune in the leading articles of the Geographical Institution of Brazil, the Royal Academy of Science of Berlin, the British Association, the Smithsonian Institution of Washington, in the discussions of the “Indian Archipelago,” of the Cosmos of the Abbé Moigno, in the Mittheilungen of Petermann, in the scientific chronicles of the great journals of France and other countries. The cheaper journals replied keenly and with inexhaustible zest. These satirical writers parodied a remark of Linnæus, quoted by the adversaries of the monster, maintaining “that nature did not make fools,” and adjured their contemporaries not to give the lie to nature, by admitting the existence of krakens, sea-serpents, “Moby Dicks,” and other lucubrations of delirious sailors. At length an article in a well-known satirical journal by a favourite contributor, the chief of the staff, settled the monster, like Hippolytus, giving it the death-blow amidst an universal burst of laughter. Wit had conquered science. During the first months of the year 1867 the question seemed buried, never to revive, when new facts were brought before the public. It was then no longer a scientific problem to be solved, but a real danger seriously to be avoided. The question took quite another shape. The monster became a small island, a rock, a reef, but a reef of indefinite and shifting proportions. On the 5th of March, 1867, the , of the Montreal Ocean Company, finding herself during the night in 27° 30′ lat. and 72° 15′ long., struck on her starboard quarter a rock, marked in no chart for that part of the sea. Under the combined efforts of the wind and its four hundred horse-power, it was going at the rate of thirteen knots. Had it not been for the superior strength of the hull of the , she would have been broken by the shock and gone down with the 237 passengers she was bringing home from Canada. Moravian Moravian The accident happened about five o’clock in the morning, as the day was breaking. The officers of the quarter-deck hurried to the after-part of the vessel. They examined the sea with the most scrupulous attention. They saw nothing but a strong eddy about three cables’ length distant, as if the surface had been violently agitated. The bearings of the place were taken exactly, and the continued its route without apparent damage. Had it struck on a submerged rock, or on an enormous wreck? they could not tell; but on examination of the ship’s bottom when undergoing repairs, it was found that part of her keel was broken. Moravian This fact, so grave in itself, might perhaps have been forgotten like many others if, three weeks after, it had not been re-enacted under similar circumstances. But, thanks to the nationality of the victim of the shock, thanks to the reputation of the company to which the vessel belonged, the circumstance became extensively circulated. The 13th of April, 1867, the sea being beautiful, the breeze favourable, the , of the Cunard Company’s line, found herself in 15° 12′ long. and 45° 37′ lat. She was going at the speed of thirteen knots and a half. Scotia At seventeen minutes past four in the afternoon, whilst the passengers were assembled at lunch in the great saloon, a slight shock was felt on the hull of the , on her quarter, a little aft of the port-paddle. Scotia The had not struck, but she had been struck, and seemingly by something rather sharp and penetrating than blunt. The shock had been so slight that no one had been alarmed, had it not been for the shouts of the carpenter’s watch, who rushed on to the bridge, exclaiming, “We are sinking! we are sinking!” At first the passengers were much frightened, but Captain Anderson hastened to reassure them. The danger could not be imminent. The , divided into seven compartments by strong partitions, could brave with impunity any leak. Captain Anderson went down immediately into the hold. He found that the sea was pouring into the fifth compartment; and the rapidity of the influx proved that the force of the water was considerable. Fortunately this compartment did not hold the boilers, or the fires would have been immediately extinguished. Captain Anderson ordered the engines to be stopped at once, and one of the men went down to ascertain the extent of the injury. Some minutes afterwards they discovered the existence of a large hole, of two yards in diameter, in the ship’s bottom. Such a leak could not be stopped; and the , her paddles half submerged, was obliged to continue her course. She was then three hundred miles from Cape Clear, and after three days’ delay, which caused great uneasiness in Liverpool, she entered the basin of the company. Scotia Scotia Scotia The engineers visited the , which was put in dry dock. They could scarcely believe it possible; at two yards and a half below water-mark was a regular rent, in the form of an isosceles triangle. The broken place in the iron plates was so perfectly defined that it could not have been more neatly done by a punch. It was clear, then, that the instrument producing the perforation was not of a common stamp; and after having been driven with prodigious strength, and piercing an iron plate 1-3/8 inches thick, had withdrawn itself by a retrograde motion truly inexplicable. Scotia Such was the last fact, which resulted in exciting once more the torrent of public opinion. From this moment all unlucky casualties which could not be otherwise accounted for were put down to the monster. Upon this imaginary creature rested the responsibility of all these shipwrecks, which unfortunately were considerable; for of three thousand ships whose loss was annually recorded at Lloyd’s, the number of sailing and steam ships supposed to be totally lost, from the absence of all news, amounted to not less than two hundred! Now, it was the “monster” who, justly or unjustly, was accused of their disappearance, and, thanks to it, communication between the different continents became more and more dangerous. The public demanded peremptorily that the seas should at any price be relieved from this formidable cetacean. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Jules Verne (1994). Twenty Thousand Leagues under the Sea. Urbana, Illinois: Project Gutenberg. Retrieved October 2022 https://www.gutenberg.org/files/164/164-h/164-h.htm This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at . www.gutenberg.org https://www.gutenberg.org/policy/license.html

A SHIFTING REEF

NEW ALLIES

TURJUN, THE PANTHAN

KULA

joinkula.io

-- employees

Since n.d.

COMPANY RANKING

KULA

EVERGREEN INDEX #5971

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

Kula WIKI

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

KULA

joinkula.io

-- employees

Since n.d.

COMPANY RANKING

KULA

EVERGREEN INDEX #5971

RELATED COMPANIESRANK

STORIES ABOUT KULA

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

Kula WIKI

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