diff --git a/data/sampled_jsons/2405.19550_Section_6.2_MATH_Pythia-1B_poor_performance_exploration.jsonl b/data/sampled_jsons/2405.19550_Section_6.2_MATH_Pythia-1B_poor_performance_exploration.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b2255a654334fc51bea1a0609f2bf3e775e08e63
--- /dev/null
+++ b/data/sampled_jsons/2405.19550_Section_6.2_MATH_Pythia-1B_poor_performance_exploration.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Integrated II Answers and Solutions | Mathleaks", "date": "", "ddg_snippet": "Mathleaks offers help with learning-focused solutions for textbooks in Integrated Mathematics II, 9th and 10th grade. They include theory, hints, answers and cover textbooks from publishers such as Houghton Mifflin, McGraw Hill, CPM, Big Ideas, and Pearson.", "subpage_snippet": "", "source": "mathleaks.com", "link": "https://mathleaks.com/courses/integrated-ii", "content": "Mathleaks offers help with learning-focused solutions for textbooks in Integrated Mathematics II, 9th and 10th grade. They include theory, hints, answers and cover textbooks from publishers such as Houghton Mifflin, McGraw Hill, CPM, Big Ideas, and Pearson."}
+{"idx": 1, "title": "Factors associated with high school learners' poor performance: a ...", "date": "", "ddg_snippet": "This study, using anon-experimental, exploratory and descriptive method, established learners' and educators' views about factors that contribute to poor performance in mathematics and physical ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/228619581_Factors_associated_with_high_school_learners'_poor_performance_a_spotlight_on_mathematics_and_physical_science", "content": "This study, using anon-experimental, exploratory and descriptive method, established learners' and educators' views about factors that contribute to poor performance in mathematics and physical ..."}
+{"idx": 2, "title": "Maths IA - 300 Maths Exploration Topics - IB Maths Resources from ...", "date": "", "ddg_snippet": "Maths IA - 300 Maths Exploration Topics: Scroll down this page to find over 300 examples of maths IA exploration topics and ideas for IB mathematics students doing their internal assessment (IA) coursework. Topics include Algebra and Number (proof), Geometry, Calculus, Statistics and Probability, Physics, and links with other subjects.", "subpage_snippet": "", "source": "ibmathsresources.com", "link": "https://ibmathsresources.com/maths-ia-maths-exploration-topics/", "content": "Maths IA - 300 Maths Exploration Topics: Scroll down this page to find over 300 examples of maths IA exploration topics and ideas for IB mathematics students doing their internal assessment (IA) coursework. Topics include Algebra and Number (proof), Geometry, Calculus, Statistics and Probability, Physics, and links with other subjects."}
+{"idx": 3, "title": "Big Ideas Math Algebra 2 Answers Chapter 6 ... - CCSS Math Answers", "date": "", "ddg_snippet": "The Big Ideas Math Book Algebra 2 Ch 6 Exponential and Logarithmic Functions include Questions from Exercises 6.1 to 6.7, Review Tests, Chapter Tests, Cumulative Assessments, etc. Enhance your subject knowledge taking the help of the Big Ideas Math Algebra 2 Answers Chapter 6 Exponential and Logarithmic Functions prepared by subject experts.", "subpage_snippet": "", "source": "ccssmathanswers.com", "link": "https://ccssmathanswers.com/big-ideas-math-algebra-2-answers-chapter-6/", "content": "The Big Ideas Math Book Algebra 2 Ch 6 Exponential and Logarithmic Functions include Questions from Exercises 6.1 to 6.7, Review Tests, Chapter Tests, Cumulative Assessments, etc. Enhance your subject knowledge taking the help of the Big Ideas Math Algebra 2 Answers Chapter 6 Exponential and Logarithmic Functions prepared by subject experts."}
+{"idx": 4, "title": "PDF Factors Contributing to Poor Learner Performance in Mathematics: a Case ...", "date": "", "ddg_snippet": "Introduction Poor learner performance in mathematics has been a global concern that has prompted developing countries to participate in initiatives to bring positive change in their communities (Sinyosi, 2015). Mathematics excellence can bring positive change in developing countries to develop their education systems for shaping the future and prospects of young people; to develop ...", "subpage_snippet": "", "source": "files.eric.ed.gov", "link": "https://files.eric.ed.gov/fulltext/EJ1301930.pdf", "content": "Introduction Poor learner performance in mathematics has been a global concern that has prompted developing countries to participate in initiatives to bring positive change in their communities (Sinyosi, 2015). Mathematics excellence can bring positive change in developing countries to develop their education systems for shaping the future and prospects of young people; to develop ..."}
+{"idx": 5, "title": "Causes of Poor Performance in Mathematics from the", "date": "", "ddg_snippet": "ultimately causes as one of the most important factors for poor performance in mathematics [11]. By developing and raising level of student's interest and involvement means how much time, energy and effort they devote towards achieving high goals in mathematics as learning process is fast and instinctual on one hand while on the other hand ...", "subpage_snippet": "", "source": "asrjetsjournal.org", "link": "https://asrjetsjournal.org/index.php/American_Scientific_Journal/article/download/1212/654/2858", "content": "ultimately causes as one of the most important factors for poor performance in mathematics [11]. By developing and raising level of student's interest and involvement means how much time, energy and effort they devote towards achieving high goals in mathematics as learning process is fast and instinctual on one hand while on the other hand ..."}
+{"idx": 6, "title": "(Pdf) Factors Contributing to Poor Learner Performance in Mathematics ...", "date": "", "ddg_snippet": "Introduction Poor learner performance in mathematics has been a global concern that has prompted developing countries to participate in initiatives to bring positive change in their communities", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/352394333_FACTORS_CONTRIBUTING_TO_POOR_LEARNER_PERFORMANCE_IN_MATHEMATICS_A_CASE_OF_SELECTED_SCHOOLS_IN_MPUMALANGA_PROVINCE_SOUTH_AFRICA", "content": "Introduction Poor learner performance in mathematics has been a global concern that has prompted developing countries to participate in initiatives to bring positive change in their communities"}
+{"idx": 7, "title": "Stress-Testing Capability Elicitation With Password-Locked Models", "date": "", "ddg_snippet": "For code generation (right), when using a Deepseek 7B (Bi et al., 2024) model fine-tuned to imitate poor answers generated by Pythia-1B (Biderman et al., 2023), both supervised fine-tuning (SFT) on few demonstrations and reinforcement learning (RL) recover most of the coding capabilities of Deepseek 7B.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2405.19550", "content": "For code generation (right), when using a Deepseek 7B (Bi et al., 2024) model fine-tuned to imitate poor answers generated by Pythia-1B (Biderman et al., 2023), both supervised fine-tuning (SFT) on few demonstrations and reinforcement learning (RL) recover most of the coding capabilities of Deepseek 7B."}
+{"idx": 8, "title": "Samarth0710/neurips-2024-peer-reviews-test-10 - Hugging Face", "date": "", "ddg_snippet": "Samarth0710/neurips-2024-peer-reviews-test-10 · Datasets at Hugging Facetrain · 4.24k rows", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/datasets/Samarth0710/neurips-2024-peer-reviews-test-10/viewer", "content": "Samarth0710/neurips-2024-peer-reviews-test-10 · Datasets at Hugging Facetrain · 4.24k rows"}
+{"idx": 9, "title": "PDF Microsoft Word - HS_Alg2_PWATP_01_ChAssess.docx", "date": "", "ddg_snippet": "4. log 2 720 8. ln 4 3y 2 11. 3 ln 6 x + ln 4 y y 3 x − 13. log 6 6 − log 6 2 + 2 log 6 In Exercises 14-17, use the change-of-base formula to evaluate the logarithm. 14. log 3 17 15. log 9 294", "subpage_snippet": "", "source": "static.bigideasmath.com", "link": "https://static.bigideasmath.com/protected/content/pwtp/aga22/aga22_algebra2_pwtp_06.pdf", "content": "4. log 2 720 8. ln 4 3y 2 11. 3 ln 6 x + ln 4 y y 3 x − 13. log 6 6 − log 6 2 + 2 log 6 In Exercises 14-17, use the change-of-base formula to evaluate the logarithm. 14. log 3 17 15. log 9 294"}
diff --git a/data/sampled_jsons/34016_EntityErasure-_Erasing_Entity_Cleanly_via_Amodal_Entity_Segmentation_and_Completion_filetypepd.jsonl b/data/sampled_jsons/34016_EntityErasure-_Erasing_Entity_Cleanly_via_Amodal_Entity_Segmentation_and_Completion_filetypepd.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..57f812b0ffcabb368788ccda11729a896c2110f1
--- /dev/null
+++ b/data/sampled_jsons/34016_EntityErasure-_Erasing_Entity_Cleanly_via_Amodal_Entity_Segmentation_and_Completion_filetypepd.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Bulk Delete Entity Framework - Entity Frame Extensions", "date": "", "ddg_snippet": "Execute HIGH performance delete operations in Entity Framework. Extend EF6 and EF Core with Bulk Delete Easy to Customize · Easy to Use · Awesome Customer Support · Over 3000 Customers", "subpage_snippet": "", "source": "www.bing.com", "link": "https://www.bing.com/aclick?ld=e8N0jM_t4RO1qz38rb3IA9PzVUCUyQ73BP5yZxVANx8XTxgtoygpp6stJhGqouRe1zhINcoUo1_fHeho5JCAvO55diHDkhLiqfI33YxBzOl3ykNKvsK0aDqSFTYTnEMBmhDWlSjzo69cSAwNM2YlGAbwJAgBrh25USoE87x9aeL92nZ-qvr3Rl_QuSvfxwdIDyr9b0mA&u=aHR0cHMlM2ElMmYlMmZlbnRpdHlmcmFtZXdvcmstZXh0ZW5zaW9ucy5uZXQlMmZidWxrLWRlbGV0ZSUzZm1zY2xraWQlM2RiYTk0OWQ5NTZjMDMxOWZhMzM3NGY1NWE2YWRmYzNhNCUyNnV0bV9zb3VyY2UlM2RiaW5nJTI2dXRtX21lZGl1bSUzZGNwYyUyNnV0bV9jYW1wYWlnbiUzZEVGRSUyNTIwLSUyNTIwVDElMjZ1dG1fdGVybSUzZGVudGl0eSUyNTIwZnJhbWV3b3JrJTI1MjBkZWxldGUlMjZ1dG1fY29udGVudCUzZEJ1bGtEZWxldGU&rlid=ba949d956c0319fa3374f55a6adfc3a4", "content": "Execute HIGH performance delete operations in Entity Framework. Extend EF6 and EF Core with Bulk Delete Easy to Customize · Easy to Use · Awesome Customer Support · Over 3000 Customers"}
+{"idx": 1, "title": "GitHub - zyxunh/ entity _ erasure", "date": "", "ddg_snippet": "EntityErasure : Erasing Entity Cleanly via Amodal Entity Segmentation and Completion [CVPR2025].", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/zyxunh/entity_erasure", "content": "EntityErasure : Erasing Entity Cleanly via Amodal Entity Segmentation and Completion [CVPR2025]."}
+{"idx": 2, "title": "EntityErasure : Erasing Entity Cleanly via Amodal Entity ...", "date": "", "ddg_snippet": "Amodal Entity Completion . Classifer-free Guidance.The proposed entity attention injects entity in-formation with amodal entity segmentation guidance.", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/papers/Zhu_EntityErasure_Erasing_Entity_Cleanly_via_Amodal_Entity_Segmentation_and_Completion_CVPR_2025_paper.pdf", "content": "Amodal Entity Completion . Classifer-free Guidance.The proposed entity attention injects entity in-formation with amodal entity segmentation guidance."}
+{"idx": 3, "title": "EntityErasure : Erasing Entity Cleanly via Amodal Entity ...", "date": "", "ddg_snippet": "Abstract. This paper presents EntityErasure , a novel diffusion-based method that can effectively erase entity without inducing unwanted sundries. To this end, we propose to address this problem by dividing it into amodal entity segmentation and completion ...", "subpage_snippet": "", "source": "zyxunh.github.io", "link": "https://zyxunh.github.io/EntityErasure-ProjectPage/", "content": "Abstract. This paper presents EntityErasure , a novel diffusion-based method that can effectively erase entity without inducing unwanted sundries. To this end, we propose to address this problem by dividing it into amodal entity segmentation and completion ..."}
+{"idx": 4, "title": "unhzyx/ entity _ erasure at main", "date": "", "ddg_snippet": "main. entity _ erasure . Ctrl+K.Upload amodal _ segmentation _model.bin with huggingface_hub. 0056599 verified 3 minutes ago.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/unhzyx/entity_erasure/tree/main", "content": "main. entity _ erasure . Ctrl+K.Upload amodal _ segmentation _model.bin with huggingface_hub. 0056599 verified 3 minutes ago."}
+{"idx": 5, "title": "[ PDF ] [EPUB] The Way of The Iceman: How The Wim Hof Method...", "date": "", "ddg_snippet": "Get Full eBook File name \"The_Way_of_The_Iceman__How_The_Wim_Hof_Met_-_Wim_Hof. pdf .epub\" Format Complete Free. Genres: \"Biography, Health, Nonfiction, Personal Development, Self Help\".", "subpage_snippet": "", "source": "oceanofpdf.com", "link": "https://oceanofpdf.com/authors/wim-hof/pdf-epub-the-way-of-the-iceman-how-the-wim-hof-method-creates-radiant-longterm-health-using-the-science-and-secrets-of-breath-control-cold-training-and-commitment-download/", "content": "Get Full eBook File name \"The_Way_of_The_Iceman__How_The_Wim_Hof_Met_-_Wim_Hof. pdf .epub\" Format Complete Free. Genres: \"Biography, Health, Nonfiction, Personal Development, Self Help\"."}
+{"idx": 6, "title": "Soft Deleting Entities Cleanly Using Entity Framework... | CodeGuru", "date": "", "ddg_snippet": "Entity Framework (EF) does not have this feature built in. However, we can use the building blocks available in EF 6 to achieve this. Entity Framework 6 has introduced two new features: ‘Interceptors’ and ‘Custom Code Conventions’.", "subpage_snippet": "", "source": "www.codeguru.com", "link": "https://www.codeguru.com/csharp/soft-deleting-entities-cleanly-using-entity-framework-6-interceptors/", "content": "Entity Framework (EF) does not have this feature built in. However, we can use the building blocks available in EF 6 to achieve this. Entity Framework 6 has introduced two new features: ‘Interceptors’ and ‘Custom Code Conventions’."}
+{"idx": 7, "title": "DEMO | PDF .ai | The best ChatPDF app", "date": "", "ddg_snippet": "We built the ultimate ChatPDF app that allows you to chat with any PDF : ask questions, get summaries, find anything you need!", "subpage_snippet": "", "source": "pdf.ai", "link": "https://pdf.ai/demo", "content": "We built the ultimate ChatPDF app that allows you to chat with any PDF : ask questions, get summaries, find anything you need!"}
+{"idx": 8, "title": "Publications", "date": "", "ddg_snippet": "EntityErasure : Erasing Entity Cleanly via Amodal Entity Segmentation and Completion Yixing Zhu, Qing Zhang, Yitong Wang, Yongwei Nie, Wei-Shi Zheng IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025.", "subpage_snippet": "", "source": "www.zhangqing-home.net", "link": "https://www.zhangqing-home.net/full_publications.html", "content": "EntityErasure : Erasing Entity Cleanly via Amodal Entity Segmentation and Completion Yixing Zhu, Qing Zhang, Yitong Wang, Yongwei Nie, Wei-Shi Zheng IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025."}
+{"idx": 9, "title": "Youtube to MP3 Converter (Ad-free)", "date": "", "ddg_snippet": "Buy me a coffee. Donate via crypto. YouTube to MP4.", "subpage_snippet": "", "source": "ezconv.com", "link": "https://ezconv.com/", "content": "Buy me a coffee. Donate via crypto. YouTube to MP4."}
diff --git a/data/sampled_jsons/51x0dfsD8A_Algorithm_2_k-HOC_O(n^4)_OR_O(n^3)_OR_O(n^2)_OR_O(n_log_n).jsonl b/data/sampled_jsons/51x0dfsD8A_Algorithm_2_k-HOC_O(n^4)_OR_O(n^3)_OR_O(n^2)_OR_O(n_log_n).jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..6e761cc95ad14632478eb70b60b1319fbe8121ea
--- /dev/null
+++ b/data/sampled_jsons/51x0dfsD8A_Algorithm_2_k-HOC_O(n^4)_OR_O(n^3)_OR_O(n^2)_OR_O(n_log_n).jsonl
@@ -0,0 +1,8 @@
+{"idx": 0, "title": "How to Fix ‘The Signature Uses an Unsupported Algorithm. The ... [2112.05682] Self-attention Does Not Need $O (n^2)$ Memory What Is SHA-256? | Boot.dev big o - Big O Question about an algorithm with (n^2 + n) / 2 ... Big O Cheat Sheet – Time Complexity Chart - freeCodeCamp.org Fast Fourier Transformation for polynomial multiplication big o - Big-O complexity for n + n-1 + n-2 - Stack Overflow", "date": "", "ddg_snippet": "Save Up to 80% On Top SSL Brands Low Prices. 24/7 Support. SSL Experts. Dec 10, 2021 · We present a very simple algorithm for attention that requires O (1) memory with respect to sequence length and an extension to self-attention that requires O ( log n ) memory. This is in contrast with the frequently stated belief that self-attention requires O (n2) memory. While the time complexity is still O (n2), device memory rather than compute capability is often the limiting factor on modern ... If you’re not familiar with SHA-256 , try this online generator. Enter any message and click the “hash” button. See full list on blog.boot.dev SHA-256 is a standard hash function, so the real question is, “what’s a hash function”? A cryptographic hash function generates a “fingerprint” of an input string. For example, if we were to hash the entire text of JRR Tolkien’s “The Lord of The Rings” series using the SHA 256 algorithm , we would get a 256-bit output unique to that book’s text. If ... See full list on blog.boot.dev SHA-256 is useful in so manycircumstances! It’s a fast and secure hash function, here are some of the most common ways that it’s used: 1. To create website authentication schemes, using JWTs, HMACs and MACs 2 . To create digital signatures 3 . To secure blockchainslike Bitcoin and Ethereum 4 . In anti-viruses, to compare the fingerprints of files and ... See full list on blog.boot.dev SHA- 2 is known for its security (it hasn’t broken down like SHA-1) and its speed. In cases where keys are not generated, such as proof-of-work Bitcoin mining, a fast hash algorithm like SHA- 2 often has the upper hand. SHA-256 is formally defined in the National Institute of Standards and Technology’s FIPS 180- 4 . Along with standardization and forma... See full list on blog.boot.dev Let’s go through an example of the SHA-256 hashing algorithm step-by-step, by hand. If you can stay awake through this whole walkthrough, you’ll understand all of its nuts and bolts. See full list on blog.boot.dev After the compression loop, but still, within the chunkloop, we modify the hash values by adding their respective variables to them, a-h. As usual, all addition is modulo 2 ^32. See full list on blog.boot.dev Last but not least, slap them all together, a simple string concatenationwill do. Done! We’ve been through every step (sans some iterations) of SHA-256 in excruciating detail :) I’m glad you’ve made it this far! Going step-by-step through the SHA-256 algorithm isn’t exactly a walk in the park. Learning the fundamentals that underpin web security ca... See full list on blog.boot.dev If you want to see all the steps we just did above in pseudocode form, then here it is, straight from WikiPedia: See full list on blog.boot.dev SHA- 2 is an algorithm , or a generalized idea of how to hash data. SHA- 2 has several variants, all of which use the same algorithm but use different constants. SHA-256, for example, sets additional constants that define the behavior of the SHA- 2 algorithm , one of these constants is the output size, 256. The 256 and 512 in SHA-256 and SHA-512 refer t... See full list on blog.boot.dev SHA- 2 is a successor to the SHA-1 hash and remains one of the strongest hash functions in use today. SHA-256, as opposed to SHA-1, hasn’t been compromised. For this reason, there’s really no reason to use SHA-1 these days, it isn’t safe. The flexibility of output size (224, 256, 512, etc) also allows SHA- 2 to pair well with popular KDFs and ciphers... See full list on blog.boot.dev Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc... Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time. May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect.", "subpage_snippet": "", "source": "www.thesslstore.com", "link": "https://www.thesslstore.com/resources/how-to-fix-the-signature-uses-an-unsupported-algorithm-the-digital-signature-is-not-valid/", "content": "Save Up to 80% On Top SSL Brands Low Prices. 24/7 Support. SSL Experts. Dec 10, 2021 · We present a very simple algorithm for attention that requires O (1) memory with respect to sequence length and an extension to self-attention that requires O ( log n ) memory. This is in contrast with the frequently stated belief that self-attention requires O (n2) memory. While the time complexity is still O (n2), device memory rather than compute capability is often the limiting factor on modern ... If you’re not familiar with SHA-256 , try this online generator. Enter any message and click the “hash” button. See full list on blog.boot.dev SHA-256 is a standard hash function, so the real question is, “what’s a hash function”? A cryptographic hash function generates a “fingerprint” of an input string. For example, if we were to hash the entire text of JRR Tolkien’s “The Lord of The Rings” series using the SHA 256 algorithm , we would get a 256-bit output unique to that book’s text. If ... See full list on blog.boot.dev SHA-256 is useful in so manycircumstances! It’s a fast and secure hash function, here are some of the most common ways that it’s used: 1. To create website authentication schemes, using JWTs, HMACs and MACs 2 . To create digital signatures 3 . To secure blockchainslike Bitcoin and Ethereum 4 . In anti-viruses, to compare the fingerprints of files and ... See full list on blog.boot.dev SHA- 2 is known for its security (it hasn’t broken down like SHA-1) and its speed. In cases where keys are not generated, such as proof-of-work Bitcoin mining, a fast hash algorithm like SHA- 2 often has the upper hand. SHA-256 is formally defined in the National Institute of Standards and Technology’s FIPS 180- 4 . Along with standardization and forma... See full list on blog.boot.dev Let’s go through an example of the SHA-256 hashing algorithm step-by-step, by hand. If you can stay awake through this whole walkthrough, you’ll understand all of its nuts and bolts. See full list on blog.boot.dev After the compression loop, but still, within the chunkloop, we modify the hash values by adding their respective variables to them, a-h. As usual, all addition is modulo 2 ^32. See full list on blog.boot.dev Last but not least, slap them all together, a simple string concatenationwill do. Done! We’ve been through every step (sans some iterations) of SHA-256 in excruciating detail :) I’m glad you’ve made it this far! Going step-by-step through the SHA-256 algorithm isn’t exactly a walk in the park. Learning the fundamentals that underpin web security ca... See full list on blog.boot.dev If you want to see all the steps we just did above in pseudocode form, then here it is, straight from WikiPedia: See full list on blog.boot.dev SHA- 2 is an algorithm , or a generalized idea of how to hash data. SHA- 2 has several variants, all of which use the same algorithm but use different constants. SHA-256, for example, sets additional constants that define the behavior of the SHA- 2 algorithm , one of these constants is the output size, 256. The 256 and 512 in SHA-256 and SHA-512 refer t... See full list on blog.boot.dev SHA- 2 is a successor to the SHA-1 hash and remains one of the strongest hash functions in use today. SHA-256, as opposed to SHA-1, hasn’t been compromised. For this reason, there’s really no reason to use SHA-1 these days, it isn’t safe. The flexibility of output size (224, 256, 512, etc) also allows SHA- 2 to pair well with popular KDFs and ciphers... See full list on blog.boot.dev Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc... Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time. May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect."}
+{"idx": 1, "title": "What Is SHA-256? | Boot.dev big o - Big O Question about an algorithm with (n^2 + n) / 2 ... Big O Cheat Sheet – Time Complexity Chart - freeCodeCamp.org Fast Fourier Transformation for polynomial multiplication big o - Big-O complexity for n + n-1 + n-2 - Stack Overflow", "date": "", "ddg_snippet": "If you’re not familiar with SHA-256 , try this online generator. Enter any message and click the “hash” button. See full list on blog.boot.dev SHA-256 is a standard hash function, so the real question is, “what’s a hash function”? A cryptographic hash function generates a “fingerprint” of an input string. For example, if we were to hash the entire text of JRR Tolkien’s “The Lord of The Rings” series using the SHA 256 algorithm , we would get a 256-bit output unique to that book’s text. If ... See full list on blog.boot.dev SHA-256 is useful in so manycircumstances! It’s a fast and secure hash function, here are some of the most common ways that it’s used: 1. To create website authentication schemes, using JWTs, HMACs and MACs 2 . To create digital signatures 3 . To secure blockchainslike Bitcoin and Ethereum 4 . In anti-viruses, to compare the fingerprints of files and ... See full list on blog.boot.dev SHA- 2 is known for its security (it hasn’t broken down like SHA-1) and its speed. In cases where keys are not generated, such as proof-of-work Bitcoin mining, a fast hash algorithm like SHA- 2 often has the upper hand. SHA-256 is formally defined in the National Institute of Standards and Technology’s FIPS 180- 4 . Along with standardization and forma... See full list on blog.boot.dev Let’s go through an example of the SHA-256 hashing algorithm step-by-step, by hand. If you can stay awake through this whole walkthrough, you’ll understand all of its nuts and bolts. See full list on blog.boot.dev After the compression loop, but still, within the chunkloop, we modify the hash values by adding their respective variables to them, a-h. As usual, all addition is modulo 2 ^32. See full list on blog.boot.dev Last but not least, slap them all together, a simple string concatenationwill do. Done! We’ve been through every step (sans some iterations) of SHA-256 in excruciating detail :) I’m glad you’ve made it this far! Going step-by-step through the SHA-256 algorithm isn’t exactly a walk in the park. Learning the fundamentals that underpin web security ca... See full list on blog.boot.dev If you want to see all the steps we just did above in pseudocode form, then here it is, straight from WikiPedia: See full list on blog.boot.dev SHA- 2 is an algorithm , or a generalized idea of how to hash data. SHA- 2 has several variants, all of which use the same algorithm but use different constants. SHA-256, for example, sets additional constants that define the behavior of the SHA- 2 algorithm , one of these constants is the output size, 256. The 256 and 512 in SHA-256 and SHA-512 refer t... See full list on blog.boot.dev SHA- 2 is a successor to the SHA-1 hash and remains one of the strongest hash functions in use today. SHA-256, as opposed to SHA-1, hasn’t been compromised. For this reason, there’s really no reason to use SHA-1 these days, it isn’t safe. The flexibility of output size (224, 256, 512, etc) also allows SHA- 2 to pair well with popular KDFs and ciphers... See full list on blog.boot.dev Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc... Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time. May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect.", "subpage_snippet": "", "source": "blog.boot.dev", "link": "https://blog.boot.dev/cryptography/how-sha-2-works-step-by-step-sha-256/", "content": "If you’re not familiar with SHA-256 , try this online generator. Enter any message and click the “hash” button. See full list on blog.boot.dev SHA-256 is a standard hash function, so the real question is, “what’s a hash function”? A cryptographic hash function generates a “fingerprint” of an input string. For example, if we were to hash the entire text of JRR Tolkien’s “The Lord of The Rings” series using the SHA 256 algorithm , we would get a 256-bit output unique to that book’s text. If ... See full list on blog.boot.dev SHA-256 is useful in so manycircumstances! It’s a fast and secure hash function, here are some of the most common ways that it’s used: 1. To create website authentication schemes, using JWTs, HMACs and MACs 2 . To create digital signatures 3 . To secure blockchainslike Bitcoin and Ethereum 4 . In anti-viruses, to compare the fingerprints of files and ... See full list on blog.boot.dev SHA- 2 is known for its security (it hasn’t broken down like SHA-1) and its speed. In cases where keys are not generated, such as proof-of-work Bitcoin mining, a fast hash algorithm like SHA- 2 often has the upper hand. SHA-256 is formally defined in the National Institute of Standards and Technology’s FIPS 180- 4 . Along with standardization and forma... See full list on blog.boot.dev Let’s go through an example of the SHA-256 hashing algorithm step-by-step, by hand. If you can stay awake through this whole walkthrough, you’ll understand all of its nuts and bolts. See full list on blog.boot.dev After the compression loop, but still, within the chunkloop, we modify the hash values by adding their respective variables to them, a-h. As usual, all addition is modulo 2 ^32. See full list on blog.boot.dev Last but not least, slap them all together, a simple string concatenationwill do. Done! We’ve been through every step (sans some iterations) of SHA-256 in excruciating detail :) I’m glad you’ve made it this far! Going step-by-step through the SHA-256 algorithm isn’t exactly a walk in the park. Learning the fundamentals that underpin web security ca... See full list on blog.boot.dev If you want to see all the steps we just did above in pseudocode form, then here it is, straight from WikiPedia: See full list on blog.boot.dev SHA- 2 is an algorithm , or a generalized idea of how to hash data. SHA- 2 has several variants, all of which use the same algorithm but use different constants. SHA-256, for example, sets additional constants that define the behavior of the SHA- 2 algorithm , one of these constants is the output size, 256. The 256 and 512 in SHA-256 and SHA-512 refer t... See full list on blog.boot.dev SHA- 2 is a successor to the SHA-1 hash and remains one of the strongest hash functions in use today. SHA-256, as opposed to SHA-1, hasn’t been compromised. For this reason, there’s really no reason to use SHA-1 these days, it isn’t safe. The flexibility of output size (224, 256, 512, etc) also allows SHA- 2 to pair well with popular KDFs and ciphers... See full list on blog.boot.dev Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc... Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time. May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect."}
+{"idx": 2, "title": "big o - Big O Question about an algorithm with (n^2 + n) / 2 ...", "date": "", "ddg_snippet": "Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc...", "subpage_snippet": "", "source": "softwareengineering.stackexchange.com", "link": "https://softwareengineering.stackexchange.com/questions/279609/big-o-question-about-an-algorithm-with-n2-n-2-growth-rate", "content": "Apr 20, 2015 · So we call the first two algorithms O(n^2 ), and we call the second O(n). It nicely divides the world up into classes of algorithms. This is what big-O is all about. It's like dividing vehicles up into cars and trucks and busses etc..."}
+{"idx": 3, "title": "[2112.05682] Self-attention Does Not Need $O (n^2)$ Memory", "date": "", "ddg_snippet": "Dec 10, 2021 · We present a very simple algorithm for attention that requires O (1) memory with respect to sequence length and an extension to self-attention that requires O ( log n ) memory. This is in contrast with the frequently stated belief that self-attention requires O (n2) memory. While the time complexity is still O (n2), device memory rather than compute capability is often the limiting factor on modern ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2112.05682", "content": "Dec 10, 2021 · We present a very simple algorithm for attention that requires O (1) memory with respect to sequence length and an extension to self-attention that requires O ( log n ) memory. This is in contrast with the frequently stated belief that self-attention requires O (n2) memory. While the time complexity is still O (n2), device memory rather than compute capability is often the limiting factor on modern ..."}
+{"idx": 4, "title": "Big O Cheat Sheet – Time Complexity Chart - freeCodeCamp.org", "date": "", "ddg_snippet": "Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart", "subpage_snippet": "", "source": "www.freecodecamp.org", "link": "https://www.freecodecamp.org/news/big-o-cheat-sheet-time-complexity-chart/", "content": "Oct 5, 2022 · In Big O, there are six major types of complexities (time and space): Constant: O (1) Linear time: O (n) Logarithmic time: O (n log n) Quadratic time: O (n^2 ) Exponential time: O (2^n) Factorial time: O (n!) Before we look at examples for each time complexity, let's understand the Big O time complexity chart. Big O Complexity Chart"}
+{"idx": 5, "title": "Fast Fourier Transformation for polynomial multiplication", "date": "", "ddg_snippet": "Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time.", "subpage_snippet": "", "source": "www.geeksforgeeks.org", "link": "https://www.geeksforgeeks.org/dsa/fast-fourier-transformation-poynomial-multiplication/", "content": "Mar 15, 2023 · This idea still solves the problem in O (n 2 n2 ) time complexity. We can use any points we want as evaluation points, but by choosing the evaluation points carefully, we can convert between representations in only O (n log n) time."}
+{"idx": 6, "title": "Hierarchical Overlapping Clustering on Graphs: Cost ...", "date": "", "ddg_snippet": "by Y Pan — They pro- posed the algorithms that achieve O ( log n )-approximation factors for the four kinds of versions, where n is the number of vertices. The techniques ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=51x0dfsD8A", "content": "by Y Pan — They pro- posed the algorithms that achieve O ( log n )-approximation factors for the four kinds of versions, where n is the number of vertices. The techniques ..."}
+{"idx": 7, "title": "big o - Big-O complexity for n + n-1 + n-2 - Stack Overflow", "date": "", "ddg_snippet": "May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect.", "subpage_snippet": "", "source": "stackoverflow.com", "link": "https://stackoverflow.com/questions/44252596/big-o-complexity-for-n-n-1-n-2-n-3-1", "content": "May 30, 2017 · Strictly speaking, shouldn’t you drop the constant and leave only the dominant term? Hence (n^2-n)/2 would end up having a complexity of n^2 or O(n^2 ). In other words saying that the complexity will be (n^2-n)/2 is incorrect."}
diff --git a/data/sampled_jsons/ACIDDnTbSJ_Feint_Behaviors_and_Strategies_Section_4.2.2_scheduler_weights_lambda_short_lambda_long.jsonl b/data/sampled_jsons/ACIDDnTbSJ_Feint_Behaviors_and_Strategies_Section_4.2.2_scheduler_weights_lambda_short_lambda_long.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..135e0e26683ff7fd21a31ad82e56b059e4fb12a5
--- /dev/null
+++ b/data/sampled_jsons/ACIDDnTbSJ_Feint_Behaviors_and_Strategies_Section_4.2.2_scheduler_weights_lambda_short_lambda_long.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Feint Behaviors and Strategies : Formalization, Implementation and...", "date": "", "ddg_snippet": "Feint behaviors refer to a set of deceptive behaviors in a nuanced manner, which enable players to obtain temporal and spatial advantages over opponents in competitive games.", "subpage_snippet": "", "source": "papers.cool", "link": "https://papers.cool/venue/ACIDDnTbSJ@OpenReview", "content": "Feint behaviors refer to a set of deceptive behaviors in a nuanced manner, which enable players to obtain temporal and spatial advantages over opponents in competitive games."}
+{"idx": 1, "title": "xAI launches Grok-4-Fast: Unified Reasoning and... - MarkTechPost", "date": "", "ddg_snippet": "xAI introduced Grok-4-Fast, a cost-optimized successor to Grok-4 that merges “reasoning” and “non-reasoning” behaviors into a single set of weights controllable via system prompts.", "subpage_snippet": "", "source": "www.marktechpost.com", "link": "https://www.marktechpost.com/2025/09/20/xai-launches-grok-4-fast-unified-reasoning-and-non-reasoning-model-with-2m-token-context-and-trained-end-to-end-with-tool-use-reinforcement-learning-rl/", "content": "xAI introduced Grok-4-Fast, a cost-optimized successor to Grok-4 that merges “reasoning” and “non-reasoning” behaviors into a single set of weights controllable via system prompts."}
+{"idx": 2, "title": "Feint Behaviors and Strategies : Formalization, Implementation and...", "date": "", "ddg_snippet": "Full paper. ACIDDnTbSJ . This figure shows an example of unsuccessful Feint behavior due to its long duration. The timeline is divided into three key time points: tB1 (end of NPC B’s first defense behavior ), tA2 (estimated start of NPC A’s second beha...", "subpage_snippet": "", "source": "deep-diver.github.io", "link": "https://deep-diver.github.io/neurips2024/posters/aciddntbsj/", "content": "Full paper. ACIDDnTbSJ . This figure shows an example of unsuccessful Feint behavior due to its long duration. The timeline is divided into three key time points: tB1 (end of NPC B’s first defense behavior ), tA2 (estimated start of NPC A’s second beha..."}
+{"idx": 3, "title": "Lambda expressions (since C++11) - cppreference.com", "date": "", "ddg_snippet": "Metaprogramming library. General utilities library. Containers library. Iterators library. Ranges library. Algorithms library. Strings library. Text processing library. Numerics library. Date and time library. Input/output library. Filesystem library...", "subpage_snippet": "", "source": "en.cppreference.com", "link": "https://en.cppreference.com/w/cpp/language/lambda.html", "content": "Metaprogramming library. General utilities library. Containers library. Iterators library. Ranges library. Algorithms library. Strings library. Text processing library. Numerics library. Date and time library. Input/output library. Filesystem library..."}
+{"idx": 4, "title": "Google Scholar", "date": "", "ddg_snippet": "Google Scholar provides a simple way to broadly search for scholarly literature. Search across a wide variety of disciplines and sources: articles, theses, books, abstracts and court opinions.", "subpage_snippet": "", "source": "scholar.google.com", "link": "https://scholar.google.com/schhp?hl=en&as_sdt=0,5", "content": "Google Scholar provides a simple way to broadly search for scholarly literature. Search across a wide variety of disciplines and sources: articles, theses, books, abstracts and court opinions."}
+{"idx": 5, "title": "Lambda Anti-Pattern: Functions Calling Functions | Orchestra", "date": "", "ddg_snippet": "Learn about Anti-patterns in Lambda -based applications in AWS lambda and how Orchestra can help you integrate AWS Lambda with data pipelines: Lambda functions calling Lambda functions.", "subpage_snippet": "", "source": "www.getorchestra.io", "link": "https://www.getorchestra.io/guides/lambda-anti-pattern-functions-calling-functions", "content": "Learn about Anti-patterns in Lambda -based applications in AWS lambda and how Orchestra can help you integrate AWS Lambda with data pipelines: Lambda functions calling Lambda functions."}
+{"idx": 6, "title": "ГДЗ по английскому языку 6 класс (spotlight) Ваулина - рабочая...", "date": "", "ddg_snippet": "24 4c My favourite day 25 4d Writing (a short text about a person) 26 Grammar Practice 27. Module 5. 29.Pairwork activities 65. Student A 65 Student B 71. Revision Section 77.", "subpage_snippet": "", "source": "gdz.top", "link": "https://gdz.top/6-klass/english/vaulina-spotlight-rabochaja-tetrad", "content": "24 4c My favourite day 25 4d Writing (a short text about a person) 26 Grammar Practice 27. Module 5. 29.Pairwork activities 65. Student A 65 Student B 71. Revision Section 77."}
+{"idx": 7, "title": "NHL Schedule - 2025 Season - ESPN", "date": "", "ddg_snippet": "The complete 2025 NHL season schedule on ESPN. Includes game times, TV listings and ticket information for all NHL games.", "subpage_snippet": "", "source": "www.espn.com", "link": "https://www.espn.com/nhl/schedule", "content": "The complete 2025 NHL season schedule on ESPN. Includes game times, TV listings and ticket information for all NHL games."}
+{"idx": 8, "title": "Lambda в Python: синтаксис, аргументы и много примеров...", "date": "", "ddg_snippet": "Все, что нужно знать о lambda -функциях в Python. В этой статье вы узнаете о том, что такое лямбда-функции в Python.", "subpage_snippet": "", "source": "PythonRu.com", "link": "https://PythonRu.com/osnovy/vse-chto-nuzhno-znat-o-lambda-funkcijah-v-python", "content": "Все, что нужно знать о lambda -функциях в Python. В этой статье вы узнаете о том, что такое лямбда-функции в Python."}
+{"idx": 9, "title": "[LYRICS] Feint - Snake Eyes (ft. CoMa) - YouTube", "date": "", "ddg_snippet": "This is a lyrics video. I am not affiliated with any artists or labels.- - - ᴅᴏᴡɴʟᴏᴀᴅs - - -Download the song ↓Beatport: http://btprt.dj/RDu7HNDNB, so cl...", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=tx6v2lWRrZ4", "content": "This is a lyrics video. I am not affiliated with any artists or labels.- - - ᴅᴏᴡɴʟᴏᴀᴅs - - -Download the song ↓Beatport: http://btprt.dj/RDu7HNDNB, so cl..."}
diff --git a/data/sampled_jsons/A_Likelihood_Based_Approach_to_Distribution_Regression_Using_Conditional_Deep_Generative_Models.jsonl b/data/sampled_jsons/A_Likelihood_Based_Approach_to_Distribution_Regression_Using_Conditional_Deep_Generative_Models.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9f184a67f2e4de6905d39c8c9ea25eca11dc9ca4
--- /dev/null
+++ b/data/sampled_jsons/A_Likelihood_Based_Approach_to_Distribution_Regression_Using_Conditional_Deep_Generative_Models.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Flow-based generative model - Wikipedia", "date": "", "ddg_snippet": "A flow- based generative model is a generative model used in machine learning that explicitly models a probability distribution by leveraging ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Flow-based_generative_model", "content": "A flow- based generative model is a generative model used in machine learning that explicitly models a probability distribution by leveraging ..."}
+{"idx": 1, "title": "Graphical model - Wikipedia", "date": "", "ddg_snippet": "Generally, probabilistic graphical models use a graph- based representation as the foundation for encoding a distribution over a multi-dimensional ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Graphical_model", "content": "Generally, probabilistic graphical models use a graph- based representation as the foundation for encoding a distribution over a multi-dimensional ..."}
+{"idx": 2, "title": "Generative model - Wikipedia", "date": "", "ddg_snippet": "... generative model \" is also used to describe models that generate instances of output variables in a way that has no clear relationship to probability ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Generative_model", "content": "... generative model \" is also used to describe models that generate instances of output variables in a way that has no clear relationship to probability ..."}
+{"idx": 3, "title": "A LIKELIHOOD BASED APPROACH TO DISTRIBUTION REGRESSION USING ...", "date": "", "ddg_snippet": "solely on the intrinsic dimension and smoothness of the true conditional distribution . These findings provide an explanation of why conditional deep gen - erative models can circumvent the curse of dimensionality from the perspective of statistical foundations and demons", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=V6hhhXoTSq", "content": "solely on the intrinsic dimension and smoothness of the true conditional distribution . These findings provide an explanation of why conditional deep gen - erative models can circumvent the curse of dimensionality from the perspective of statistical foundations and demons"}
+{"idx": 4, "title": "A Likelihood Based Approach to Distribution Regression Using ...", "date": "", "ddg_snippet": "Oct 2, 2024 · In this work, we explore the theoretical properties of conditional deep generative models under the statistical framework of distribution regression where the response variable lies in...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/384630603_A_Likelihood_Based_Approach_to_Distribution_Regression_Using_Conditional_Deep_Generative_Models", "content": "Oct 2, 2024 · In this work, we explore the theoretical properties of conditional deep generative models under the statistical framework of distribution regression where the response variable lies in..."}
+{"idx": 5, "title": "ICML Poster A Likelihood Based Approach to Distribution ...", "date": "", "ddg_snippet": "In this work, we explore the theoretical properties of conditional deep generative models under the statistical framework of distribution regression where the response variable lies in a high-dimensional ambient space but concentrates around a potentially lower-dimensional manifold.", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46645", "content": "In this work, we explore the theoretical properties of conditional deep generative models under the statistical framework of distribution regression where the response variable lies in a high-dimensional ambient space but concentrates around a potentially lower-dimensional manifold."}
+{"idx": 6, "title": "A Likelihood Based Approach to Distribution Regression Using ...", "date": "", "ddg_snippet": "Our approach simultaneously estimates a regression function and a conditional generator using a generative learning framework, where a conditional generator is a function that can generate samples from a conditional distribution .", "subpage_snippet": "", "source": "bohrium.dp.tech", "link": "https://bohrium.dp.tech/paper/arxiv/2410.02025", "content": "Our approach simultaneously estimates a regression function and a conditional generator using a generative learning framework, where a conditional generator is a function that can generate samples from a conditional distribution ."}
+{"idx": 7, "title": "A Likelihood Based Approach to Distribution Regression Using ...", "date": "", "ddg_snippet": "To the best of our knowledge, our study is the first attempt to explore the likelihood - based approach for distributional regression using a conditional deep generative model , considering full-dimensional noise and the potential presence of singular underlying support.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.02025v1", "content": "To the best of our knowledge, our study is the first attempt to explore the likelihood - based approach for distributional regression using a conditional deep generative model , considering full-dimensional noise and the potential presence of singular underlying support."}
+{"idx": 8, "title": "A Likelihood Approach to Nonparametric Estimation of a ...", "date": "", "ddg_snippet": "We investigate statistical properties of a likelihood approach to nonparametric estimation of a singular distribution using deep generative models . More speci cally, a deep generative model is used to model high-dimensional data that are assumed to concentrate around some low-dimensional structure.", "subpage_snippet": "", "source": "jmlr.org", "link": "https://jmlr.org/papers/volume24/21-1099/21-1099.pdf", "content": "We investigate statistical properties of a likelihood approach to nonparametric estimation of a singular distribution using deep generative models . More speci cally, a deep generative model is used to model high-dimensional data that are assumed to concentrate around some low-dimensional structure."}
+{"idx": 9, "title": "Newest 'normal-distribution' Questions - Cross Validated", "date": "", "ddg_snippet": "... distribution comes when we try ... I'm trying to calculate the 50% confidence interval of the true mean based on a sample of a gaussian distribution .", "subpage_snippet": "", "source": "stats.stackexchange.com", "link": "https://stats.stackexchange.com/questions/tagged/normal-distribution", "content": "... distribution comes when we try ... I'm trying to calculate the 50% confidence interval of the true mean based on a sample of a gaussian distribution ."}
diff --git a/data/sampled_jsons/A_Mathematical_Framework_for_Transformer_Circuits_abstract_Elhage_year_2021.jsonl b/data/sampled_jsons/A_Mathematical_Framework_for_Transformer_Circuits_abstract_Elhage_year_2021.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..cd5ecc21c3c88835554edb2821febe9622035ac0
--- /dev/null
+++ b/data/sampled_jsons/A_Mathematical_Framework_for_Transformer_Circuits_abstract_Elhage_year_2021.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "A Mathematical Framework for Transformer Circuits", "date": "", "ddg_snippet": "Nelson Elhage ∗†, Neel Nanda ∗, Catherine Olsson ∗, Tom Henighan †, Nicholas Joseph †, Ben Mann †, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark, Jared ...", "subpage_snippet": "", "source": "transformer-circuits.pub", "link": "https://transformer-circuits.pub/2021/framework/index.html", "content": "Nelson Elhage ∗†, Neel Nanda ∗, Catherine Olsson ∗, Tom Henighan †, Nicholas Joseph †, Ben Mann †, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark, Jared ..."}
+{"idx": 1, "title": "framework-transformer-circuits.pdf", "date": "", "ddg_snippet": "A foundational article for understanding inner workings of transformer architecture. - Antrophic, A Mathematical Framework for Transformer Circuits . Page 5. 1 ...", "subpage_snippet": "", "source": "jungwooyang.com", "link": "https://jungwooyang.com/assets/files/framework-transformer-circuits.pdf", "content": "A foundational article for understanding inner workings of transformer architecture. - Antrophic, A Mathematical Framework for Transformer Circuits . Page 5. 1 ..."}
+{"idx": 2, "title": "A Mathematical Framework For Transformer Circuits", "date": "", "ddg_snippet": "The document discusses a mathematical framework for understanding transformer circuits, focusing on reverse engineering simple transformer models to uncover their internal operations. It highlights the importance of attention heads, particularly 'induction heads', in facilitating in-context learning and outlines key findings about the algorithms used by one and two-layer attention-only ...", "subpage_snippet": "", "source": "www.scribd.com", "link": "https://www.scribd.com/document/866284321/A-Mathematical-Framework-for-Transformer-Circuits", "content": "The document discusses a mathematical framework for understanding transformer circuits, focusing on reverse engineering simple transformer models to uncover their internal operations. It highlights the importance of attention heads, particularly 'induction heads', in facilitating in-context learning and outlines key findings about the algorithms used by one and two-layer attention-only ..."}
+{"idx": 3, "title": "A Mathematical Framework for Transformer Circuits - Oxen", "date": "", "ddg_snippet": "Mathematical Framework for Transformer Circuits The paper can be found here:", "subpage_snippet": "", "source": "www.oxen.ai", "link": "https://www.oxen.ai/blog/arxiv-dives-a-mathematical-framework-for-transformer-circuits", "content": "Mathematical Framework for Transformer Circuits The paper can be found here:"}
+{"idx": 4, "title": "A Mathematical Framework for Transformer Circuits - Tom Bewley", "date": "", "ddg_snippet": "A Mathematical Framework for Transformer Circuits 2 minute read 2021 #Content/Paper by Nelson Elhage , Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark ...", "subpage_snippet": "", "source": "tombewley.com", "link": "https://tombewley.com/notes/A+Mathematical+Framework+for+Transformer+Circuits", "content": "A Mathematical Framework for Transformer Circuits 2 minute read 2021 #Content/Paper by Nelson Elhage , Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark ..."}
+{"idx": 5, "title": "A Mathematical Framework for Transformer Circuits", "date": "", "ddg_snippet": "We don't attempt to apply to our insights to larger models in this first paper, but in a forthcoming paper, we will show that both our mathematical framework for understanding transformers, and the concept of induction heads, continues to be at least partially relevant for much larger and more realistic models - though we remain a very long way from being able to fully reverse engineer ...", "subpage_snippet": "", "source": "dgcnz.github.io", "link": "https://dgcnz.github.io/second-brain/100+Reference+notes/101+Literature/A+Mathematical+Framework+for+Transformer+Circuits/", "content": "We don't attempt to apply to our insights to larger models in this first paper, but in a forthcoming paper, we will show that both our mathematical framework for understanding transformers, and the concept of induction heads, continues to be at least partially relevant for much larger and more realistic models - though we remain a very long way from being able to fully reverse engineer ..."}
+{"idx": 6, "title": "Transformer Circuits: A Mathematical Framework - studylib.net", "date": "", "ddg_snippet": "Explore a mathematical framework for transformer circuits , focusing on attention mechanisms and in-context learning. AI, Machine Learning research.", "subpage_snippet": "", "source": "studylib.net", "link": "https://studylib.net/doc/27496897/transformer-circui...--1-", "content": "Explore a mathematical framework for transformer circuits , focusing on attention mechanisms and in-context learning. AI, Machine Learning research."}
+{"idx": 7, "title": "A Mathematical Framework for Transformer Circuits - Amanda Askell", "date": "", "ddg_snippet": "Recommended citation: Nelson Elhage , Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark, Jared Kaplan ...", "subpage_snippet": "", "source": "askell.io", "link": "https://askell.io/publication/transformer-circuits", "content": "Recommended citation: Nelson Elhage , Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Jack Clark, Jared Kaplan ..."}
+{"idx": 8, "title": "A Walkthrough of A Mathematical Framework for Transformer Circuits", "date": "", "ddg_snippet": "A Mathematical Framework for Transformer Circuits is, in my opinion, the coolest paper I've ever had the privilege of working on. But it's also very long and dense and at times confusing, and this makes me sad!", "subpage_snippet": "", "source": "www.neelnanda.io", "link": "https://www.neelnanda.io/mechanistic-interpretability/a-walkthrough-of-a-mathematical-framework-for-transformer-circuits", "content": "A Mathematical Framework for Transformer Circuits is, in my opinion, the coolest paper I've ever had the privilege of working on. But it's also very long and dense and at times confusing, and this makes me sad!"}
+{"idx": 9, "title": "A Mathematical Framework for Transformer Circuits", "date": "", "ddg_snippet": "A Mathematical Framework for Transformer Circuits ; Anthropic Economic Index report: Uneven geographic and enterprise AI adoption. Sep 15, 2025.", "subpage_snippet": "", "source": "www.anthropic.com", "link": "https://www.anthropic.com/research/a-mathematical-framework-for-transformer-circuits", "content": "A Mathematical Framework for Transformer Circuits ; Anthropic Economic Index report: Uneven geographic and enterprise AI adoption. Sep 15, 2025."}
diff --git a/data/sampled_jsons/A_cost_function_for_similarity-based_hierarchical_clustering_Dasgupta_2016_abstract_year_2016.jsonl b/data/sampled_jsons/A_cost_function_for_similarity-based_hierarchical_clustering_Dasgupta_2016_abstract_year_2016.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b5f4cfbf944b55565ccbc9efdd31aa92f1119a73
--- /dev/null
+++ b/data/sampled_jsons/A_cost_function_for_similarity-based_hierarchical_clustering_Dasgupta_2016_abstract_year_2016.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "A cost function for similarity-based hierarchical clustering", "date": "", "ddg_snippet": "by S Dasgupta · 2015 · Cited by 275 — We introduce a simple cost function on hierarchies over a set of points, given pairwise similarities between those points.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/1510.05043", "content": "by S Dasgupta · 2015 · Cited by 275 — We introduce a simple cost function on hierarchies over a set of points, given pairwise similarities between those points."}
+{"idx": 1, "title": "A cost function for similarity-based hierarchical clustering", "date": "", "ddg_snippet": "We introduce a simple cost function on hierarchies over a set of points, given pairwise similarities between those points.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/10.1145/2897518.2897527", "content": "We introduce a simple cost function on hierarchies over a set of points, given pairwise similarities between those points."}
+{"idx": 2, "title": "A cost function for similarity-based hierarchical clustering", "date": "", "ddg_snippet": "by S Dasgupta · 2016 · Cited by 275 — In this work we introduce a simple cost function for hierarchical clustering: a function that, given pairwise similarities between data points, assigns a score ... 18 pages", "subpage_snippet": "", "source": "cseweb.ucsd.edu", "link": "https://cseweb.ucsd.edu/~dasgupta/papers/hier-cost.pdf", "content": "by S Dasgupta · 2016 · Cited by 275 — In this work we introduce a simple cost function for hierarchical clustering: a function that, given pairwise similarities between data points, assigns a score ... 18 pages"}
+{"idx": 3, "title": "Hierarchical Overlapping Clustering on Graphs: Cost ...", "date": "", "ddg_snippet": "by Y Pan — Dasgupta, S. A cost function for similarity-based hierarchi- cal clustering . In Proceedings of the Forty-eighth Annual. ACM Symposium on Theory of Computing ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=51x0dfsD8A", "content": "by Y Pan — Dasgupta, S. A cost function for similarity-based hierarchi- cal clustering . In Proceedings of the Forty-eighth Annual. ACM Symposium on Theory of Computing ..."}
+{"idx": 4, "title": "A Revenue Function for Comparison-Based Hierarchical ...", "date": "", "ddg_snippet": "by A Mandal · 2022 · Cited by 3 — We show that this function is closely related to Dasgupta's cost for hierarchical clustering that uses pairwise similarities. On the theoretical ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2211.16459", "content": "by A Mandal · 2022 · Cited by 3 — We show that this function is closely related to Dasgupta's cost for hierarchical clustering that uses pairwise similarities. On the theoretical ..."}
+{"idx": 5, "title": "Foundations of Comparison-Based Hierarchical Clustering", "date": "", "ddg_snippet": "by D Ghoshdastidar · Cited by 41 — Dasgupta, S. (2016). A cost function for similarity-based hierarchical clustering . In Symposium on. Theory of Computing, pages 118–127. Emamjomeh-Zadeh, E ...", "subpage_snippet": "", "source": "papers.neurips.cc", "link": "http://papers.neurips.cc/paper/8964-foundations-of-comparison-based-hierarchical-clustering.pdf", "content": "by D Ghoshdastidar · Cited by 41 — Dasgupta, S. (2016). A cost function for similarity-based hierarchical clustering . In Symposium on. Theory of Computing, pages 118–127. Emamjomeh-Zadeh, E ..."}
+{"idx": 6, "title": "An Information-theoretic Perspective of Hierarchical ...", "date": "", "ddg_snippet": "by Y Pan · Cited by 10 — 2007. 395. [10] Sanjoy Dasgupta. A cost function for similarity-based hierarchical clustering . In Daniel Wichs. 396 and Yishay Mansour, editors, Proceedings ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=LPB2BFZvncQ", "content": "by Y Pan · Cited by 10 — 2007. 395. [10] Sanjoy Dasgupta. A cost function for similarity-based hierarchical clustering . In Daniel Wichs. 396 and Yishay Mansour, editors, Proceedings ..."}
+{"idx": 7, "title": "Hierarchical Clustering: New Bounds and Objective", "date": "", "ddg_snippet": "by M Rahgoshay · 2021 · Cited by 4 — He defined a cost function for a tree $T$ to be $ Cost(T) = \\sum_{i,j \\in [n]} \\big(w_{i,j} \\times |T_{i,j}| \\big )$ where $T_{i,j}$ is the subtree rooted at the ...", "subpage_snippet": "", "source": "ui.adsabs.harvard.edu", "link": "https://ui.adsabs.harvard.edu/abs/2021arXiv211106863R/abstract", "content": "by M Rahgoshay · 2021 · Cited by 4 — He defined a cost function for a tree $T$ to be $ Cost(T) = \\sum_{i,j \\in [n]} \\big(w_{i,j} \\times |T_{i,j}| \\big )$ where $T_{i,j}$ is the subtree rooted at the ..."}
+{"idx": 8, "title": "Hierarchical Clustering: Objective Functions and Algorithms", "date": "", "ddg_snippet": "[21] Sanjoy Dasgupta. A cost function for similarity-based hierarchical clustering . In Proceedings of the 48th Annual ACM SIGACT Symposium on Theory of ...", "subpage_snippet": "", "source": "dspace.mit.edu", "link": "https://dspace.mit.edu/bitstream/handle/1721.1/121430/main_openaccess.pdf;jsessionid=91FF840BBE185F0BEDA735618FE63D41?sequence=1", "content": "[21] Sanjoy Dasgupta. A cost function for similarity-based hierarchical clustering . In Proceedings of the 48th Annual ACM SIGACT Symposium on Theory of ..."}
+{"idx": 9, "title": "UCSB CS Theory Colloquium Series", "date": "", "ddg_snippet": "A cost function for similarity-based hierarchical clustering . STOC 2016. 2. Explainable k-means clustering. Common clustering algorithms return clusters ...", "subpage_snippet": "", "source": "sites.cs.ucsb.edu", "link": "https://sites.cs.ucsb.edu/~vigoda/theory/F21/Sanjoy.html", "content": "A cost function for similarity-based hierarchical clustering . STOC 2016. 2. Explainable k-means clustering. Common clustering algorithms return clusters ..."}
diff --git "a/data/sampled_jsons/Algorithm_2_Tanh_Demon_adaptive_temperature_\317\204_calculation_standard_deviation_R_k_reward.jsonl" "b/data/sampled_jsons/Algorithm_2_Tanh_Demon_adaptive_temperature_\317\204_calculation_standard_deviation_R_k_reward.jsonl"
new file mode 100644
index 0000000000000000000000000000000000000000..bced0ed4aa0fd843c70009d0fe9b0ab0a9e8c56e
--- /dev/null
+++ "b/data/sampled_jsons/Algorithm_2_Tanh_Demon_adaptive_temperature_\317\204_calculation_standard_deviation_R_k_reward.jsonl"
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Standard deviation - Wikipedia", "date": "", "ddg_snippet": "A useful property of the standard deviation is that, unlike the variance, it is expressed in the same unit as the data. Standard deviation can also be used to calculate standard error for a finite sample, and to determine statistical significance.", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Standard_deviation", "content": "A useful property of the standard deviation is that, unlike the variance, it is expressed in the same unit as the data. Standard deviation can also be used to calculate standard error for a finite sample, and to determine statistical significance."}
+{"idx": 1, "title": "Training-free Diffusion Model Alignment with Sampling Demons", "date": "", "ddg_snippet": "Algorithm 2 Tanh Demon with Adaptive Temperature . 1: Input: A list of ODE reward estimate.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.05760v1", "content": "Algorithm 2 Tanh Demon with Adaptive Temperature . 1: Input: A list of ODE reward estimate."}
+{"idx": 2, "title": "How To Calculate The Standard Deviation - YouTube", "date": "", "ddg_snippet": "This Statistics video tutorial explains how to calculate the standard deviation using 2 examples.", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=IaTFpp-uzp0", "content": "This Statistics video tutorial explains how to calculate the standard deviation using 2 examples."}
+{"idx": 3, "title": "Standard Deviation - Formula | How to Calculate Standard Deviation ?", "date": "", "ddg_snippet": "The calculations for standard deviation differ for different data. Distribution measures the deviation of data from its mean or average position. There are three methods to find the standard deviation .", "subpage_snippet": "", "source": "www.cuemath.com", "link": "https://www.cuemath.com/data/standard-deviation/", "content": "The calculations for standard deviation differ for different data. Distribution measures the deviation of data from its mean or average position. There are three methods to find the standard deviation ."}
+{"idx": 4, "title": "Standard Deviation Formulas", "date": "", "ddg_snippet": "Standard Deviation Formulas. Deviation means how far from the normal.To calculate the standard deviation of those numbers: 1. Work out the Mean (the simple average of the numbers). 2 . Then for each number: subtract the Mean and square the result.", "subpage_snippet": "", "source": "www.mathsisfun.com", "link": "https://www.mathsisfun.com/data/standard-deviation-formulas.html", "content": "Standard Deviation Formulas. Deviation means how far from the normal.To calculate the standard deviation of those numbers: 1. Work out the Mean (the simple average of the numbers). 2 . Then for each number: subtract the Mean and square the result."}
+{"idx": 5, "title": "Standard Deviation Formula and Uses, vs. Variance", "date": "", "ddg_snippet": "Calculating Standard Deviation . Standard deviation is calculated as follows: Calculate the mean of all data points: Add the data point values and divide by the number of data points.", "subpage_snippet": "", "source": "www.investopedia.com", "link": "https://www.investopedia.com/terms/s/standarddeviation.asp", "content": "Calculating Standard Deviation . Standard deviation is calculated as follows: Calculate the mean of all data points: Add the data point values and divide by the number of data points."}
+{"idx": 6, "title": "How to Calculate Standard Deviation ? - GeeksforGeeks", "date": "", "ddg_snippet": "Steps for Calculations of Standard Deviation . We can calculate the standard deviation using following steps: Step 1: Calculate the mean (average) of the data set.", "subpage_snippet": "", "source": "www.geeksforgeeks.org", "link": "https://www.geeksforgeeks.org/maths/how-to-calculate-standard-deviation/", "content": "Steps for Calculations of Standard Deviation . We can calculate the standard deviation using following steps: Step 1: Calculate the mean (average) of the data set."}
+{"idx": 7, "title": "Standard Deviation Calculator", "date": "", "ddg_snippet": "Use this calculator to easily calculate the standard deviation of a sample, or to estimate the population standard deviation based on a random sample from it. Standard deviation for binomial data.", "subpage_snippet": "", "source": "www.gigacalculator.com", "link": "https://www.gigacalculator.com/calculators/standard-deviation-calculator.php", "content": "Use this calculator to easily calculate the standard deviation of a sample, or to estimate the population standard deviation based on a random sample from it. Standard deviation for binomial data."}
+{"idx": 8, "title": "Standard deviation : calculating step by step (article) | Khan Academy", "date": "", "ddg_snippet": "Calculating standard deviation step by step. Standard deviation of a population.", "subpage_snippet": "", "source": "www.khanacademy.org", "link": "https://www.khanacademy.org/math/statistics-probability/summarizing-quantitative-data/variance-standard-deviation-population/a/calculating-standard-deviation-step-by-step", "content": "Calculating standard deviation step by step. Standard deviation of a population."}
+{"idx": 9, "title": "Calculate the standard deviation of the following data: 10, 20, 15...", "date": "", "ddg_snippet": "For calculating mean, we will divide the sum of given terms by the number of terms. After that, we will take derivatives from mean in the frequency distribution table and then square those deviations . Using all the columns we will find standard deviation of the data using formula.", "subpage_snippet": "", "source": "www.vedantu.com", "link": "https://www.vedantu.com/question-answer/calculate-the-standard-deviation-of-the-class-11-maths-cbse-5faf9c9a4855db14f9d07879", "content": "For calculating mean, we will divide the sum of given terms by the number of terms. After that, we will take derivatives from mean in the frequency distribution table and then square those deviations . Using all the columns we will find standard deviation of the data using formula."}
diff --git a/data/sampled_jsons/Beimel_Dynamic_Algorithms_Adaptive_Adversary_full_paper_returns_solution.jsonl b/data/sampled_jsons/Beimel_Dynamic_Algorithms_Adaptive_Adversary_full_paper_returns_solution.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..2e924f2f46a0efbe105fe9c9adf73361fb8ffb2c
--- /dev/null
+++ b/data/sampled_jsons/Beimel_Dynamic_Algorithms_Adaptive_Adversary_full_paper_returns_solution.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Dynamic Algorithms against an Adaptive Adversary: Generic ... DynamicAlgorithmsAgainstanAdaptiveAdversary ... [PDF] Dynamic algorithms against an adaptive adversary ... Dynamic Algorithms Against an Adaptive Adversary: Generic ... Rounding Dynamic Matchings Against an Adaptive Adversary Dynamic algorithms against an adaptive adversary: generic ... Dynamic Algorithms Against an Adaptive Adversary: Generic ...", "date": "", "ddg_snippet": "A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adap- tive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. Nov 9, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... A general reduction is given transforming a dynamic algorithm against an oblivious adversary to a dynamic algorithms robust against an adaptive adversary , which maintains several copies of the oblivious algorithm and uses differential privacy to protect their random bits. Given an input that undergoes a sequence of updates, a dynamic algorithm maintains a valid solution to some predefined ... Nov 7, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem. Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ... We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis.", "subpage_snippet": "", "source": "par.nsf.gov", "link": "https://par.nsf.gov/servlets/purl/10440813", "content": "A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adap- tive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. Nov 9, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... A general reduction is given transforming a dynamic algorithm against an oblivious adversary to a dynamic algorithms robust against an adaptive adversary , which maintains several copies of the oblivious algorithm and uses differential privacy to protect their random bits. Given an input that undergoes a sequence of updates, a dynamic algorithm maintains a valid solution to some predefined ... Nov 7, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem. Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ... We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis."}
+{"idx": 1, "title": "DynamicAlgorithmsAgainstanAdaptiveAdversary ...", "date": "", "ddg_snippet": "Nov 9, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2111.03980", "content": "Nov 9, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ..."}
+{"idx": 2, "title": "[PDF] Dynamic algorithms against an adaptive adversary ...", "date": "", "ddg_snippet": "A general reduction is given transforming a dynamic algorithm against an oblivious adversary to a dynamic algorithms robust against an adaptive adversary , which maintains several copies of the oblivious algorithm and uses differential privacy to protect their random bits. Given an input that undergoes a sequence of updates, a dynamic algorithm maintains a valid solution to some predefined ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Dynamic-algorithms-against-an-adaptive-adversary:-Beimel-Kaplan/aa9988bf55a092f8050b11818ca3647543263351", "content": "A general reduction is given transforming a dynamic algorithm against an oblivious adversary to a dynamic algorithms robust against an adaptive adversary , which maintains several copies of the oblivious algorithm and uses differential privacy to protect their random bits. Given an input that undergoes a sequence of updates, a dynamic algorithm maintains a valid solution to some predefined ..."}
+{"idx": 3, "title": "Dynamic Algorithms Against an Adaptive Adversary: Generic ... Rounding Dynamic Matchings Against an Adaptive Adversary Dynamic algorithms against an adaptive adversary: generic ... Dynamic Algorithms Against an Adaptive Adversary: Generic ...", "date": "", "ddg_snippet": "Nov 7, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem. Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ... We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2111.03980", "content": "Nov 7, 2021 · A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after seeing the previous outputs of the algorithm . We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy ... This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem. Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ... We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis."}
+{"idx": 4, "title": "Rounding Dynamic Matchings Against an Adaptive Adversary", "date": "", "ddg_snippet": "This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem.", "subpage_snippet": "", "source": "www.cs.cmu.edu", "link": "https://www.cs.cmu.edu/~dwajc/pdfs/wajc19.pdf", "content": "This paper provides the first randomized dynamic matching algorithms which work against adaptive adversaries and outperform deterministic algorithms for this problem."}
+{"idx": 5, "title": "Dynamic algorithms against an adaptive adversary: generic ...", "date": "", "ddg_snippet": "Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ...", "subpage_snippet": "", "source": "par.nsf.gov", "link": "https://par.nsf.gov/biblio/10440813-dynamic-algorithms-against-adaptive-adversary-generic-constructions-lower-bounds", "content": "Author (s) / Creator (s): Beimel , Amos; Kaplan, Haim; Mansour, Yishay; Nissim, Kobbi; Saranurak, Thatchaphol; Stemmer, Uri Date Published: 2022-06-01 Journal Name: STOC 2022: Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing Page Range / eLocation ID: 1671 to 1684 Format (s): Medium: X Sponsoring Org: National Science ..."}
+{"idx": 6, "title": "Dynamic Algorithms Against an Adaptive Adversary: Generic ...", "date": "", "ddg_snippet": "We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis.", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2111.03980", "content": "We obtain faster dynamic algorithms against an adaptive adversary and separation results between what is achievable in the oblivious vs. adaptive settings. To get these results we exploit techniques from differential privacy, cryptography, and adaptive data analysis."}
+{"idx": 7, "title": "[2111.03980v1] Dynamic Algorithms Against an Adaptive Adversary ...", "date": "", "ddg_snippet": "View a PDF of the paper titled Dynamic Algorithms Against an Adaptive Adversary : Generic Constructions and Lower Bounds, by Amos Beimel and 5 other authors.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2111.03980v1", "content": "View a PDF of the paper titled Dynamic Algorithms Against an Adaptive Adversary : Generic Constructions and Lower Bounds, by Amos Beimel and 5 other authors."}
+{"idx": 8, "title": "Dynamic algorithms against an adaptive adversary : generic...", "date": "", "ddg_snippet": "We present the first fully dynamic algorithm that maintains the shortest paths against an adaptive adversary in subquadratic update time. This is obtained via a combinatorial reduction that allows reconstructing the shortest paths with only a few distance estimates.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/361232054_Dynamic_algorithms_against_an_adaptive_adversary_generic_constructions_and_lower_bounds", "content": "We present the first fully dynamic algorithm that maintains the shortest paths against an adaptive adversary in subquadratic update time. This is obtained via a combinatorial reduction that allows reconstructing the shortest paths with only a few distance estimates."}
+{"idx": 9, "title": "STOC 2022 - Dynamic Algorithms Against an Adaptive Adversary ...", "date": "", "ddg_snippet": "Dynamic Algorithms Against an Adaptive Adversary : Generic Constructions and Lower Bounds Amos Beimel (Ben-Gurion University), Haim Kaplan (Tel Aviv University and Google research), Yishay Mansour (Tel Aviv University and Google research), Kobbi Nissim...", "subpage_snippet": "", "source": "rutube.ru", "link": "https://rutube.ru/video/289472930250da35af1e1357416079ac/", "content": "Dynamic Algorithms Against an Adaptive Adversary : Generic Constructions and Lower Bounds Amos Beimel (Ben-Gurion University), Haim Kaplan (Tel Aviv University and Google research), Yishay Mansour (Tel Aviv University and Google research), Kobbi Nissim..."}
diff --git a/data/sampled_jsons/Book_Crossing_dataset_parameters_alpha_x_min_X_2.38_8_10^5.jsonl b/data/sampled_jsons/Book_Crossing_dataset_parameters_alpha_x_min_X_2.38_8_10^5.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..276ff11876e0269043d9593e23cc2d3de335e627
--- /dev/null
+++ b/data/sampled_jsons/Book_Crossing_dataset_parameters_alpha_x_min_X_2.38_8_10^5.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "BookCorpus - Wikipedia", "date": "", "ddg_snippet": "BookCorpus (also sometimes referred to as the Toronto Book Corpus) is a dataset consisting of the text of around 7,000 self-published books scraped from the indie ebook distribution website Smashwords. [1]", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/BookCorpus", "content": "BookCorpus (also sometimes referred to as the Toronto Book Corpus) is a dataset consisting of the text of around 7,000 self-published books scraped from the indie ebook distribution website Smashwords. [1]"}
+{"idx": 1, "title": "Book-Crossing-Dataset/Project Part I.docx at main - GitHub", "date": "", "ddg_snippet": "Recommendation system for Book - Crossing Dataset. Contribute to AlexKuoTW/ Book - Crossing -Dataset development by creating an account on GitHub .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/AlexKuoTW/Book-Crossing-Dataset/blob/main/Project+Part+I.docx", "content": "Recommendation system for Book - Crossing Dataset. Contribute to AlexKuoTW/ Book - Crossing -Dataset development by creating an account on GitHub ."}
+{"idx": 2, "title": "An Introduction to the COCO Dataset - Roboflow Blog", "date": "", "ddg_snippet": "Oct 18, 2020 · The COCO dataset also provides a base dataset to train computer vision models in a supervised training method. Once the model is trained on the COCO dataset , it can be fine-tuned to learn other tasks, with a custom dataset .", "subpage_snippet": "", "source": "blog.roboflow.com", "link": "https://blog.roboflow.com/coco-dataset/", "content": "Oct 18, 2020 · The COCO dataset also provides a base dataset to train computer vision models in a supervised training method. Once the model is trained on the COCO dataset , it can be fine-tuned to learn other tasks, with a custom dataset ."}
+{"idx": 3, "title": "MaziyarPanahi/arxflix- dataset -dup-12290-alpaca · Datasets at...", "date": "", "ddg_snippet": "Inference over this number of examples times 10 models evaluated amounts to 2 million inference examples. Finetuning on the other hand, requires all parameters to be trained and training dataset sizes are considerably larger.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/datasets/MaziyarPanahi/arxflix-dataset-dup-12290-alpaca", "content": "Inference over this number of examples times 10 models evaluated amounts to 2 million inference examples. Finetuning on the other hand, requires all parameters to be trained and training dataset sizes are considerably larger."}
+{"idx": 4, "title": "1.1. Linear Models — scikit-learn 1.7.2 documentation", "date": "", "ddg_snippet": "1.1. 10 . Bayesian Regression#. Bayesian regression techniques can be used to include regularization parameters in the estimation procedure: the regularization parameter is not set in a hard sense but tuned to the data at hand.", "subpage_snippet": "", "source": "scikit-learn.org", "link": "https://scikit-learn.org/stable/modules/linear_model.html", "content": "1.1. 10 . Bayesian Regression#. Bayesian regression techniques can be used to include regularization parameters in the estimation procedure: the regularization parameter is not set in a hard sense but tuned to the data at hand."}
+{"idx": 5, "title": "Group sparse Bayesian learning for data -driven discovery of explicit...", "date": "", "ddg_snippet": "Moreover, it is unclear how to efficiently learn a parametric system from multiple data sets with different parameters . This paper presents a group sparse Bayesian learning approaches to uncover the explicit model forms of a parametric dynamical system with estimated uncertainties.", "subpage_snippet": "", "source": "www.aimsciences.org", "link": "https://www.aimsciences.org/article/doi/10.3934/naco.2022040", "content": "Moreover, it is unclear how to efficiently learn a parametric system from multiple data sets with different parameters . This paper presents a group sparse Bayesian learning approaches to uncover the explicit model forms of a parametric dynamical system with estimated uncertainties."}
+{"idx": 6, "title": "Калькулятор уравнений", "date": "", "ddg_snippet": "•lambda — lambda. •= — geq. •pi — pi. alpha — alpha .", "subpage_snippet": "", "source": "mathdf.com", "link": "https://mathdf.com/equ/ru/", "content": "•lambda — lambda. •= — geq. •pi — pi. alpha — alpha ."}
+{"idx": 7, "title": "Create multiple subplots using plt.subplots — Matplotlib...", "date": "", "ddg_snippet": "Matplotlib 3. 10 .6 documentation - Home. Plot types. User guide.When stacking in two directions, the returned axs is a 2D NumPy array. If you have to set parameters for each subplot it's handy to iterate over all subplots in a 2D grid using for ax in axs.flat", "subpage_snippet": "", "source": "matplotlib.org", "link": "https://matplotlib.org/stable/gallery/subplots_axes_and_figures/subplots_demo.html", "content": "Matplotlib 3. 10 .6 documentation - Home. Plot types. User guide.When stacking in two directions, the returned axs is a 2D NumPy array. If you have to set parameters for each subplot it's handy to iterate over all subplots in a 2D grid using for ax in axs.flat"}
+{"idx": 8, "title": "7 Scikit-learn Tricks for Optimized Cross-Validation", "date": "", "ddg_snippet": "Sep 8 , 2025 · This article reveals seven scikit-learn tricks for optimizing cross-validation, along with code examples of their implementation.", "subpage_snippet": "", "source": "machinelearningmastery.com", "link": "https://machinelearningmastery.com/7-scikit-learn-tricks-for-optimized-cross-validation/", "content": "Sep 8 , 2025 · This article reveals seven scikit-learn tricks for optimizing cross-validation, along with code examples of their implementation."}
+{"idx": 9, "title": "Top 7 Cross-Validation Techniques with Python Code", "date": "", "ddg_snippet": "", "subpage_snippet": "", "source": "www.analyticsvidhya.com", "link": "https://www.analyticsvidhya.com/blog/2021/11/top-cross-validation-techniques-with-python-code/", "content": ""}
diff --git a/data/sampled_jsons/CLIP_image_encoder_semantic_space_vs_VAE_latent_space_technical_differences_dimensionality.jsonl b/data/sampled_jsons/CLIP_image_encoder_semantic_space_vs_VAE_latent_space_technical_differences_dimensionality.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..07b68af80085511bfa9d34e4b4320762f2ba4291
--- /dev/null
+++ b/data/sampled_jsons/CLIP_image_encoder_semantic_space_vs_VAE_latent_space_technical_differences_dimensionality.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "TokLIP: Marry Visual Tokens to CLIP for Multimodal", "date": "", "ddg_snippet": "The token encoder is initialized from a pre-trained bidirectional CLIP vision encoder and semanticizes the discrete visual codes under a data ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.05422v1", "content": "The token encoder is initialized from a pre-trained bidirectional CLIP vision encoder and semanticizes the discrete visual codes under a data ..."}
+{"idx": 1, "title": "Visual Lexicon: Rich Image Features in Language Space", "date": "", "ddg_snippet": "We propose ViLex , an image encoder that maps images into the vocabulary space , effectively preserving semantic information and intricate visual ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.06774v1", "content": "We propose ViLex , an image encoder that maps images into the vocabulary space , effectively preserving semantic information and intricate visual ..."}
+{"idx": 2, "title": "REPA-E: Unlocking VAE for End-to-End Tuning with Latent", "date": "", "ddg_snippet": "... tuning the VAE and latent diffusion model during training , automatically improves the latent space structure across different VAE architectures.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.10483v1", "content": "... tuning the VAE and latent diffusion model during training , automatically improves the latent space structure across different VAE architectures."}
+{"idx": 3, "title": "Master Fine-Grained Image Control: Discover Playground", "date": "", "ddg_snippet": "... encoders (such as T5 or CLIP ), PGv3 can better capture the complex semantics , logical relationships, and detailed descriptions in text, and convert ...", "subpage_snippet": "", "source": "kcgod.com", "link": "https://kcgod.com/Playground-v3-technical-report", "content": "... encoders (such as T5 or CLIP ), PGv3 can better capture the complex semantics , logical relationships, and detailed descriptions in text, and convert ..."}
+{"idx": 4, "title": "DisCo-Diff: Enhancing Continuous Diffusion Models with Discrete", "date": "", "ddg_snippet": "A careful hierarchical design can encourage different discrete latents to encode different image characteristics, such as shape vs .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2407.03300v1", "content": "A careful hierarchical design can encourage different discrete latents to encode different image characteristics, such as shape vs ."}
+{"idx": 5, "title": "Discrete Distribution Networks", "date": "", "ddg_snippet": "VAEs encode data into a simple distribution’s latent space through an Encoder , and the Decoder is trained to reconstruct the original data from this ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2401.00036v3", "content": "VAEs encode data into a simple distribution’s latent space through an Encoder , and the Decoder is trained to reconstruct the original data from this ..."}
+{"idx": 6, "title": "Diffusion Models", "date": "", "ddg_snippet": "Unlike VAE or flow models , diffusion models are learned with a fixed procedure and the latent variable has high dimensionality (same as the original ...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/blog/Esmail-AGumaan/diffusion-models", "content": "Unlike VAE or flow models , diffusion models are learned with a fixed procedure and the latent variable has high dimensionality (same as the original ..."}
+{"idx": 7, "title": "ComfyUI Beginner's Guide: Master Features with Free Online", "date": "", "ddg_snippet": "Latent Space Transformation: The feature vectors from the Text Encoder and a random noise image are transformed into a latent space .", "subpage_snippet": "", "source": "www.runcomfy.com", "link": "https://www.runcomfy.com/tutorials/comfyui-beginners-guide", "content": "Latent Space Transformation: The feature vectors from the Text Encoder and a random noise image are transformed into a latent space ."}
+{"idx": 8, "title": "Color-Diffusion: using diffusion models to colorize black and", "date": "", "ddg_snippet": "... the 3x64x64 pixel space image is smaller than the latents you would get from encoding a higher-res image with models like VQGAN or the stablediff VAE ...", "subpage_snippet": "", "source": "news.ycombinator.com", "link": "https://news.ycombinator.com/item?id=36991293", "content": "... the 3x64x64 pixel space image is smaller than the latents you would get from encoding a higher-res image with models like VQGAN or the stablediff VAE ..."}
+{"idx": 9, "title": "Ilya's 30 Papers to Carmack: VLAEs - by theahura", "date": "", "ddg_snippet": "... practitioners use models like variational autoencoders ( VAEs ) to do representation learning, because they naturally have a clear hierarchy of latent ...", "subpage_snippet": "", "source": "theahura.substack.com", "link": "https://theahura.substack.com/p/ilyas-30-papers-to-carmack-vlaes", "content": "... practitioners use models like variational autoencoders ( VAEs ) to do representation learning, because they naturally have a clear hierarchy of latent ..."}
diff --git a/data/sampled_jsons/CoPINN_weight_scheduler_linear_interpolation_easiest_hardest_samples_vie_vih_epoch_i_Ne_formula_year_2024.jsonl b/data/sampled_jsons/CoPINN_weight_scheduler_linear_interpolation_easiest_hardest_samples_vie_vih_epoch_i_Ne_formula_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..95831306286d3c10d71e66968e31ba898652a863
--- /dev/null
+++ b/data/sampled_jsons/CoPINN_weight_scheduler_linear_interpolation_easiest_hardest_samples_vie_vih_epoch_i_Ne_formula_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "CoPINN : Cognitive Physics-Informed Neural Networks", "date": "", "ddg_snippet": "Therefore, the learning objective of CoPINN can be formulated as followsFor a training phase with Ne epochs , we expect the weight of the easiest sample to be 1 in the first epoch and 0 in Ne epochs .", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=4vAa0A98xI", "content": "Therefore, the learning objective of CoPINN can be formulated as followsFor a training phase with Ne epochs , we expect the weight of the easiest sample to be 1 in the first epoch and 0 in Ne epochs ."}
+{"idx": 1, "title": "ICML Poster CoPINN : Physical Informed Neural Network", "date": "", "ddg_snippet": "To deal with this daunting problem, we propose a novel framework named Cognitive Physical Informed Neural Network ( CoPINN ) that imitates the human cognitive learning manner from easy to hard .", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46458", "content": "To deal with this daunting problem, we propose a novel framework named Cognitive Physical Informed Neural Network ( CoPINN ) that imitates the human cognitive learning manner from easy to hard ."}
+{"idx": 2, "title": "machinelearningmastery.com/difference-between-a-batch-and-an- epoch", "date": "", "ddg_snippet": "The number of epochs can be set to an integer value between one and infinity.", "subpage_snippet": "", "source": "machinelearningmastery.com", "link": "https://machinelearningmastery.com/difference-between-a-batch-and-an-epoch/", "content": "The number of epochs can be set to an integer value between one and infinity."}
+{"idx": 3, "title": "Formula 55 | Ставки на спорт онлайн - Букмекерская контора ставок", "date": "", "ddg_snippet": "Австралия. Hard . Квалификация (3-й сет - супер тай-брейк).", "subpage_snippet": "", "source": "formula55.me", "link": "https://formula55.me/", "content": "Австралия. Hard . Квалификация (3-й сет - супер тай-брейк)."}
+{"idx": 4, "title": "PyTorch Loss Functions: The Ultimate Guide", "date": "", "ddg_snippet": "NLL does not only care about the prediction being correct but also about the model being certain about the prediction with a high score. The Pytorch NLL Loss is expressed as: where x is the input, y is the target, w is the weight , and N is the batch size. When could it be used?", "subpage_snippet": "", "source": "neptune.ai", "link": "https://neptune.ai/blog/pytorch-loss-functions", "content": "NLL does not only care about the prediction being correct but also about the model being certain about the prediction with a high score. The Pytorch NLL Loss is expressed as: where x is the input, y is the target, w is the weight , and N is the batch size. When could it be used?"}
+{"idx": 5, "title": "Video Generation with Wan2.1 - Chutes Documentation", "date": "", "ddg_snippet": "Video Interpolation . Create smooth transitions between keyframes: classInterpolationInput(BaseModel): start_prompt: str end_prompt: str interpolation _steps: int = Field(5, ge=3, le=10) resolution: Resolution = Resolution.WIDESCREEN @.", "subpage_snippet": "", "source": "chutes.ai", "link": "https://chutes.ai/docs/examples/video-generation", "content": "Video Interpolation . Create smooth transitions between keyframes: classInterpolationInput(BaseModel): start_prompt: str end_prompt: str interpolation _steps: int = Field(5, ge=3, le=10) resolution: Resolution = Resolution.WIDESCREEN @."}
+{"idx": 6, "title": "“ I Want To Enjoy Every Minute That I Can”: For... | British Vogue", "date": "", "ddg_snippet": "In just a few short years, one name has overtaken all others in Formula 1: Lando Norris. As the British driver turned Gen Z pin-up eyes the world championship, he tells Kate Lloyd about life in the fast lane.", "subpage_snippet": "", "source": "www.vogue.co.uk", "link": "https://www.vogue.co.uk/article/lando-norris-interview", "content": "In just a few short years, one name has overtaken all others in Formula 1: Lando Norris. As the British driver turned Gen Z pin-up eyes the world championship, he tells Kate Lloyd about life in the fast lane."}
+{"idx": 7, "title": "ONE : eCommerce", "date": "", "ddg_snippet": "...HTTPS ecomm.one- line .com/one-ecom/ schedule /vessel- schedule .", "subpage_snippet": "", "source": "ecomm.one-line.com", "link": "https://ecomm.one-line.com/one-ecom/schedule/vessel-schedule", "content": "...HTTPS ecomm.one- line .com/one-ecom/ schedule /vessel- schedule ."}
+{"idx": 8, "title": "Wolfram|Alpha: Computational Intelligence", "date": "", "ddg_snippet": "Compute answers using Wolfram's breakthrough technology & knowledgebase, relied on by millions of students & professionals. For math, science, nutrition, history, geography, engineering, mathematics, linguistics, sports, finance, music…", "subpage_snippet": "", "source": "www.wolframalpha.com", "link": "https://www.wolframalpha.com/", "content": "Compute answers using Wolfram's breakthrough technology & knowledgebase, relied on by millions of students & professionals. For math, science, nutrition, history, geography, engineering, mathematics, linguistics, sports, finance, music…"}
+{"idx": 9, "title": "Perplexity", "date": "", "ddg_snippet": "Perplexity is a free AI-powered answer engine that provides accurate, trusted, and real-time answers to any question.", "subpage_snippet": "", "source": "www.perplexity.ai", "link": "https://www.perplexity.ai/", "content": "Perplexity is a free AI-powered answer engine that provides accurate, trusted, and real-time answers to any question."}
diff --git a/data/sampled_jsons/Concept_Bottleneck_Models_Koh_2020_abstract_arxiv_year_2020.jsonl b/data/sampled_jsons/Concept_Bottleneck_Models_Koh_2020_abstract_arxiv_year_2020.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..641b187f827dcb9e4cd42a617ab39d454b0be13b
--- /dev/null
+++ b/data/sampled_jsons/Concept_Bottleneck_Models_Koh_2020_abstract_arxiv_year_2020.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Explainable artificial intelligence - Wikipedia", "date": "", "ddg_snippet": "20 Concept Bottleneck Models , which use concept -level abstractions to explain model reasoning, are examples of this and can be applied in both image ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Explainable_artificial_intelligence", "content": "20 Concept Bottleneck Models , which use concept -level abstractions to explain model reasoning, are examples of this and can be applied in both image ..."}
+{"idx": 1, "title": "[2007.04612] Concept Bottleneck Models - arXiv.org", "date": "", "ddg_snippet": "Jul 9, 2020 · On x-ray grading and bird identification, concept bottleneck models achieve competitive accuracy with standard end-to-end models, while enabling interpretation in terms of high-level clinical concepts (\"bone spurs\") or bird attributes (\"wing color\").", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2007.04612", "content": "Jul 9, 2020 · On x-ray grading and bird identification, concept bottleneck models achieve competitive accuracy with standard end-to-end models, while enabling interpretation in terms of high-level clinical concepts (\"bone spurs\") or bird attributes (\"wing color\")."}
+{"idx": 2, "title": "Concept Bottleneck Models - arXiv.org", "date": "", "ddg_snippet": "Concept bottleneck models can compete on task accuracy while supporting intervention and interpretation, allowing practitioners to reason about these models in terms of high-level concepts they are familiar with, and enabling more effective human-model collaboration through test-time in-tervention.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2007.04612", "content": "Concept bottleneck models can compete on task accuracy while supporting intervention and interpretation, allowing practitioners to reason about these models in terms of high-level concepts they are familiar with, and enabling more effective human-model collaboration through test-time in-tervention."}
+{"idx": 3, "title": "Concept Bottleneck Models - PMLR", "date": "", "ddg_snippet": "Proceedings of Machine Learning Research [edit] Concept Bottleneck Models Pang Wei Koh , Thao Nguyen, Yew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, Percy Liang Proceedings of the 37th International Conference on Machine Learning, PMLR 119:5338-5348, 2020 .", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v119/koh20a.html", "content": "Proceedings of Machine Learning Research [edit] Concept Bottleneck Models Pang Wei Koh , Thao Nguyen, Yew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, Percy Liang Proceedings of the 37th International Conference on Machine Learning, PMLR 119:5338-5348, 2020 ."}
+{"idx": 4, "title": "Bayesian Concept Bottleneck Models with LLM Priors - arXiv.org", "date": "", "ddg_snippet": "5 hours ago · Abstract Concept Bottleneck Models (CBMs) have been proposed as a compromise between white-box and black-box models, aiming to achieve interpretability without sacrificing accuracy.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.15555v2", "content": "5 hours ago · Abstract Concept Bottleneck Models (CBMs) have been proposed as a compromise between white-box and black-box models, aiming to achieve interpretability without sacrificing accuracy."}
+{"idx": 5, "title": "Diverse Concept Proposals for Concept Bottleneck Models", "date": "", "ddg_snippet": "Dec 24, 2024 · Recently popularized by Koh et al. ( 2020 ), concept bottleneck models (CBM) prove to be a highly flexible and (hopefully) interpretable model class. Inputs 𝑥 x italic_x are first converted to a small set of concepts 𝑐 c italic_c that are used to make the ultimate prediction 𝑦 y italic_y. This allows the human operator to inspect the concepts that form the basis of the prediction and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.18059v1", "content": "Dec 24, 2024 · Recently popularized by Koh et al. ( 2020 ), concept bottleneck models (CBM) prove to be a highly flexible and (hopefully) interpretable model class. Inputs 𝑥 x italic_x are first converted to a small set of concepts 𝑐 c italic_c that are used to make the ultimate prediction 𝑦 y italic_y. This allows the human operator to inspect the concepts that form the basis of the prediction and ..."}
+{"idx": 6, "title": "Concept bottleneck models | Proceedings of the 37th ...", "date": "", "ddg_snippet": "Jul 13, 2020 · On x-ray grading and bird identification, concept bottleneck models achieve competitive accuracy with standard end-to-end models, while enabling interpretation in terms of high-level clinical concepts (\"bone spurs\") or bird attributes (\"wing color\").", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/abs/10.5555/3524938.3525433", "content": "Jul 13, 2020 · On x-ray grading and bird identification, concept bottleneck models achieve competitive accuracy with standard end-to-end models, while enabling interpretation in terms of high-level clinical concepts (\"bone spurs\") or bird attributes (\"wing color\")."}
+{"idx": 7, "title": "Concept Bottleneck Language Models For protein design", "date": "", "ddg_snippet": "To achieve this, we modify the standard masked language model architecture by adding a concept bottleneck layer to explicitly incorporate human ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.06090v2", "content": "To achieve this, we modify the standard masked language model architecture by adding a concept bottleneck layer to explicitly incorporate human ..."}
+{"idx": 8, "title": "If Concept Bottlenecks are the Question, are Foundation Models", "date": "", "ddg_snippet": "Concept Bottleneck Models (CBMs) [ 41 ] combine two neural modules: a concept extractor f : 𝒳 → ℝ k : 𝑓 → 𝒳 superscript ℝ 𝑘 f ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.19774v2", "content": "Concept Bottleneck Models (CBMs) [ 41 ] combine two neural modules: a concept extractor f : 𝒳 → ℝ k : 𝑓 → 𝒳 superscript ℝ 𝑘 f ..."}
+{"idx": 9, "title": "CLIP-QDA: An Explainable Concept Bottleneck Model", "date": "", "ddg_snippet": "A class of networks that effectively exploits this notion is Concept Bottleneck Models (CBMs) ( Koh et al., 2020 ) .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2312.00110v3", "content": "A class of networks that effectively exploits this notion is Concept Bottleneck Models (CBMs) ( Koh et al., 2020 ) ."}
diff --git a/data/sampled_jsons/DART_CVPR_Park_Sang-Jun_disease-matching_constraint_lambda_m_=_1.jsonl b/data/sampled_jsons/DART_CVPR_Park_Sang-Jun_disease-matching_constraint_lambda_m_=_1.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..44296beb3852e677fd5005ac7c820508378fa88f
--- /dev/null
+++ b/data/sampled_jsons/DART_CVPR_Park_Sang-Jun_disease-matching_constraint_lambda_m_=_1.jsonl
@@ -0,0 +1 @@
+{"idx": 0, "title": "", "date": "", "ddg_snippet": "", "subpage_snippet": "", "source": "", "link": "", "content": ""}
diff --git a/data/sampled_jsons/DART_Disease-aware_Image-Text_Alignment_equation_5_total_loss_stage_1_lambda_m_disease-matching_cons_year_2024.jsonl b/data/sampled_jsons/DART_Disease-aware_Image-Text_Alignment_equation_5_total_loss_stage_1_lambda_m_disease-matching_cons_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..103c49dff21f8b65ce714e55c0269607af57de71
--- /dev/null
+++ b/data/sampled_jsons/DART_Disease-aware_Image-Text_Alignment_equation_5_total_loss_stage_1_lambda_m_disease-matching_cons_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "DART : Disease - aware Image - Text Alignment and Self-correcting...", "date": "", "ddg_snippet": "Image -to- text Retrieval with Disease - matching . After aligning the image and text embeddings in a shared embedding space using contrastive learning, we retrieve reports relevant to the image features.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.11786v1", "content": "Image -to- text Retrieval with Disease - matching . After aligning the image and text embeddings in a shared embedding space using contrastive learning, we retrieve reports relevant to the image features."}
+{"idx": 1, "title": "(PDF) DART : Disease - aware Image - Text Alignment and...", "date": "", "ddg_snippet": "disease - aware image -to- text alignment . First , an input im-. loss (CL), image -to- text retrieval ( I 2T), disease - matching constraint (DM), and self-correction (SC). A “” indicates the presence of each. component, while “-” denotes its absence.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/390845711_DART_Disease-aware_Image-Text_Alignment_and_Self-correcting_Re-alignment_for_Trustworthy_Radiology_Report_Generation", "content": "disease - aware image -to- text alignment . First , an input im-. loss (CL), image -to- text retrieval ( I 2T), disease - matching constraint (DM), and self-correction (SC). A “” indicates the presence of each. component, while “-” denotes its absence."}
+{"idx": 2, "title": "Brat Text Generator - Charli XCX Inspired Text Maker", "date": "", "ddg_snippet": "Generate Charli XCX Brat-inspired text with our Brat Text Generator. Customize your text with stylish fonts and effects!", "subpage_snippet": "", "source": "bratgenerator.io", "link": "https://bratgenerator.io/old/", "content": "Generate Charli XCX Brat-inspired text with our Brat Text Generator. Customize your text with stylish fonts and effects!"}
+{"idx": 3, "title": "Dart", "date": "", "ddg_snippet": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation. Request Code.", "subpage_snippet": "", "source": "www.catalyzex.com", "link": "https://www.catalyzex.com/s/Dart", "content": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation. Request Code."}
+{"idx": 4, "title": "Block of Equations Alignment \\begin{ equation *} - LaTeX.org | Forum", "date": "", "ddg_snippet": "The imgshare.io links are dead, no images there. But you can post the images here as attachments. The \"Attachments\" link is below the text edit field, when you write a post.", "subpage_snippet": "", "source": "latex.org", "link": "https://latex.org/forum/viewtopic.php?t=34561", "content": "The imgshare.io links are dead, no images there. But you can post the images here as attachments. The \"Attachments\" link is below the text edit field, when you write a post."}
+{"idx": 5, "title": "Articles by Dong-Ho Shin | Synthical", "date": "", "ddg_snippet": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation.", "subpage_snippet": "", "source": "synthical.com", "link": "https://synthical.com/profile/3e7eed8b-226c-4317-a4f7-ae93f39b25b8/articles", "content": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation."}
+{"idx": 6, "title": "Latest 15 Papers - May 21, 2025", "date": "", "ddg_snippet": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation.", "subpage_snippet": "", "source": "viao.co.uk", "link": "https://viao.co.uk/blog/latest-15-papers-may-21-1747789981215", "content": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation."}
+{"idx": 7, "title": "Face Swap Online Free", "date": "", "ddg_snippet": "You can use Remaker's Face Swapper to swap faces online for free. Here are the steps: 1 . Upload the original image that contains the face you want to replace.", "subpage_snippet": "", "source": "remaker.ai", "link": "https://remaker.ai/face-swap-free/", "content": "You can use Remaker's Face Swapper to swap faces online for free. Here are the steps: 1 . Upload the original image that contains the face you want to replace."}
+{"idx": 8, "title": "Joseph is a super minimal content focus theme for Jekyll.", "date": "", "ddg_snippet": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation. Sang-Jun Park, Keun-Soo Heo, Dong-Hee Shin, Young-Han Son, Ji-Hye Oh, Tae-Eui Kam.", "subpage_snippet": "", "source": "donghee-shin-page.github.io", "link": "https://donghee-shin-page.github.io/publications/", "content": "DART : Disease - aware Image - Text Alignment and Self-correcting Re- alignment for Trustworthy Radiology Report Generation. Sang-Jun Park, Keun-Soo Heo, Dong-Hee Shin, Young-Han Son, Ji-Hye Oh, Tae-Eui Kam."}
+{"idx": 9, "title": "CVPR2025 Accepted Papers-CSDN博客", "date": "", "ddg_snippet": "Single Image Plant Modeling Zhihao Liu · Zhanglin Cheng · Naoto Yokoya Alignment , Mining and Fusion: Representation Alignment with Hard Negative Mining and Selective Knowledge Fusion for Medical Visual Question Answering Yuanhao Zou · Zhaozheng Yin BlobGEN-Vid...", "subpage_snippet": "", "source": "blog.csdn.net", "link": "https://blog.csdn.net/u013963578/article/details/146183100", "content": "Single Image Plant Modeling Zhihao Liu · Zhanglin Cheng · Naoto Yokoya Alignment , Mining and Fusion: Representation Alignment with Hard Negative Mining and Selective Knowledge Fusion for Medical Visual Question Answering Yuanhao Zou · Zhaozheng Yin BlobGEN-Vid..."}
diff --git a/data/sampled_jsons/DCBM_Data-Efficient_Visual_Concept_Bottleneck_Models_OpenReview_Section_5.jsonl b/data/sampled_jsons/DCBM_Data-Efficient_Visual_Concept_Bottleneck_Models_OpenReview_Section_5.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..d98e0483dd41d76057cb3041653344e7ff2c9ce4
--- /dev/null
+++ b/data/sampled_jsons/DCBM_Data-Efficient_Visual_Concept_Bottleneck_Models_OpenReview_Section_5.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models - OpenReview", "date": "", "ddg_snippet": "We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes during concept generation while preserving interpretability. DCBMs define concepts as image regions detected by segmentation or detection foundation models , allowing each image to generate multiple concepts across different granularities.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=BdO4R6XxUH", "content": "We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes during concept generation while preserving interpretability. DCBMs define concepts as image regions detected by segmentation or detection foundation models , allowing each image to generate multiple concepts across different granularities."}
+{"idx": 1, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "data /: Directories for classes, concepts , datasets, embeddings, and segments. experiments/: Code for experiments detailed in the main paper and supplementary material.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/KathPra/DCBM", "content": "data /: Directories for classes, concepts , datasets, embeddings, and segments. experiments/: Code for experiments detailed in the main paper and supplementary material."}
+{"idx": 2, "title": "ICML DCBM: Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "Poster in Workshop: Actionable Interpretability DCBM : Data-Efficient Visual Concept Bottleneck Models Katharina Prasse · Patrick Knab · Sascha Marton · Christian Bartelt · Margret Keuper [ Abstract ] [ Project Page ] [ OpenReview ] Sat 19 Jul 10:40 a.m. PDT — 11:40 a.m. PDT", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/49558", "content": "Poster in Workshop: Actionable Interpretability DCBM : Data-Efficient Visual Concept Bottleneck Models Katharina Prasse · Patrick Knab · Sascha Marton · Christian Bartelt · Margret Keuper [ Abstract ] [ Project Page ] [ OpenReview ] Sat 19 Jul 10:40 a.m. PDT — 11:40 a.m. PDT"}
+{"idx": 3, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "Abstract Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on con-cept sets extracted from large language models or extensive image corpora, limiting their effec-tiveness in data -sparse scenarios.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2412.11576", "content": "Abstract Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on con-cept sets extracted from large language models or extensive image corpora, limiting their effec-tiveness in data -sparse scenarios."}
+{"idx": 4, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models :: MPG.PuRe", "date": "", "ddg_snippet": "Content show hide Free keywords: Computer Science, Computer Vision and Pattern Recognition, cs.CV Abstract: Concept Bottleneck Models (CBMs) enhance the interpretability of neural", "subpage_snippet": "", "source": "pure.mpg.de", "link": "https://pure.mpg.de/pubman/faces/ViewItemFullPage.jsp?itemId=item_3636912_1&view=ACTIONS", "content": "Content show hide Free keywords: Computer Science, Computer Vision and Pattern Recognition, cs.CV Abstract: Concept Bottleneck Models (CBMs) enhance the interpretability of neural"}
+{"idx": 5, "title": "Revisions | OpenReview", "date": "", "ddg_snippet": "We introduce Data-efficient Concept Bottleneck Models ( DCBMs ) — a new method that works even when we have few images. Instead of relying on predefined concepts , DCBMs use modern AI tools to detect meaningful regions of each image, automatically creating visual concepts that are specific to the dataset.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/revisions?id=BdO4R6XxUH", "content": "We introduce Data-efficient Concept Bottleneck Models ( DCBMs ) — a new method that works even when we have few images. Instead of relying on predefined concepts , DCBMs use modern AI tools to detect meaningful regions of each image, automatically creating visual concepts that are specific to the dataset."}
+{"idx": 6, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on concept sets extracted from large language models or extensive image corpora, limiting their effectiveness in data -sparse scenarios. We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2412.11576", "content": "Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on concept sets extracted from large language models or extensive image corpora, limiting their effectiveness in data -sparse scenarios. We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes ..."}
+{"idx": 7, "title": "GitHub - deepopo/DCBM", "date": "", "ddg_snippet": "This is an implementation of the IEEE TPAMI paper The Decoupling Concept Bottleneck Model ( DCBM ). The vision-language- model (VLM) part is being refined and will be available soon. Figure 1: DCBM Pipeline. (A) DCBM for prediction and interpretation. (B) DCBM for human-machine interaction, including forward intervention and backward rectification.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/deepopo/DCBM", "content": "This is an implementation of the IEEE TPAMI paper The Decoupling Concept Bottleneck Model ( DCBM ). The vision-language- model (VLM) part is being refined and will be available soon. Figure 1: DCBM Pipeline. (A) DCBM for prediction and interpretation. (B) DCBM for human-machine interaction, including forward intervention and backward rectification."}
+{"idx": 8, "title": "DCBM: Data-Efficient Visual Concept Bottleneck Models - OpenReview", "date": "", "ddg_snippet": "Abstract Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on con-cept sets extracted from large language models or extensive image corpora, limiting their effec-tiveness in data -sparse scenarios.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=BdO4R6XxUH", "content": "Abstract Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts . However, current CBMs typically rely on con-cept sets extracted from large language models or extensive image corpora, limiting their effec-tiveness in data -sparse scenarios."}
+{"idx": 9, "title": "www.mpi-inf.mpg.de", "date": "", "ddg_snippet": "scenarios. We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes during concept generation while preserving interpretability.", "subpage_snippet": "", "source": "www.mpi-inf.mpg.de", "link": "https://www.mpi-inf.mpg.de/fileadmin/inf/bibtex/3636912.bib", "content": "scenarios. We propose Data-efficient CBMs ( DCBMs ), which reduce the need for large sample sizes during concept generation while preserving interpretability."}
diff --git a/data/sampled_jsons/DCBM_applicability_limited_by_segmentation_detection_foundation_model.jsonl b/data/sampled_jsons/DCBM_applicability_limited_by_segmentation_detection_foundation_model.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..3f37f80caa92930ea5365e51bddd755929be9265
--- /dev/null
+++ b/data/sampled_jsons/DCBM_applicability_limited_by_segmentation_detection_foundation_model.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "DCBM : Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of the CLIP embedding space for a given use case. Further advances in vision-language models, segmentation , and detection models can enhance results...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.11576v2", "content": "DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of the CLIP embedding space for a given use case. Further advances in vision-language models, segmentation , and detection models can enhance results..."}
+{"idx": 1, "title": "DCBM : Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of the CLIP embedding space for a given use case. Further advances in vision-language models, segmentation , and detection models can enhance results...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.11576v3", "content": "DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of the CLIP embedding space for a given use case. Further advances in vision-language models, segmentation , and detection models can enhance results..."}
+{"idx": 2, "title": "DCBM : Data-Efficient Visual Concept Bottleneck Models", "date": "", "ddg_snippet": "DCBMs use segmentation and detection foundation models to extract image regions as visual concept proposals. DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2412.11576", "content": "DCBMs use segmentation and detection foundation models to extract image regions as visual concept proposals. DCBM ’s applicability is limited by two factors, the suitability of a segmentation foundation model and the expressiveness of."}
+{"idx": 3, "title": "Publications - Max Planck Institut für Informatik", "date": "", "ddg_snippet": "FAIR-TAT: Improving Model Fairness Using Targeted Adversarial Training,” in IEEE/CVF Winter Conference on Applications of Computer Vision (WACV ...", "subpage_snippet": "", "source": "www.mpi-inf.mpg.de", "link": "https://www.mpi-inf.mpg.de/de/departments/computer-vision-and-machine-learning/publications", "content": "FAIR-TAT: Improving Model Fairness Using Targeted Adversarial Training,” in IEEE/CVF Winter Conference on Applications of Computer Vision (WACV ..."}
+{"idx": 4, "title": "Prof. Dr.-Ing. Margret Keuper | Universität Mannheim", "date": "", "ddg_snippet": "Grouping Problems (in applications such as Image and Motion Segmentation and Multiple Object Tracking) ... Text-Guided Graphics Program Synthesis, by ...", "subpage_snippet": "", "source": "www.uni-mannheim.de", "link": "https://www.uni-mannheim.de/dws/people/professors/prof-dr-ing-margret-keuper/", "content": "Grouping Problems (in applications such as Image and Motion Segmentation and Multiple Object Tracking) ... Text-Guided Graphics Program Synthesis, by ..."}
+{"idx": 5, "title": "Computer Vision & Machine Learning (Prof. Dr.-Ing. Margret", "date": "", "ddg_snippet": "Grouping Problems (in applications such as Image and Motion Segmentation and Multiple Object Tracking) ... Benchmarking Of Semantic Segmentation .", "subpage_snippet": "", "source": "www.uni-mannheim.de", "link": "https://www.uni-mannheim.de/dws/research/focus-groups/computer-vision-machine-learning-prof-dr-ing-margret-keuper/", "content": "Grouping Problems (in applications such as Image and Motion Segmentation and Multiple Object Tracking) ... Benchmarking Of Semantic Segmentation ."}
+{"idx": 6, "title": "How Do I Download Pictures From Verizon Phone", "date": "", "ddg_snippet": "DCbm Another how do i download of why we screw to away contain ' idea '? practice of possible segments + room of mysterious Age updates so is us to ...", "subpage_snippet": "", "source": "imerinc.com", "link": "http://imerinc.com/yatchclub/node/how-do-i-download-pictures-from-verizon-phone.htm", "content": "DCbm Another how do i download of why we screw to away contain ' idea '? practice of possible segments + room of mysterious Age updates so is us to ..."}
+{"idx": 7, "title": "CN107306087B - A dual-stage converter and its control method -", "date": "", "ddg_snippet": "... by static converters using discharge tubes with control electrode or semiconductor devices with control electrode using devices of a triode or ...", "subpage_snippet": "", "source": "patents.google.com", "link": "https://patents.google.com/patent/CN107306087B/en", "content": "... by static converters using discharge tubes with control electrode or semiconductor devices with control electrode using devices of a triode or ..."}
+{"idx": 8, "title": "US20060064583A1 - Programming interface for configuring a", "date": "", "ddg_snippet": "H04N21/45 — Management operations performed by the client for facilitating the reception of or the interaction with the content or ...", "subpage_snippet": "", "source": "patents.google.com", "link": "https://patents.google.com/patent/US20060064583A1/en", "content": "H04N21/45 — Management operations performed by the client for facilitating the reception of or the interaction with the content or ..."}
+{"idx": 9, "title": "Filopodyan: An open-source pipeline for the analysis of", "date": "", "ddg_snippet": "Our highly customizable pipeline is widely applicable, capable of detecting filopodia in four different cell types in vitro and in vivo.", "subpage_snippet": "", "source": "rupress.org", "link": "https://rupress.org/jcb/article/216/10/3405/38925/Filopodyan-An-open-source-pipeline-for-the", "content": "Our highly customizable pipeline is widely applicable, capable of detecting filopodia in four different cell types in vitro and in vivo."}
diff --git a/data/sampled_jsons/DIKE-ERIS_duality_limitations.jsonl b/data/sampled_jsons/DIKE-ERIS_duality_limitations.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c94e99d06a442d2cf54dad40e4d5d429c1ad0e18
--- /dev/null
+++ b/data/sampled_jsons/DIKE-ERIS_duality_limitations.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Leuce (mythology) - Wikipedia", "date": "", "ddg_snippet": "The double color, Servius says, made a wreath that represented the duality of the hero's labors in both the upper and the underworld.", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Leuce_(mythology)", "content": "The double color, Servius says, made a wreath that represented the duality of the hero's labors in both the upper and the underworld."}
+{"idx": 1, "title": "A Three-Branch Checks-and-Balances Framework for Context ...", "date": "", "ddg_snippet": "Oct 12, 2024 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=o2afWIxjKD", "content": "Oct 12, 2024 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning."}
+{"idx": 2, "title": "Edward Y. Chang on LinkedIn: How can we address the ...", "date": "", "ddg_snippet": "The DIKE-ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles.", "subpage_snippet": "", "source": "www.linkedin.com", "link": "https://www.linkedin.com/posts/edward-y-chang-218b182_how-can-we-address-the-challenges-of-genai-activity-7255955883786948608-CqQ8", "content": "The DIKE-ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles."}
+{"idx": 3, "title": "[2502.00136v1] A Three-Branch Checks-and-Balances ... A Three-Branch Checks-and-Balances Framework for Context ... Ethical Guardrails for AI: A Checks-and-Balances Approach Edward Y. Chang on LinkedIn: How can we address the ... It Takes a Mirror to Find Flaws: Uncovering Biases with ... Integrating Emotional and Linguistic Mod...", "date": "", "ddg_snippet": "Jan 31, 2025 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning. Oct 12, 2024 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning. Adversarial dynamics: DIKE-ERIS duality creates robust ethical safeguards Practical security: Addresses critical concerns around safe AI deployment and harmful outputs The DIKE-ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles. Feb 17, 2024 · The \\ DIKE-\\ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles. This research develops advanced methodologies for Large Language Models (LLMs) to better manage linguistic behaviors related to emotions and ethics. We introduce DIKE , an adversarial framework that enhances the LLMs' ability to internalize and reflect global human values, adapting to varied cultural contexts to promote transparency and trust among users. The methodology involves detailed ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2502.00136v1", "content": "Jan 31, 2025 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning. Oct 12, 2024 · The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and culturally-aware ethical reasoning. Adversarial dynamics: DIKE-ERIS duality creates robust ethical safeguards Practical security: Addresses critical concerns around safe AI deployment and harmful outputs The DIKE-ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles. Feb 17, 2024 · The \\ DIKE-\\ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles. This research develops advanced methodologies for Large Language Models (LLMs) to better manage linguistic behaviors related to emotions and ethics. We introduce DIKE , an adversarial framework that enhances the LLMs' ability to internalize and reflect global human values, adapting to varied cultural contexts to promote transparency and trust among users. The methodology involves detailed ..."}
+{"idx": 4, "title": "Ethical Guardrails for AI: A Checks-and-Balances Approach", "date": "", "ddg_snippet": "Adversarial dynamics: DIKE-ERIS duality creates robust ethical safeguards Practical security: Addresses critical concerns around safe AI deployment and harmful outputs", "subpage_snippet": "", "source": "www.zerna.io", "link": "https://www.zerna.io/page/security/presentation_set/security-llm-research/presentation/security-ethical-alignment-fairness/slide/security-paper-2502_00136", "content": "Adversarial dynamics: DIKE-ERIS duality creates robust ethical safeguards Practical security: Addresses critical concerns around safe AI deployment and harmful outputs"}
+{"idx": 5, "title": "It Takes a Mirror to Find Flaws: Uncovering Biases with ...", "date": "", "ddg_snippet": "Feb 17, 2024 · The \\ DIKE-\\ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/378268332_It_Takes_a_Mirror_to_Find_Flaws_Uncovering_Biases_with_Reflective_Large_Language_Models", "content": "Feb 17, 2024 · The \\ DIKE-\\ERIS duality , through their adversarial interaction, enables adaptation to diverse cultural contexts while maintaining consistent ethical principles."}
+{"idx": 6, "title": "Integrating Emotional and Linguistic Mod...", "date": "", "ddg_snippet": "This research develops advanced methodologies for Large Language Models (LLMs) to better manage linguistic behaviors related to emotions and ethics. We introduce DIKE , an adversarial framework that enhances the LLMs' ability to internalize and reflect global human values, adapting to varied cultural contexts to promote transparency and trust among users. The methodology involves detailed ...", "subpage_snippet": "", "source": "axi.lims.ac.uk", "link": "https://axi.lims.ac.uk/paper/2405.07076", "content": "This research develops advanced methodologies for Large Language Models (LLMs) to better manage linguistic behaviors related to emotions and ethics. We introduce DIKE , an adversarial framework that enhances the LLMs' ability to internalize and reflect global human values, adapting to varied cultural contexts to promote transparency and trust among users. The methodology involves detailed ..."}
+{"idx": 7, "title": "[2502.00136v1] A Three-Branch Checks-and-Balances ...", "date": "", "ddg_snippet": "31 Jan 2025 — The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This ...", "subpage_snippet": "", "source": "www.arxiv.org", "link": "https://www.arxiv.org/abs/2502.00136v1", "content": "31 Jan 2025 — The adversarial DIKE-ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This ..."}
+{"idx": 8, "title": "A Three-Branch Checks-and-Balances Framework", "date": "", "ddg_snippet": "The adversarial DIKE - ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=o2afWIxjKD", "content": "The adversarial DIKE - ERIS duality enables adaptation to diverse cultural contexts while upholding consistent ethical principles. This architecture addresses limitations of reinforcement learning with human feedback (RLHF) by providing interpretable, adaptable, and..."}
+{"idx": 9, "title": "(PDF) Multi-LLM Agent Collaborative Intelligence: The Path to Artificial...", "date": "", "ddg_snippet": "The DIKE - ERIS duality , through their adversarial interaction, enables. adaptation to diverse cultural contexts while maintaining consistent. ethical principles.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/387541177_Multi-LLM_Agent_Collaborative_Intelligence_The_Path_to_Artificial_General_Intelligence_3rd_edition", "content": "The DIKE - ERIS duality , through their adversarial interaction, enables. adaptation to diverse cultural contexts while maintaining consistent. ethical principles."}
diff --git a/data/sampled_jsons/Definition_4.2_(Per-Instance_Privacy_Loss)_Leveraging_Per-Instance_Privacy_for_Machine_Unlearning.jsonl b/data/sampled_jsons/Definition_4.2_(Per-Instance_Privacy_Loss)_Leveraging_Per-Instance_Privacy_for_Machine_Unlearning.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..2af8008bb0831dbf51eb30adc894dfa21a466c0c
--- /dev/null
+++ b/data/sampled_jsons/Definition_4.2_(Per-Instance_Privacy_Loss)_Leveraging_Per-Instance_Privacy_for_Machine_Unlearning.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Leveraging Per - Instance Privacy for Machine Unlearning", "date": "", "ddg_snippet": ". 4 Per - Instance Unlearning Difficulty Analysis. Our goal is to bound the number of steps Definition 4 . 2 ( Per - Instance Privacy Loss ). Recall that. νT,Dsubscript𝜈𝑇𝐷\\nu_{T,D}italic_ν start_POSTSUBSCRIPT italic_T , italic_D end_POSTSUBSCRIPT.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.18786v1", "content": ". 4 Per - Instance Unlearning Difficulty Analysis. Our goal is to bound the number of steps Definition 4 . 2 ( Per - Instance Privacy Loss ). Recall that. νT,Dsubscript𝜈𝑇𝐷\\nu_{T,D}italic_ν start_POSTSUBSCRIPT italic_T , italic_D end_POSTSUBSCRIPT."}
+{"idx": 1, "title": "Leveraging Per-Instance Privacy for Machine Unlearning", "date": "", "ddg_snippet": "We present a principled, per - instance approach to quantifying the dificulty of unlearning via fine-tuning.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2505.18786", "content": "We present a principled, per - instance approach to quantifying the dificulty of unlearning via fine-tuning."}
+{"idx": 2, "title": "Leveraging Per - Instance Privacy for Machine Unlearning", "date": "", "ddg_snippet": "Definition 4 . 2 ( Per - Instance Privacy Loss ). Per - instance Privacy loss We compute the terms in the privacy loss P (x, α) stated in Definition 4 . 2 by taking a Monte-Carlo estimate from a single training run with check-points w0, ws1 , ws2 , · · · , wsN , i.e.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=0A4Y9qRnu9", "content": "Definition 4 . 2 ( Per - Instance Privacy Loss ). Per - instance Privacy loss We compute the terms in the privacy loss P (x, α) stated in Definition 4 . 2 by taking a Monte-Carlo estimate from a single training run with check-points w0, ws1 , ws2 , · · · , wsN , i.e."}
+{"idx": 3, "title": "RL-Finetuned LLMs for Privacy-Preserving Synthetic Rewriting", "date": "", "ddg_snippet": "... focus on pattern-matching for structured PII, often fall short in addressing complex, context-dependent implicit and stylometric privacy leakages [ 4 ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.19286v1", "content": "... focus on pattern-matching for structured PII, often fall short in addressing complex, context-dependent implicit and stylometric privacy leakages [ 4 ..."}
+{"idx": 4, "title": "How to choose the right reserved instance in Azure | Microsoft", "date": "", "ddg_snippet": "A reserved instance is a way to prepay for a certain amount of compute usage in Azure for a fixed period of time (either one or three years [1]), at ...", "subpage_snippet": "", "source": "techcommunity.microsoft.com", "link": "https://techcommunity.microsoft.com/blog/finopsblog/how-to-choose-the-right-reserved-instance-in-azure/4124218", "content": "A reserved instance is a way to prepay for a certain amount of compute usage in Azure for a fixed period of time (either one or three years [1]), at ..."}
+{"idx": 5, "title": "Decentralised, collaborative, and privacy-preserving machine", "date": "", "ddg_snippet": "... training frameworks either do not provide the correct level of privacy protection for individual patients or achieve the best utility- privacy trade ...", "subpage_snippet": "", "source": "www.thelancet.com", "link": "https://www.thelancet.com/journals/ebiom/article/PIIS2352-3964(24)00041-0/fulltext", "content": "... training frameworks either do not provide the correct level of privacy protection for individual patients or achieve the best utility- privacy trade ..."}
+{"idx": 6, "title": "Why Production Machine Vision Systems Matter Today", "date": "", "ddg_snippet": "For instance , Averroes.ai reduced weekly false rejects from 12,000 units to just 246 units, saving over $18 million annually per production line.", "subpage_snippet": "", "source": "www.unitxlabs.com", "link": "https://www.unitxlabs.com/resources/why-production-machine-vision-systems-matter/", "content": "For instance , Averroes.ai reduced weekly false rejects from 12,000 units to just 246 units, saving over $18 million annually per production line."}
+{"idx": 7, "title": "US11188791B2 - Anonymizing data for preserving privacy during", "date": "", "ddg_snippet": "2019-11-18 Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION reassignment INTERNATIONAL BUSINESS MACHINES CORPORATION ASSIGNMENT OF ASSIGNORS ...", "subpage_snippet": "", "source": "patents.google.com", "link": "https://patents.google.com/patent/US11188791B2/en", "content": "2019-11-18 Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION reassignment INTERNATIONAL BUSINESS MACHINES CORPORATION ASSIGNMENT OF ASSIGNORS ..."}
+{"idx": 8, "title": "Daniel Roy", "date": "", "ddg_snippet": "I co-organized the first workshop on probabilistic programming for statistics and machine learning at NIPS*2008 (with Vikash Mansinghka, John Winn ...", "subpage_snippet": "", "source": "danroy.org", "link": "http://danroy.org/", "content": "I co-organized the first workshop on probabilistic programming for statistics and machine learning at NIPS*2008 (with Vikash Mansinghka, John Winn ..."}
+{"idx": 9, "title": "Click Without Compromise: Online Advertising Measurement via", "date": "", "ddg_snippet": "... raised increasing privacy concerns regarding user consent, data security, and the potential misuse of personal information [ 48 , 57 , 27 , 49 ] .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2406.02463v3", "content": "... raised increasing privacy concerns regarding user consent, data security, and the potential misuse of personal information [ 48 , 57 , 27 , 49 ] ."}
diff --git a/data/sampled_jsons/Emerging_Properties_in_Self-Supervised_Vision_Transformers_Caron_ICCV_2021.jsonl b/data/sampled_jsons/Emerging_Properties_in_Self-Supervised_Vision_Transformers_Caron_ICCV_2021.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..a1563d3d6a98f01eb834b343166f1432f4fb25d2
--- /dev/null
+++ b/data/sampled_jsons/Emerging_Properties_in_Self-Supervised_Vision_Transformers_Caron_ICCV_2021.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "View a PDF of the paper titled Emerging Properties in Self - Supervised Vision Transformers , by Mathilde Caron and 6 other authors.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2104.14294", "content": "View a PDF of the paper titled Emerging Properties in Self - Supervised Vision Transformers , by Mathilde Caron and 6 other authors."}
+{"idx": 1, "title": "ICCV 2021 Open Access Repository", "date": "", "ddg_snippet": "Emerging Properties in Self - Supervised Vision Transformers . Mathilde Caron , Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, Armand Joulin; Proceedings of the IEEE/CVF International Conference on Computer Vision ( ICCV ), 2021 , pp. 9650-9660.", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/ICCV2021/html/Caron_Emerging_Properties_in_Self-Supervised_Vision_Transformers_ICCV_2021_paper", "content": "Emerging Properties in Self - Supervised Vision Transformers . Mathilde Caron , Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, Armand Joulin; Proceedings of the IEEE/CVF International Conference on Computer Vision ( ICCV ), 2021 , pp. 9650-9660."}
+{"idx": 2, "title": "[PDF] Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "@article{ Caron 2021 EmergingPI, title={ Emerging Properties in Self - Supervised Vision Transformers }, author={Mathilde Caron and Hugo Touvron and Ishan Misra and Herv'e J'egou and Julien Mairal and Piotr Bojanowski and Armand Joulin}, journal={ 2021 IEEE/CVF...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Emerging-Properties-in-Self-Supervised-Vision-Caron-Touvron/ad4a0938c48e61b7827869e4ac3baffd0aefab35", "content": "@article{ Caron 2021 EmergingPI, title={ Emerging Properties in Self - Supervised Vision Transformers }, author={Mathilde Caron and Hugo Touvron and Ishan Misra and Herv'e J'egou and Julien Mairal and Piotr Bojanowski and Armand Joulin}, journal={ 2021 IEEE/CVF..."}
+{"idx": 3, "title": "Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "Conference: 2021 IEEE/CVF International Conference on Computer Vision ( ICCV ). Authors... DINO [63] (short for Self-DIstillation with NO Labels) is a self - supervised learning framework that trains vision transformers (ViTs) [44] without requiring any labeled data.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/359005314_Emerging_Properties_in_Self-Supervised_Vision_Transformers", "content": "Conference: 2021 IEEE/CVF International Conference on Computer Vision ( ICCV ). Authors... DINO [63] (short for Self-DIstillation with NO Labels) is a self - supervised learning framework that trains vision transformers (ViTs) [44] without requiring any labeled data."}
+{"idx": 4, "title": "Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "In this paper, we question if self - supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets).", "subpage_snippet": "", "source": "deepai.org", "link": "https://deepai.org/publication/emerging-properties-in-self-supervised-vision-transformers", "content": "In this paper, we question if self - supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets)."}
+{"idx": 5, "title": "Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "Communication Dans Un Congrès Année : 2021 . Emerging Properties in Self - Supervised Vision Transformers .In this paper, we question if self - supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets).", "subpage_snippet": "", "source": "hal.science", "link": "https://hal.science/hal-03323359", "content": "Communication Dans Un Congrès Année : 2021 . Emerging Properties in Self - Supervised Vision Transformers .In this paper, we question if self - supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets)."}
+{"idx": 6, "title": "Mathilde Caron - Google Akademik", "date": "", "ddg_snippet": "Emerging properties in self - supervised vision transformers .Proceedings of the International Conference on Computer Vision ( ICCV ), 2021 .", "subpage_snippet": "", "source": "scholar.google.co.id", "link": "https://scholar.google.co.id/citations?user=eiB0s-kAAAAJ&hl=tr", "content": "Emerging properties in self - supervised vision transformers .Proceedings of the International Conference on Computer Vision ( ICCV ), 2021 ."}
+{"idx": 7, "title": "DINO: Self - Supervised Vision Transformers and Their Emerging ...", "date": "", "ddg_snippet": "The paper “ Emerging Properties in Self - Supervised Vision Transformers ” introduces DINO, a novel self - supervised learning approach that reveals unique properties of Vision Transformers and achieves state-of-the-art performance.", "subpage_snippet": "", "source": "medium.com", "link": "https://medium.com/@jimcanary/dino-self-supervised-vision-transformers-and-their-emerging-properties-7f9e5f4adac4", "content": "The paper “ Emerging Properties in Self - Supervised Vision Transformers ” introduces DINO, a novel self - supervised learning approach that reveals unique properties of Vision Transformers and achieves state-of-the-art performance."}
+{"idx": 8, "title": "DINO: Emerging Properties in Self - Supervised Vision Transformers", "date": "", "ddg_snippet": "This study questions if Self Supervised Learning provides new properties to Vision Transformers [1] that standout from convolutional networks and underlines the importance of momentum encoder[2], multi-crop training[3] and the use of small patches.", "subpage_snippet": "", "source": "wandb.ai", "link": "https://wandb.ai/self-supervised-learning/dino/reports/DINO-Emerging-Properties-in-Self-Supervised-Vision-Transformers--VmlldzoxMzM2MTAz", "content": "This study questions if Self Supervised Learning provides new properties to Vision Transformers [1] that standout from convolutional networks and underlines the importance of momentum encoder[2], multi-crop training[3] and the use of small patches."}
+{"idx": 9, "title": "DINO - Emerging properties in self - supervised vision transformers", "date": "", "ddg_snippet": "The emerging properties which have been identified to come up especially when using a vision transformer are: The attention features contain semantic segmentation like properties without being trained in this way (no labels, no supervision ).", "subpage_snippet": "", "source": "www.paepper.com", "link": "https://www.paepper.com/blog/posts/dino-emerging-properties-in-self-supervised-vision-transformers/", "content": "The emerging properties which have been identified to come up especially when using a vision transformer are: The attention features contain semantic segmentation like properties without being trained in this way (no labels, no supervision )."}
diff --git a/data/sampled_jsons/Emotion_Analysis_GPT-4_analyzes_each_rewritten_document_to_identify_the_top_M_emotions.jsonl b/data/sampled_jsons/Emotion_Analysis_GPT-4_analyzes_each_rewritten_document_to_identify_the_top_M_emotions.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..77d8de07c7f5e12cac150f7dd1b58627b5fdc9e0
--- /dev/null
+++ b/data/sampled_jsons/Emotion_Analysis_GPT-4_analyzes_each_rewritten_document_to_identify_the_top_M_emotions.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "(PDF) Integrating Emotional and Linguistic Models for Ethical...", "date": "", "ddg_snippet": "Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top . M .", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/380515639_Integrating_Emotional_and_Linguistic_Models_for_Ethical_Compliance_in_Large_Language_Models", "content": "Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top . M ."}
+{"idx": 1, "title": "Integrating Emotional and Linguistic Models for", "date": "", "ddg_snippet": "linguistic behaviors on the behavior spectrum Ψ. 2. Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top M emotions .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2405.07076", "content": "linguistic behaviors on the behavior spectrum Ψ. 2. Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top M emotions ."}
+{"idx": 2, "title": "integrating-emotional-and-linguistic-models-for-ethical- ...", "date": "", "ddg_snippet": "... Emotion Analysis: GPT-4 analyzes each rewritten document to identify the top M emotions . It then tallies the frequencies of these top emotions across all N ...", "subpage_snippet": "", "source": "scispace.com", "link": "https://scispace.com/pdf/integrating-emotional-and-linguistic-models-for-ethical-c8c5drc4bk.pdf", "content": "... Emotion Analysis: GPT-4 analyzes each rewritten document to identify the top M emotions . It then tallies the frequencies of these top emotions across all N ..."}
+{"idx": 3, "title": "The Influence of Text Variation on User Engagement in", "date": "", "ddg_snippet": "To answer these questions, we build a large-scale dataset of Reddit posts sharing YouTube videos to enable a comprehensive cross-platform, multimodal ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.03769v1", "content": "To answer these questions, we build a large-scale dataset of Reddit posts sharing YouTube videos to enable a comprehensive cross-platform, multimodal ..."}
+{"idx": 4, "title": "Top 15 ChatGPT Rewriter Tools to Improve Your Content Creation", "date": "", "ddg_snippet": "Leveraging the capabilities of the GPT (Generative Pre-trained Transformer) model developed by OpenAI, these tools offer a range of functionalities ...", "subpage_snippet": "", "source": "speedybrand.io", "link": "https://speedybrand.io/blogs/chatgpt-rewriter", "content": "Leveraging the capabilities of the GPT (Generative Pre-trained Transformer) model developed by OpenAI, these tools offer a range of functionalities ..."}
+{"idx": 5, "title": "source code analysis of Amazon Kiro", "date": "", "ddg_snippet": "It uses OpenVSX to attempt to work around the fractured ecosystem problem (see below) which means that developers who use programming languages such ...", "subpage_snippet": "", "source": "ghuntley.com", "link": "https://ghuntley.com/amazon-kiro-source-code/", "content": "It uses OpenVSX to attempt to work around the fractured ecosystem problem (see below) which means that developers who use programming languages such ..."}
+{"idx": 6, "title": "Top 13 Advanced RAG Techniques for Your Next Project", "date": "", "ddg_snippet": "The first step is to load the documents , then split or chunk it using various chunking techniques and then embed it using an embedding model so that ...", "subpage_snippet": "", "source": "www.analyticsvidhya.com", "link": "https://www.analyticsvidhya.com/blog/2025/04/advanced-rag-techniques/", "content": "The first step is to load the documents , then split or chunk it using various chunking techniques and then embed it using an embedding model so that ..."}
+{"idx": 7, "title": "Top Tech Tidbits for Thursday, November 23, 2023 - Volume 938", "date": "", "ddg_snippet": "I am excited to report that we have been working on a custom GPT for Top Tech Tidbits, using the new ChatGPT Plus Custom GPT feature , that ...", "subpage_snippet": "", "source": "www.toptechtidbits.com", "link": "https://www.toptechtidbits.com/tidbits2023/11232023/index.html", "content": "I am excited to report that we have been working on a custom GPT for Top Tech Tidbits, using the new ChatGPT Plus Custom GPT feature , that ..."}
+{"idx": 8, "title": "A Three-Branch Checks-and-Balances Framework", "date": "", "ddg_snippet": "2. Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top M emotions . It then tallies the frequencies of these top emotions across all N × L instances.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=o2afWIxjKD", "content": "2. Emotion Analysis : GPT - 4 analyzes each rewritten document to identify the top M emotions . It then tallies the frequencies of these top emotions across all N × L instances."}
+{"idx": 9, "title": "How Generative AI Works and How to Use LLMs Wisely — Halo Lab", "date": "", "ddg_snippet": "Each token has its own probability (weight) of use, which helps an LLM choose the most likely options. ... mechanism allows you to take into account ...", "subpage_snippet": "", "source": "www.halo-lab.com", "link": "https://www.halo-lab.com/blog/llms-how-they-work-and-how-to-use-them", "content": "Each token has its own probability (weight) of use, which helps an LLM choose the most likely options. ... mechanism allows you to take into account ..."}
diff --git a/data/sampled_jsons/Equation_(1)_FBox_score_box_embeddings_siteopenreview.net.jsonl b/data/sampled_jsons/Equation_(1)_FBox_score_box_embeddings_siteopenreview.net.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..31e42ff78b5e0d4984185f8dee7776ed8ba0c42e
--- /dev/null
+++ b/data/sampled_jsons/Equation_(1)_FBox_score_box_embeddings_siteopenreview.net.jsonl
@@ -0,0 +1,7 @@
+{"idx": 0, "title": "Representing Sentence Interpretations with Overlapping Box ...", "date": "", "ddg_snippet": "3.1.1 Box Embeddings Box embeddings represent items as n-dimensional hyperrectangles. A box embedding b is con-structed from two vectors: a center vector c and an offset vector o. For each ith dimension, the area of a box embedding is defined as the interval [ci − oi, ci + oi].", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=DEB8muQHYh", "content": "3.1.1 Box Embeddings Box embeddings represent items as n-dimensional hyperrectangles. A box embedding b is con-structed from two vectors: a center vector c and an offset vector o. For each ith dimension, the area of a box embedding is defined as the interval [ci − oi, ci + oi]."}
+{"idx": 1, "title": "Dual Box Embeddings for the Description Logic EL++ - OpenReview", "date": "", "ddg_snippet": "Figure 1 illustrates 2-dimensional Box2EL embeddings 326 that form a logical model of the TBox in Example 1, since, e.g., 327 Box (Father) ⊆ Box (Parent) ∩ Box (Male) and Equation (1 ) holds 328 for all relevant axioms.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=gTPTgixn3r", "content": "Figure 1 illustrates 2-dimensional Box2EL embeddings 326 that form a logical model of the TBox in Example 1, since, e.g., 327 Box (Father) ⊆ Box (Parent) ∩ Box (Male) and Equation (1 ) holds 328 for all relevant axioms."}
+{"idx": 2, "title": "A GEOMETRIC APPROACH TO PERSONALIZED RECOMMENDATION WITH SET ...", "date": "", "ddg_snippet": "193 Let, Box (m) = QD d=1[m⌞ d, m⌝ d] be the box embeddings for a movie m. At dimension d, the volume 194 of intersection between user u and movie m is defined as - 195 196 ⌝ d s completely contained by us d,u⌝d),(m⌞d,m⌝d)) = 1. This objective creates a set-theoretic interpretation with box embeddings , Vol((u⌞", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=0HWAbWgI3T", "content": "193 Let, Box (m) = QD d=1[m⌞ d, m⌝ d] be the box embeddings for a movie m. At dimension d, the volume 194 of intersection between user u and movie m is defined as - 195 196 ⌝ d s completely contained by us d,u⌝d),(m⌞d,m⌝d)) = 1. This objective creates a set-theoretic interpretation with box embeddings , Vol((u⌞"}
+{"idx": 3, "title": "S moothing the g eometry", "date": "", "ddg_snippet": "Table 5: F 1 scores of the box lattice, order embeddings , and our smoothed model, for different levels of label imbalance on the WordNet mammal subset.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=H1xSNiRcF7", "content": "Table 5: F 1 scores of the box lattice, order embeddings , and our smoothed model, for different levels of label imbalance on the WordNet mammal subset."}
+{"idx": 4, "title": "A Geometric Approach to Personalized Recommendation with...", "date": "", "ddg_snippet": "May 1 , 2025 · This paper presents a geometric approach to recommender systems using box embeddings where users, items, and attributes are all represented as boxes in the latent space to facilitate set-theoretic operations, such as negation and intersection, e.g., recommending a comedy but not romance. Overall the paper received positive-leaning scores where all reviewers praise the innovative usage of the ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=27tMzmzDjO", "content": "May 1 , 2025 · This paper presents a geometric approach to recommender systems using box embeddings where users, items, and attributes are all represented as boxes in the latent space to facilitate set-theoretic operations, such as negation and intersection, e.g., recommending a comedy but not romance. Overall the paper received positive-leaning scores where all reviewers praise the innovative usage of the ..."}
+{"idx": 5, "title": "Locality Sensitive Hashing in Fourier Frequency Domain For ...", "date": "", "ddg_snippet": "Box Embeddings The box embedding based volume score can be expressed as a shift invariant score a (q q x x). Here, the query and corpus items are expressed as boxes denoted by (z z q, Z Z q) and (z z x, Z Z x) respectively (the lower and upper corner coordinate vectors).", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=rUf0GV5CuU", "content": "Box Embeddings The box embedding based volume score can be expressed as a shift invariant score a (q q x x). Here, the query and corpus items are expressed as boxes denoted by (z z q, Z Z q) and (z z x, Z Z x) respectively (the lower and upper corner coordinate vectors)."}
+{"idx": 6, "title": "B : HIERARCHICAL CONCEPT REPRESENTATION O EMBEDDING OF BINARY ...", "date": "", "ddg_snippet": "s as HEC and probabilistic box embeddings with temperatures as T- Box . We report balanced accuracy, meaning we either ensure the positive and negative test sizes are equal (for reconstr ction) or re-weight the negative samples (for trans tive", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=zkE2js9qRe", "content": "s as HEC and probabilistic box embeddings with temperatures as T- Box . We report balanced accuracy, meaning we either ensure the positive and negative test sizes are equal (for reconstr ction) or re-weight the negative samples (for trans tive"}
diff --git a/data/sampled_jsons/Evaluating_Generative_AI_Systems_Is_a_Social_Science_Measurement_Challenge_Section_6_a_lot_of_work_c.jsonl b/data/sampled_jsons/Evaluating_Generative_AI_Systems_Is_a_Social_Science_Measurement_Challenge_Section_6_a_lot_of_work_c.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..570a280ef18f2cf71284593027d12f0af58b8bf3
--- /dev/null
+++ b/data/sampled_jsons/Evaluating_Generative_AI_Systems_Is_a_Social_Science_Measurement_Challenge_Section_6_a_lot_of_work_c.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Position: Evaluating Generative AI Systems Is a Social ...", "date": "", "ddg_snippet": "by H Wallach · 2025 · Cited by 12 — We present a four-level framework, grounded in measurement theory from the social sciences , for measuring concepts related to the capabilities, behaviors, and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2502.00561", "content": "by H Wallach · 2025 · Cited by 12 — We present a four-level framework, grounded in measurement theory from the social sciences , for measuring concepts related to the capabilities, behaviors, and ..."}
+{"idx": 1, "title": "A Validity-Centered Framework for AI Evaluation", "date": "", "ddg_snippet": "Position: Evaluating generative ai systems is a social science measurement challenge . arXiv preprint arXiv:2502.00561, 2025. Kuhn (1997) ↑ Thomas S Kuhn ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.10573v3", "content": "Position: Evaluating generative ai systems is a social science measurement challenge . arXiv preprint arXiv:2502.00561, 2025. Kuhn (1997) ↑ Thomas S Kuhn ..."}
+{"idx": 2, "title": "Methodological Challenges in Agentic Evaluations of AI ...", "date": "", "ddg_snippet": "by K Wei — Posi- tion: Evaluating Generative AI Systems is a Social Sci- ence Measurement Challenge , February 2025. URL https://arxiv.org/abs/2502.00561. Wang, A ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=ZhSKG8IslC", "content": "by K Wei — Posi- tion: Evaluating Generative AI Systems is a Social Sci- ence Measurement Challenge , February 2025. URL https://arxiv.org/abs/2502.00561. Wang, A ..."}
+{"idx": 3, "title": "Integrating Generative Artificial Intelligence into Social ...", "date": "", "ddg_snippet": "7 May 2025 — This essay introduces a special issue that examines how these and other affordances of generative AI can advance social science research.", "subpage_snippet": "", "source": "journals.sagepub.com", "link": "https://journals.sagepub.com/doi/10.1177/00491241251339184", "content": "7 May 2025 — This essay introduces a special issue that examines how these and other affordances of generative AI can advance social science research."}
+{"idx": 4, "title": "Generative AI in Computer Science Education", "date": "", "ddg_snippet": "by D Franklin · 2025 · Cited by 6 — Generative AI is a disruptive technology that has the potential to transform many aspects of how computer science is taught.", "subpage_snippet": "", "source": "www.cambridge.org", "link": "https://www.cambridge.org/core/elements/generative-ai-in-computer-science-education/0A22106CBD7FCB391FD120C56E21420F", "content": "by D Franklin · 2025 · Cited by 6 — Generative AI is a disruptive technology that has the potential to transform many aspects of how computer science is taught."}
+{"idx": 5, "title": "Challenges in evaluating AI systems", "date": "", "ddg_snippet": "4 Oct 2023 — We want readers of this post to have two main takeaways: robust evaluations are extremely difficult to develop and implement , and effective AI governance ...", "subpage_snippet": "", "source": "www.anthropic.com", "link": "https://www.anthropic.com/research/evaluating-ai-systems", "content": "4 Oct 2023 — We want readers of this post to have two main takeaways: robust evaluations are extremely difficult to develop and implement , and effective AI governance ..."}
+{"idx": 6, "title": "Impacts of generative artificial intelligence on the future ...", "date": "", "ddg_snippet": "by N Salari · 2025 · Cited by 9 — This systematic review is conducted to examine the impacts of GenAI on the future of employment, focusing on concerns about rising unemployment.", "subpage_snippet": "", "source": "www.sciencedirect.com", "link": "https://www.sciencedirect.com/science/article/pii/S2451958825000673", "content": "by N Salari · 2025 · Cited by 9 — This systematic review is conducted to examine the impacts of GenAI on the future of employment, focusing on concerns about rising unemployment."}
+{"idx": 7, "title": "AI Governance Needs Sociotechnical Expertise", "date": "", "ddg_snippet": "DeepMind researchers have proposed a sociotechnical safety evaluation of generative AI systems by incorporating two additional layers to safety evaluation ...", "subpage_snippet": "", "source": "datasociety.net", "link": "https://datasociety.net/wp-content/uploads/2024/05/DS_AI_Governance_Policy_Brief.pdf", "content": "DeepMind researchers have proposed a sociotechnical safety evaluation of generative AI systems by incorporating two additional layers to safety evaluation ..."}
+{"idx": 8, "title": "ChatGPT and Beyond: Exploring the Responsible Use of ...", "date": "", "ddg_snippet": "by M Söllner · 2025 · Cited by 15 — ChatGPT and Beyond: Exploring the Responsible Use of Generative AI in the Workplace. An Interdisciplinary Perspective. Download PDF.", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/article/10.1007/s12599-025-00932-8", "content": "by M Söllner · 2025 · Cited by 15 — ChatGPT and Beyond: Exploring the Responsible Use of Generative AI in the Workplace. An Interdisciplinary Perspective. Download PDF."}
+{"idx": 9, "title": "Unpacking the Gap in Human-Centered Evaluations of AI ...", "date": "", "ddg_snippet": "by A Khullar · 2025 — We argue for assessing broader achievements enabled through AI's use when conducting human-centered evaluations of AI .", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/full/10.1145/3706598.3713278", "content": "by A Khullar · 2025 — We argue for assessing broader achievements enabled through AI's use when conducting human-centered evaluations of AI ."}
diff --git a/data/sampled_jsons/FlowDec_is_not_streaming-capable_due_to_the_noncausal_architecture_of_the_used_DNNs.jsonl b/data/sampled_jsons/FlowDec_is_not_streaming-capable_due_to_the_noncausal_architecture_of_the_used_DNNs.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b0751762efb4a220eb2cc22d9e1d33bc724bd643
--- /dev/null
+++ b/data/sampled_jsons/FlowDec_is_not_streaming-capable_due_to_the_noncausal_architecture_of_the_used_DNNs.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Results obtained for the Voicebank-Demand. Values indicate mean and...", "date": "", "ddg_snippet": "... While FlowDec , like DAC, is currently not streaming - capable due to the noncausal architecture of the used DNNs , our postfilter approach can be modified for a causal DNN as in (Richter et al., 2024a), which would pave the way for real-time communication and audio streaming...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/figure/Results-obtained-for-the-Voicebank-Demand-Values-indicate-mean-and-standard-deviation_tbl2_379089990", "content": "... While FlowDec , like DAC, is currently not streaming - capable due to the noncausal architecture of the used DNNs , our postfilter approach can be modified for a causal DNN as in (Richter et al., 2024a), which would pave the way for real-time communication and audio streaming..."}
+{"idx": 1, "title": "FlowDec : A flow -based full-band general audio codec with high...", "date": "", "ddg_snippet": "While FlowDec , like DAC, is currently not streaming - capable due to the noncausal architecture of the used DNNs , our postfilter approach can be modified for a causal DNN as in (Richter et al., 2024a) , which would pave the way for real-time communication and audio streaming applications.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.01485v1", "content": "While FlowDec , like DAC, is currently not streaming - capable due to the noncausal architecture of the used DNNs , our postfilter approach can be modified for a causal DNN as in (Richter et al., 2024a) , which would pave the way for real-time communication and audio streaming applications."}
+{"idx": 2, "title": "Ethereal Prism Rod | Fisch Wiki | Fandom", "date": "", "ddg_snippet": "The Ethereal Prism Rod is an extremely powerful, money-making, all-rounder stat late-game rod costing 15,000,000 C$ , due to its mutation passive, having a 50% chance to apply the 8x sell value Prismize mutation .", "subpage_snippet": "", "source": "fisch.fandom.com", "link": "https://fisch.fandom.com/wiki/Ethereal_Prism_Rod", "content": "The Ethereal Prism Rod is an extremely powerful, money-making, all-rounder stat late-game rod costing 15,000,000 C$ , due to its mutation passive, having a 50% chance to apply the 8x sell value Prismize mutation ."}
+{"idx": 3, "title": "AI Architecture Generator – Realize Your Vision with Leonardo.Ai", "date": "", "ddg_snippet": "Enhance your architectural design process with Leonardo.Ai's AI architecture generator tools. Create stunning mockups, concept designs, and dynamic videos to streamline your workflow.", "subpage_snippet": "", "source": "leonardo.ai", "link": "https://leonardo.ai/ai-architecture-generator/", "content": "Enhance your architectural design process with Leonardo.Ai's AI architecture generator tools. Create stunning mockups, concept designs, and dynamic videos to streamline your workflow."}
+{"idx": 4, "title": "Humanize AI Text Free- Get 100% Human Score", "date": "", "ddg_snippet": "Students can use AI Humanizer to refine essays and homework, improving the flow and readability while adhering to academic standards. It’s the perfect tool for boosting grades and enhancing writing skills.", "subpage_snippet": "", "source": "decopy.ai", "link": "https://decopy.ai/ai-humanizer/", "content": "Students can use AI Humanizer to refine essays and homework, improving the flow and readability while adhering to academic standards. It’s the perfect tool for boosting grades and enhancing writing skills."}
+{"idx": 5, "title": "ChatPlanet - Random Video Chat Roulette", "date": "", "ddg_snippet": "Here, you can meet interesting people from around the globe through random video chats. Whether you're looking to make new friends, share experiences, or simply enjoy some light-hearted banter, ChatPlanet is the perfect place for you. Just click to start, and let the conversations flow !", "subpage_snippet": "", "source": "videochatplanet.com", "link": "https://videochatplanet.com/", "content": "Here, you can meet interesting people from around the globe through random video chats. Whether you're looking to make new friends, share experiences, or simply enjoy some light-hearted banter, ChatPlanet is the perfect place for you. Just click to start, and let the conversations flow !"}
+{"idx": 6, "title": "Can AI Actually Create a Python Trading Bot Specifically for...", "date": "", "ddg_snippet": "The potential lies in using AI models to create custom strategies, manage data, and even automate certain aspects of trading, though the actual level of automation achievable directly through Python code on Thinkorswim is constrained by the platform’s architecture .", "subpage_snippet": "", "source": "trading-strategies.academy", "link": "https://trading-strategies.academy/archives/1645", "content": "The potential lies in using AI models to create custom strategies, manage data, and even automate certain aspects of trading, though the actual level of automation achievable directly through Python code on Thinkorswim is constrained by the platform’s architecture ."}
+{"idx": 7, "title": "“I Want To Enjoy Every Minute That I Can”: For... | British Vogue", "date": "", "ddg_snippet": "They are capable of tracking your browser across other sites and building up a profile of your interests. This may impact the content and messages you see on other websites you visit.You can refuse the use of this cookie by switching off the slider to the right.", "subpage_snippet": "", "source": "www.vogue.co.uk", "link": "https://www.vogue.co.uk/article/lando-norris-interview", "content": "They are capable of tracking your browser across other sites and building up a profile of your interests. This may impact the content and messages you see on other websites you visit.You can refuse the use of this cookie by switching off the slider to the right."}
+{"idx": 8, "title": "Historical Consciousness: What Germany Could Learn from Russia", "date": "", "ddg_snippet": "I use the term “historical consciousness” because I consider it the best antidote to the repetition of catastrophes.“Despite this betrayal of the Soviet Union, Russia still uses diplomatic and friendly words toward the US, which is actually astonishing.”", "subpage_snippet": "", "source": "sonar21.com", "link": "https://sonar21.com/historical-consciousness-what-germany-could-learn-from-russia/", "content": "I use the term “historical consciousness” because I consider it the best antidote to the repetition of catastrophes.“Despite this betrayal of the Soviet Union, Russia still uses diplomatic and friendly words toward the US, which is actually astonishing.”"}
+{"idx": 9, "title": "International Departures", "date": "", "ddg_snippet": "Fraport TAV Antalya Airport Operator cannot be held responsible for any damages that may arise from the use of this information, inaccuracies or deficiencies. Please refer to the airline company for more detailed information on flights.", "subpage_snippet": "", "source": "www.antalya-airport.aero", "link": "https://www.antalya-airport.aero/passengers-visitors/flight-info/international-departures", "content": "Fraport TAV Antalya Airport Operator cannot be held responsible for any damages that may arise from the use of this information, inaccuracies or deficiencies. Please refer to the airline company for more detailed information on flights."}
diff --git a/data/sampled_jsons/FourCastNet_A_Global_Data-driven_High-resolution_Weather_Model_using_Adaptive_Fourier_Neural_Operato.jsonl b/data/sampled_jsons/FourCastNet_A_Global_Data-driven_High-resolution_Weather_Model_using_Adaptive_Fourier_Neural_Operato.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c782f616e652a151e99994d686ba053e16512164
--- /dev/null
+++ b/data/sampled_jsons/FourCastNet_A_Global_Data-driven_High-resolution_Weather_Model_using_Adaptive_Fourier_Neural_Operato.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using ...", "date": "", "ddg_snippet": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution . FourCastNet accurately forecasts high-resolution , fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2202.11214", "content": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution . FourCastNet accurately forecasts high-resolution , fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for ..."}
+{"idx": 1, "title": "FourCastNet: Accelerating Global High-Resolution Weather Forecasting ...", "date": "", "ddg_snippet": "We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range forecasts five orders-of-magnitude faster than NWP while approaching state-of-the-art accuracy.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/10.1145/3592979.3593412", "content": "We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range forecasts five orders-of-magnitude faster than NWP while approaching state-of-the-art accuracy."}
+{"idx": 2, "title": "Global Data-driven High-resolution Weather Model - GitHub", "date": "", "ddg_snippet": "@article {pathak2022fourcastnet, title = { Fourcastnet : A global data-driven high-resolution weather model using adaptive fourier neural operators }, author = {Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and others}, journal = { arXiv ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/yrqUni/FourCastNet", "content": "@article {pathak2022fourcastnet, title = { Fourcastnet : A global data-driven high-resolution weather model using adaptive fourier neural operators }, author = {Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and others}, journal = { arXiv ..."}
+{"idx": 3, "title": "FourCastNet: A Data-driven Model for High-resolution Weather ... - ICLR", "date": "", "ddg_snippet": "Poster in Workshop: AI for Earth and Space Science FourCastNet : A Data-driven Model for High-resolution Weather Forecasts using Adaptive Fourier Neural Operators", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2022/7656", "content": "Poster in Workshop: AI for Earth and Space Science FourCastNet : A Data-driven Model for High-resolution Weather Forecasts using Adaptive Fourier Neural Operators"}
+{"idx": 4, "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using ...", "date": "", "ddg_snippet": "How data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models is discussed. FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution . FourCastNet accurately forecasts high ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/FourCastNet:-A-Global-Data-driven-High-resolution-Pathak-Subramanian/10194e9d1d6b8ca8870445c990d4933c1dac1125", "content": "How data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models is discussed. FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution . FourCastNet accurately forecasts high ..."}
+{"idx": 5, "title": "FourCastNet: Accelerating Global High-Resolution Weather Forecasting ...", "date": "", "ddg_snippet": "Abstract: Extreme weather amplified by climate change is causing increasingly devastating impacts across the globe. The current use of physics-based numerical weather prediction (NWP) limits accuracy and resolution due to high computational cost and strict time-to-solution limits. We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=Hnd0NDzUWL", "content": "Abstract: Extreme weather amplified by climate change is causing increasingly devastating impacts across the globe. The current use of physics-based numerical weather prediction (NWP) limits accuracy and resolution due to high computational cost and strict time-to-solution limits. We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and ..."}
+{"idx": 6, "title": "Skilful global seasonal predictions from a machine learning weather ...", "date": "", "ddg_snippet": "Kurth, T. et al. Fourcastnet : Accelerating global high-resolution weather forecasting using adaptive fourier neural operators . In Proceedings of the platform for advanced scientific computing ...", "subpage_snippet": "", "source": "www.nature.com", "link": "https://www.nature.com/articles/s41612-025-01198-3", "content": "Kurth, T. et al. Fourcastnet : Accelerating global high-resolution weather forecasting using adaptive fourier neural operators . In Proceedings of the platform for advanced scientific computing ..."}
+{"idx": 7, "title": "Innovative Short-Term Weather Forecasting System Combining Data-Driven ...", "date": "", "ddg_snippet": "Abstract Regional short-term high-resolution meteorological prediction systems are critical for applications requiring precise and frequent updates, such as renewable energy forecasting. However, existing forecasting methods are constrained by limitations such as coarse resolution , high computational demand, and a narrow range of predictive variables. This study introduces an innovative ...", "subpage_snippet": "", "source": "journals.ametsoc.org", "link": "https://journals.ametsoc.org/view/journals/aies/4/3/AIES-D-24-0125.1.xml", "content": "Abstract Regional short-term high-resolution meteorological prediction systems are critical for applications requiring precise and frequent updates, such as renewable energy forecasting. However, existing forecasting methods are constrained by limitations such as coarse resolution , high computational demand, and a narrow range of predictive variables. This study introduces an innovative ..."}
+{"idx": 8, "title": "[2208.05419] FourCastNet: Accelerating Global High-Resolution Weather ...", "date": "", "ddg_snippet": "Extreme weather amplified by climate change is causing increasingly devastating impacts across the globe. The current use of physics-based numerical weather prediction (NWP) limits accuracy due to high computational cost and strict time-to-solution limits. We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2208.05419", "content": "Extreme weather amplified by climate change is causing increasingly devastating impacts across the globe. The current use of physics-based numerical weather prediction (NWP) limits accuracy due to high computational cost and strict time-to-solution limits. We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range ..."}
+{"idx": 9, "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using ...", "date": "", "ddg_snippet": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at 0.25∘ resolution .", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/362153109_FourCastNet_A_Global_Data-driven_High-resolution_Weather_Model_using_Adaptive_Fourier_Neural_Operators", "content": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at 0.25∘ resolution ."}
diff --git a/data/sampled_jsons/FourCastNet_Pathak_2022_evaluation_metrics_negative_log-likelihood_year_2022.jsonl b/data/sampled_jsons/FourCastNet_Pathak_2022_evaluation_metrics_negative_log-likelihood_year_2022.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..dc65a761c33cbad659f36102f79198a40c56fc9d
--- /dev/null
+++ b/data/sampled_jsons/FourCastNet_Pathak_2022_evaluation_metrics_negative_log-likelihood_year_2022.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using ...", "date": "", "ddg_snippet": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2202.11214", "content": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for ..."}
+{"idx": 1, "title": "GitHub - NVlabs/FourCastNet: Initial public release of code, data, and ...", "date": "", "ddg_snippet": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at 0.25∘ resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/NVlabs/FourCastNet", "content": "FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at 0.25∘ resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor."}
+{"idx": 2, "title": "FourCastNet: A Data-driven Model for High-resolution Weather ... - ICLR", "date": "", "ddg_snippet": "FourCastNet : A Data-driven Model for High-resolution Weather Forecasts using Adaptive Fourier Neural Operators Jaideep Pathak · Shashank Subramanian · Peter Harrington · Sanjeev Raja · Ashesh Chattopadhyay · Morteza Mardani · Thorsten Kurth · David M. Hall · Zongyi Li · Kamyar Azizzadenesheli · Pedram Hassanzadeh · Karthik Kashinath ...", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2022/7656", "content": "FourCastNet : A Data-driven Model for High-resolution Weather Forecasts using Adaptive Fourier Neural Operators Jaideep Pathak · Shashank Subramanian · Peter Harrington · Sanjeev Raja · Ashesh Chattopadhyay · Morteza Mardani · Thorsten Kurth · David M. Hall · Zongyi Li · Kamyar Azizzadenesheli · Pedram Hassanzadeh · Karthik Kashinath ..."}
+{"idx": 3, "title": "FourCastNet - Janelia CVML", "date": "", "ddg_snippet": "The FourCastNet paper [ Pathak , 2022 ] trains an Adaptive Fourier Neural Operator (AFNO) network [Guibas, 2022 ] to predict a collection of atmospheric variables at the next time step (6 hours into the future) given the current readout for those variables.", "subpage_snippet": "", "source": "janelia-cvml.github.io", "link": "https://janelia-cvml.github.io/blog/posts/FourCastNet/index.html", "content": "The FourCastNet paper [ Pathak , 2022 ] trains an Adaptive Fourier Neural Operator (AFNO) network [Guibas, 2022 ] to predict a collection of atmospheric variables at the next time step (6 hours into the future) given the current readout for those variables."}
+{"idx": 4, "title": "FourCastNet: A practical introduction to a state-of-the-art deep ...", "date": "", "ddg_snippet": "However, recent advances in deep learning, specifically the FourCastNet model, have shown that data-driven approaches can forecast important atmospheric variables with excellent skill and comparable accuracy to standard numerical methods, but at orders-of-magnitude lower computational and energy cost during inference, enabling larger ensembles ...", "subpage_snippet": "", "source": "www.climatechange.ai", "link": "https://www.climatechange.ai/papers/neurips2022/115", "content": "However, recent advances in deep learning, specifically the FourCastNet model, have shown that data-driven approaches can forecast important atmospheric variables with excellent skill and comparable accuracy to standard numerical methods, but at orders-of-magnitude lower computational and energy cost during inference, enabling larger ensembles ..."}
+{"idx": 5, "title": "[2507.12144] FourCastNet 3: A geometric approach to probabilistic ...", "date": "", "ddg_snippet": "FourCastNet 3 advances global weather modeling by implementing a scalable, geometric machine learning (ML) approach to probabilistic ensemble forecasting. The approach is designed to respect spherical geometry and to accurately model the spatially correlated probabilistic nature of the problem, resulting in stable spectra and realistic dynamics across multiple scales. FourCastNet 3 delivers ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2507.12144", "content": "FourCastNet 3 advances global weather modeling by implementing a scalable, geometric machine learning (ML) approach to probabilistic ensemble forecasting. The approach is designed to respect spherical geometry and to accurately model the spatially correlated probabilistic nature of the problem, resulting in stable spectra and realistic dynamics across multiple scales. FourCastNet 3 delivers ..."}
+{"idx": 6, "title": "FourCastNet: Accelerating Global High-Resolution Weather Forecasting ...", "date": "", "ddg_snippet": "We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range forecasts five orders-of-magnitude faster than NWP while approaching state-of-the-art accuracy.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/abs/10.1145/3592979.3593412", "content": "We report that a data-driven deep learning Earth system emulator, FourCastNet , can predict global weather and generate medium-range forecasts five orders-of-magnitude faster than NWP while approaching state-of-the-art accuracy."}
+{"idx": 7, "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using ...", "date": "", "ddg_snippet": "How data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models is discussed. FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution. FourCastNet accurately forecasts high ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/FourCastNet:-A-Global-Data-driven-High-resolution-Pathak-Subramanian/10194e9d1d6b8ca8870445c990d4933c1dac1125", "content": "How data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models is discussed. FourCastNet , short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\\\circ}$ resolution. FourCastNet accurately forecasts high ..."}
+{"idx": 8, "title": "FourCastNet 3: A geometric approach to probabilistic ...", "date": "", "ddg_snippet": "16 Jul 2025 — Appendix D Evaluation metrics . Report issue for preceding element. We define evaluation metrics to measure the performance of trained models.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.12144v1", "content": "16 Jul 2025 — Appendix D Evaluation metrics . Report issue for preceding element. We define evaluation metrics to measure the performance of trained models."}
+{"idx": 9, "title": "the logarithm trick: achieve better long term - forecast via ...", "date": "", "ddg_snippet": "by T Zhang — In this paper, we present a simpler solution: by merely integrating an element-wise logarithmic operation after the standard Mean Square Error(MSE) loss - a ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=Y29rdPpPu4", "content": "by T Zhang — In this paper, we present a simpler solution: by merely integrating an element-wise logarithmic operation after the standard Mean Square Error(MSE) loss - a ..."}
diff --git a/data/sampled_jsons/GTA_Greedy_Task_Assignment_distributed_machine_learning_performance_worker_time_inefficiency.jsonl b/data/sampled_jsons/GTA_Greedy_Task_Assignment_distributed_machine_learning_performance_worker_time_inefficiency.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..1fb982c5e4a39ad31fe1fed114b8b587c3743200
--- /dev/null
+++ b/data/sampled_jsons/GTA_Greedy_Task_Assignment_distributed_machine_learning_performance_worker_time_inefficiency.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "In this paper, we propose ATA (Adaptive Task Allocation), a method that adapts to heterogeneous and random distributions of worker computation times .", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46650", "content": "In this paper, we propose ATA (Adaptive Task Allocation), a method that adapts to heterogeneous and random distributions of worker computation times ."}
+{"idx": 1, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "18 Jun 2025 — In distributed machine learning , traditional greedy methods waste resources by assigning tasks to every worker , even when only a few results are ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=1BaC3AdG1i&noteId=hkH8Wi9zZm", "content": "18 Jun 2025 — In distributed machine learning , traditional greedy methods waste resources by assigning tasks to every worker , even when only a few results are ..."}
+{"idx": 2, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "2 Feb 2025 — However, their greedy approach can lead to inefficiencies using more computation than required, especially when computation times vary across ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.00775v1", "content": "2 Feb 2025 — However, their greedy approach can lead to inefficiencies using more computation than required, especially when computation times vary across ..."}
+{"idx": 3, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "Current distributed machine learning systems typically employ Greedy Task Allocation ( GTA ) strategies, where all available workers are assigned tasks to ...", "subpage_snippet": "", "source": "www.alphaxiv.org", "link": "https://www.alphaxiv.org/overview/2502.00775v2", "content": "Current distributed machine learning systems typically employ Greedy Task Allocation ( GTA ) strategies, where all available workers are assigned tasks to ..."}
+{"idx": 4, "title": "Task allocation for maximum cooperation in complex ...", "date": "", "ddg_snippet": "by J Wang · 2024 · Cited by 1 — Based on RULE 1 and RULE 2, we design a Greedy Task Allocation ( GTA ) algorithm as shown in Algorithm 1. At first, we obtain critical actors ...", "subpage_snippet": "", "source": "www.sciencedirect.com", "link": "https://www.sciencedirect.com/science/article/pii/S0950705124006233", "content": "by J Wang · 2024 · Cited by 1 — Based on RULE 1 and RULE 2, we design a Greedy Task Allocation ( GTA ) algorithm as shown in Algorithm 1. At first, we obtain critical actors ..."}
+{"idx": 5, "title": "Consensus-Based Group Task Assignment with Social ...", "date": "", "ddg_snippet": "by X Li · 2020 · Cited by 45 — In this paper, we propose a novel framework for group task assignment based on worker groups' preferences, which includes two components: social impact-based ...", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/article/10.1007/s41019-020-00142-0", "content": "by X Li · 2020 · Cited by 45 — In this paper, we propose a novel framework for group task assignment based on worker groups' preferences, which includes two components: social impact-based ..."}
+{"idx": 6, "title": "Multiattribute E-CARGO Task Assignment Model Based on ...", "date": "", "ddg_snippet": "by Z Liu · Cited by 8 — The CTO requires A to organize different groups to complete these tasks and achieve the most effective performance in the shortest time . Obviously, GTA is used.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=dhQ3OKfzRu", "content": "by Z Liu · Cited by 8 — The CTO requires A to organize different groups to complete these tasks and achieve the most effective performance in the shortest time . Obviously, GTA is used."}
+{"idx": 7, "title": "Coalition-based task assignment with priority-aware ...", "date": "", "ddg_snippet": "by Y Zhao · 2024 · Cited by 23 — The greedy approach forms a set of worker coalitions greedily for performing tasks and uses an acceptance probability to identify high-value ...", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/article/10.1007/s00778-023-00802-3", "content": "by Y Zhao · 2024 · Cited by 23 — The greedy approach forms a set of worker coalitions greedily for performing tasks and uses an acceptance probability to identify high-value ..."}
+{"idx": 8, "title": "Profit Optimization in Spatial Crowdsourcing - Kai Zheng", "date": "", "ddg_snippet": "To achieve computational efficiency, we propose a Greedy Task Assignment (GTA) algorithm that gives priority to tasks with the higher reward per unit of ... 16 pages", "subpage_snippet": "", "source": "zheng-kai.com", "link": "https://zheng-kai.com/paper/tkde_2022_zhao.pdf", "content": "To achieve computational efficiency, we propose a Greedy Task Assignment (GTA) algorithm that gives priority to tasks with the higher reward per unit of ... 16 pages"}
+{"idx": 9, "title": "Generating Adversarial Examples for Discrete Data", "date": "", "ddg_snippet": "by P Yang · 2020 · Cited by 138 — Greedy Attack achieves higher success rate , while Gumbel Attack requires fewer model evaluations, leading to better efficiency in real-time or large-scale ... 36 pages", "subpage_snippet": "", "source": "jmlr.org", "link": "https://jmlr.org/papers/volume21/19-569/19-569.pdf", "content": "by P Yang · 2020 · Cited by 138 — Greedy Attack achieves higher success rate , while Gumbel Attack requires fewer model evaluations, leading to better efficiency in real-time or large-scale ... 36 pages"}
diff --git a/data/sampled_jsons/GUI_automation_action_planning_execution_LLM_agent_2024.jsonl b/data/sampled_jsons/GUI_automation_action_planning_execution_LLM_agent_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..f4877f1f8cbbd117657b49f512fd56c2cc937c45
--- /dev/null
+++ b/data/sampled_jsons/GUI_automation_action_planning_execution_LLM_agent_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "You Only Look at Screens: Multimodal Chain-of-Action Agents", "date": "", "ddg_snippet": "Building intelligent autonomous agents that are capable of task planning , decision making, and action execution in a particular environment is a long ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2309.11436v4", "content": "Building intelligent autonomous agents that are capable of task planning , decision making, and action execution in a particular environment is a long ..."}
+{"idx": 1, "title": "Large Language Model-Brained GUI Agents: A Survey", "date": "", "ddg_snippet": "... LLM -brained” GUI agents capable of interpreting complex GUI elements and autonomously executing actions based on natural language instructions.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.18279v12", "content": "... LLM -brained” GUI agents capable of interpreting complex GUI elements and autonomously executing actions based on natural language instructions."}
+{"idx": 2, "title": "Mobile-Agent-v3: Foundamental Agents for GUI Automation", "date": "", "ddg_snippet": "... post-trained on large-scale, diverse GUI interaction data, GUI -Owl unifies perception, grounding, reasoning, planning , and action execution within a ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.15144v1", "content": "... post-trained on large-scale, diverse GUI interaction data, GUI -Owl unifies perception, grounding, reasoning, planning , and action execution within a ..."}
+{"idx": 3, "title": "Echoes of Understanding: Enhancing LLM-based Agents with", "date": "", "ddg_snippet": "In these roles, LLM -based agents [ 4 ; 5 ] harness the reasoning and planning capabilities of LLMs to automate complex tasks across diverse domains ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.16024v1", "content": "In these roles, LLM -based agents [ 4 ; 5 ] harness the reasoning and planning capabilities of LLMs to automate complex tasks across diverse domains ..."}
+{"idx": 4, "title": "Evaluation and Benchmarking of LLM Agents: A Survey", "date": "", "ddg_snippet": "This taxonomy serves both as a conceptual framework and a practical guide, enabling systematic comparison and analysis of LLM agents across a wide ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.21504v1", "content": "This taxonomy serves both as a conceptual framework and a practical guide, enabling systematic comparison and analysis of LLM agents across a wide ..."}
+{"idx": 5, "title": "Bridging Reasoning and Action: The Synergy of Large Concept", "date": "", "ddg_snippet": "Agentic graph systems require sophisticated reasoning, planning , and action - execution capabilities to function effectively.", "subpage_snippet": "", "source": "techaireports.com", "link": "https://techaireports.com/bridging-reasoning-and-action-the-synergy-of-large-concept-models-lcms-and-large-action-models-lams-in-agentic-systems/", "content": "Agentic graph systems require sophisticated reasoning, planning , and action - execution capabilities to function effectively."}
+{"idx": 6, "title": "Xinbei Ma - ACL Anthology", "date": "", "ddg_snippet": "Existing LLM agents mainly generate natural language plans to guide reasoning, which is verbose and inefficient. ... Planning of Thoughts (D-PoT) for ...", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/people/x/xinbei-ma/", "content": "Existing LLM agents mainly generate natural language plans to guide reasoning, which is verbose and inefficient. ... Planning of Thoughts (D-PoT) for ..."}
+{"idx": 7, "title": "o1 · GitHub Topics · GitHub", "date": "", "ddg_snippet": "LLM Agent Framework in ComfyUI includes MCP ... Official implementation of GUI -R1 : A Generalist R1-Style Vision-Language Action Model For GUI Agents", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/topics/o1", "content": "LLM Agent Framework in ComfyUI includes MCP ... Official implementation of GUI -R1 : A Generalist R1-Style Vision-Language Action Model For GUI Agents"}
+{"idx": 8, "title": "Yian Wang", "date": "", "ddg_snippet": "To facilitate future research in this area, we developed OS-Atlas - a foundational GUI action model that excels at GUI grounding and OOD agentic ...", "subpage_snippet": "", "source": "www.catalyzex.com", "link": "https://www.catalyzex.com/author/Yian+Wang", "content": "To facilitate future research in this area, we developed OS-Atlas - a foundational GUI action model that excels at GUI grounding and OOD agentic ..."}
+{"idx": 9, "title": "browser-use vs comic-mono-font - compare differences and", "date": "", "ddg_snippet": "... I'm excited to share talk2browser, which leverages LangGraph's agent orchestration capabilities to create a self-improving browser automation system ...", "subpage_snippet": "", "source": "www.libhunt.com", "link": "https://www.libhunt.com/compare-browser-use-vs-comic-mono-font", "content": "... I'm excited to share talk2browser, which leverages LangGraph's agent orchestration capabilities to create a self-improving browser automation system ..."}
diff --git a/data/sampled_jsons/Herman_Chau_algebraic_combinatorics_dataset.jsonl b/data/sampled_jsons/Herman_Chau_algebraic_combinatorics_dataset.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c3a37dfec56cc6e6cc393c17fb492bcc453a281c
--- /dev/null
+++ b/data/sampled_jsons/Herman_Chau_algebraic_combinatorics_dataset.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2503.06366] Machine Learning meets Algebraic Combinatorics: A", "date": "", "ddg_snippet": "... Machine Learning meets Algebraic Combinatorics : A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics, by Herman Chau ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2503.06366", "content": "... Machine Learning meets Algebraic Combinatorics : A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics, by Herman Chau ..."}
+{"idx": 1, "title": "AlCoVE: an Algebraic Combinatorics Virtual Expedition", "date": "", "ddg_snippet": "AlCoVE brings together researchers interested in algebraic combinatorics from around the world. ... algebra, prove its eigenvalues are polynomials in ...", "subpage_snippet": "", "source": "www.math.uwaterloo.ca", "link": "https://www.math.uwaterloo.ca/~opecheni/alcove2024.htm", "content": "AlCoVE brings together researchers interested in algebraic combinatorics from around the world. ... algebra, prove its eigenvalues are polynomials in ..."}
+{"idx": 2, "title": "Read Herman by Jim Unger - GoComics", "date": "", "ddg_snippet": "3 days ago · Discover Herman by Jim Unger, the outrageous single-panel comic known for its distinctive humor and quirky characters. Classic strips and new adventures continue this beloved cartoon's legacy.", "subpage_snippet": "", "source": "www.gocomics.com", "link": "https://www.gocomics.com/herman", "content": "3 days ago · Discover Herman by Jim Unger, the outrageous single-panel comic known for its distinctive humor and quirky characters. Classic strips and new adventures continue this beloved cartoon's legacy."}
+{"idx": 3, "title": "Herman by Jim Unger for August 23, 2025 | GoComics", "date": "", "ddg_snippet": "Aug 23, 2025 · Read Herman —a comic strip by creator Jim Unger—for today, August 23, 2025, and check out other great comics, too!", "subpage_snippet": "", "source": "www.gocomics.com", "link": "https://www.gocomics.com/herman/2025/08/23", "content": "Aug 23, 2025 · Read Herman —a comic strip by creator Jim Unger—for today, August 23, 2025, and check out other great comics, too!"}
+{"idx": 4, "title": "Read about Herman and Jim Unger | GoComics", "date": "", "ddg_snippet": "Learn more about the Herman comic strip, its cast of characters, and creator Jim Unger.", "subpage_snippet": "", "source": "www.gocomics.com", "link": "https://www.gocomics.com/herman/about", "content": "Learn more about the Herman comic strip, its cast of characters, and creator Jim Unger."}
+{"idx": 5, "title": "Herman by Jim Unger for December 21, 2024 | GoComics", "date": "", "ddg_snippet": "Dec 21, 2024 · Read Herman —a comic strip by creator Jim Unger—for today, December 21, 2024, and check out other great comics, too!", "subpage_snippet": "", "source": "www.gocomics.com", "link": "https://www.gocomics.com/herman/2024/12/21", "content": "Dec 21, 2024 · Read Herman —a comic strip by creator Jim Unger—for today, December 21, 2024, and check out other great comics, too!"}
+{"idx": 6, "title": "Herman by Jim Unger for June 12, 2022 - GoComics", "date": "", "ddg_snippet": "Jun 12, 2022 · Read Herman —a comic strip by creator Jim Unger—for today, June 12, 2022, and check out other great comics, too!", "subpage_snippet": "", "source": "www.gocomics.com", "link": "https://www.gocomics.com/herman/2022/06/12", "content": "Jun 12, 2022 · Read Herman —a comic strip by creator Jim Unger—for today, June 12, 2022, and check out other great comics, too!"}
+{"idx": 7, "title": "synthical.com/search/by_author/Davis%20Brown", "date": "", "ddg_snippet": "Machine Learning meets Algebraic Combinatorics : A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics", "subpage_snippet": "", "source": "synthical.com", "link": "https://synthical.com/search/by_author/Davis+Brown", "content": "Machine Learning meets Algebraic Combinatorics : A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics"}
+{"idx": 8, "title": "Davis Brown", "date": "", "ddg_snippet": "We build datasets and tasks desgined to evaluate a model's ability to generate new research-level algebraic combinatorics conjectures.", "subpage_snippet": "", "source": "davisrbrown.com", "link": "https://davisrbrown.com/", "content": "We build datasets and tasks desgined to evaluate a model's ability to generate new research-level algebraic combinatorics conjectures."}
+{"idx": 9, "title": "MATH-AI: The 4th Workshop on Mathematical Reasoning and AI", "date": "", "ddg_snippet": "Proving Olympiad Algebraic Inequalities without ... HARDMATH: A Benchmark Dataset for Challenging Problems in Applied Mathematics ( Poster ) > link", "subpage_snippet": "", "source": "neurips.cc", "link": "https://neurips.cc/virtual/2024/workshop/84719", "content": "Proving Olympiad Algebraic Inequalities without ... HARDMATH: A Benchmark Dataset for Challenging Problems in Applied Mathematics ( Poster ) > link"}
diff --git a/data/sampled_jsons/HtmlRAG_Section_3.2.1_exact_text.jsonl b/data/sampled_jsons/HtmlRAG_Section_3.2.1_exact_text.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9feb5f47fada698ffef66310af20934665a20bd3
--- /dev/null
+++ b/data/sampled_jsons/HtmlRAG_Section_3.2.1_exact_text.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "plageon/ HtmlRAG : HtmlRAG : HTML is Better Than Plain Text for...", "date": "", "ddg_snippet": "We propose HtmlRAG , which uses HTML instead of plain text as the format of external knowledge in RAG systems. To tackle the long context brought by HTML , we propose Lossless HTML Cleaning and Two-Step Block-Tree-Based HTML Pruning.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/plageon/HtmlRAG", "content": "We propose HtmlRAG , which uses HTML instead of plain text as the format of external knowledge in RAG systems. To tackle the long context brought by HTML , we propose Lossless HTML Cleaning and Two-Step Block-Tree-Based HTML Pruning."}
+{"idx": 1, "title": "Implementing HtmlRAG : Enhancing Retrieval-Augmented... | Dev Genius", "date": "", "ddg_snippet": "HtmlRAG is a novel approach that proposes using HTML as the format of retrieved knowledge in RAG systems instead of plain text . By retaining the HTML structure, LLMs can access richer semantic and structural information, leading to improved understanding and generation capabilities.", "subpage_snippet": "", "source": "blog.devgenius.io", "link": "https://blog.devgenius.io/implementing-htmlrag-enhancing-retrieval-augmented-generation-with-html-knowledge-91cdd6278e23", "content": "HtmlRAG is a novel approach that proposes using HTML as the format of retrieved knowledge in RAG systems instead of plain text . By retaining the HTML structure, LLMs can access richer semantic and structural information, leading to improved understanding and generation capabilities."}
+{"idx": 2, "title": "HtmlRAG : HTML is Better Than Plain Text for Modeling Retrieved...", "date": "", "ddg_snippet": "In this paper, we propose HtmlRAG , which uses HTML instead of plain text as the format of retrieved knowledge in RAG systems, aiming to keep richer semantic and structured information that is missing in plain text .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.02959v1", "content": "In this paper, we propose HtmlRAG , which uses HTML instead of plain text as the format of retrieved knowledge in RAG systems, aiming to keep richer semantic and structured information that is missing in plain text ."}
+{"idx": 3, "title": "AI Innovations and Insights 20: HtmlRAG , AFLOW, ChunkRAG, and...", "date": "", "ddg_snippet": "HtmlRAG leverages HTML format instead of plain text in RAG systems to preserve semantic and structural information.", "subpage_snippet": "", "source": "readmedium.com", "link": "https://readmedium.com/ai-innovations-and-insights-20-htmlrag-aflow-chunkrag-and-markitdown-fe102693315e", "content": "HtmlRAG leverages HTML format instead of plain text in RAG systems to preserve semantic and structural information."}
+{"idx": 4, "title": "HtmlRAG : Building an Efficient HTML Retrieval Enhanced Generation...", "date": "", "ddg_snippet": "HtmlRAG is an innovative open source project focused on improving retrieval enhancement generation ( RAG ) approach to HTML document processing in the system. The project presents a novel approach that argues that using HTML formatting in RAG systems is more efficient than plain text .", "subpage_snippet": "", "source": "www.kdjingpai.com", "link": "https://www.kdjingpai.com/en/htmlrag/", "content": "HtmlRAG is an innovative open source project focused on improving retrieval enhancement generation ( RAG ) approach to HTML document processing in the system. The project presents a novel approach that argues that using HTML formatting in RAG systems is more efficient than plain text ."}
+{"idx": 5, "title": "Paper page - HtmlRAG : HTML is Better Than Plain Text for Modeling...", "date": "", "ddg_snippet": "Abstract. HtmlRAG enhances Retrieval-Augmented Generation ( RAG ) systems by using HTML instead of plain text , improving knowledge modeling and reducing information loss through HTML cleaning, compression, and pruning. AI-generated summary.", "subpage_snippet": "", "source": "www.aifasthub.com", "link": "https://www.aifasthub.com/papers/2411.02959", "content": "Abstract. HtmlRAG enhances Retrieval-Augmented Generation ( RAG ) systems by using HTML instead of plain text , improving knowledge modeling and reducing information loss through HTML cleaning, compression, and pruning. AI-generated summary."}
+{"idx": 6, "title": "HtmlRAG : Enhancing RAG Systems with Richer... - MarkTechPost", "date": "", "ddg_snippet": "HtmlRAG ’s superior performance compared to traditional plain- text -based post-retrieval processes validates the effectiveness of utilizing HTML format for knowledge retrieval.", "subpage_snippet": "", "source": "www.marktechpost.com", "link": "https://www.marktechpost.com/2024/11/10/htmlrag-enhancing-rag-systems-with-richer-semantic-and-structural-information-through-html/", "content": "HtmlRAG ’s superior performance compared to traditional plain- text -based post-retrieval processes validates the effectiveness of utilizing HTML format for knowledge retrieval."}
+{"idx": 7, "title": "HtmlRAG : Enhancing RAG Systems with Richer Semantic and...", "date": "", "ddg_snippet": "HtmlRAG is a new method that improves Retrieval-Augmented Generation ( RAG ) systems by using HTML instead of plain text . This approach helps maintain important structural and semantic information that is often lost during conversion to plain text .", "subpage_snippet": "", "source": "itinai.com", "link": "https://itinai.com/htmlrag-enhancing-rag-systems-with-richer-semantic-and-structural-information-through-html/", "content": "HtmlRAG is a new method that improves Retrieval-Augmented Generation ( RAG ) systems by using HTML instead of plain text . This approach helps maintain important structural and semantic information that is often lost during conversion to plain text ."}
+{"idx": 8, "title": "Unleashing the Power of HtmlRAG : Transforming RAG with HTML ...", "date": "", "ddg_snippet": "Discover how HtmlRAG , combined with UnDatas.io's HTML - based table data extraction, revolutionizes RAG systems. Learn how it preserves data structure, alleviates LLMs' hallucination problem, and outperforms traditional plain - text - based RAG through its unique workflow.", "subpage_snippet": "", "source": "undatas.io", "link": "https://undatas.io/blog/posts/unleashing-the-power-of-htmlrag-transforming-rag-with-html-enhanced-table-data-from-undatas-io/", "content": "Discover how HtmlRAG , combined with UnDatas.io's HTML - based table data extraction, revolutionizes RAG systems. Learn how it preserves data structure, alleviates LLMs' hallucination problem, and outperforms traditional plain - text - based RAG through its unique workflow."}
+{"idx": 9, "title": "htmlrag · PyPI", "date": "", "ddg_snippet": "A smart toolkit for HTML cleaning and pruning for RAG systems.If you switch from htmlrag v0.0.4 to v0.0.5, please download the latest version of modeling files for Gerative HTML Pruners, which are available at modeling_llama.py, and modeling_phi3.py.", "subpage_snippet": "", "source": "pypi.org", "link": "https://pypi.org/project/htmlrag/", "content": "A smart toolkit for HTML cleaning and pruning for RAG systems.If you switch from htmlrag v0.0.4 to v0.0.5, please download the latest version of modeling files for Gerative HTML Pruners, which are available at modeling_llama.py, and modeling_phi3.py."}
diff --git a/data/sampled_jsons/ICML_2025_checks_and_balances_ethical_AI_limitations.jsonl b/data/sampled_jsons/ICML_2025_checks_and_balances_ethical_AI_limitations.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..843f42a96341c3840bae339b7f6d969e2bd5ed8b
--- /dev/null
+++ b/data/sampled_jsons/ICML_2025_checks_and_balances_ethical_AI_limitations.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "A Checks-and-Balances Framework for Context-Aware ...", "date": "", "ddg_snippet": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment. 2 months ago. ·. ICML. Paper · ICML 2025 ... limitations : (1) the challenge of decomposing ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46461", "content": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment. 2 months ago. ·. ICML. Paper · ICML 2025 ... limitations : (1) the challenge of decomposing ..."}
+{"idx": 1, "title": "A Checks-and-Balances Framework for Context-Aware ...", "date": "", "ddg_snippet": "by EY Chang · Cited by 1 — A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... RLHF Limitations : ▷ Susceptible to social biases. ▷ Vulnerable to reward ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/media/icml-2025/Slides/46461_OMgXx2a.pdf", "content": "by EY Chang · Cited by 1 — A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... RLHF Limitations : ▷ Susceptible to social biases. ▷ Vulnerable to reward ..."}
+{"idx": 2, "title": "A Checks-and-Balances Framework for Context-Aware ...", "date": "", "ddg_snippet": "1 May 2025 — back arrow Go to ICML 2025 Conference homepage. A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... limitations and ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=4uOEiitySn&noteId=cYh3zaQycT", "content": "1 May 2025 — back arrow Go to ICML 2025 Conference homepage. A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... limitations and ..."}
+{"idx": 3, "title": "ICML 2025 Papers", "date": "", "ddg_snippet": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment · Premise-Augmented Reasoning Chains Improve Error Identification in Math reasoning ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/papers.html", "content": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment · Premise-Augmented Reasoning Chains Improve Error Identification in Math reasoning ..."}
+{"idx": 4, "title": "ICML 2025 Schedule", "date": "", "ddg_snippet": "Reasoning Limitations of Multimodal Large Language Models. A case study of Bongard Problems ... A Checks-and-Balances Framework for Context-Aware Ethical AI ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/calendar", "content": "Reasoning Limitations of Multimodal Large Language Models. A case study of Bongard Problems ... A Checks-and-Balances Framework for Context-Aware Ethical AI ..."}
+{"idx": 5, "title": "ICML 2025 Tuesday 07/15", "date": "", "ddg_snippet": "... limitations . We also show that WRO underperforms in practice due optimization ... A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment.", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/day/7/15", "content": "... limitations . We also show that WRO underperforms in practice due optimization ... A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment."}
+{"idx": 6, "title": "Digital SuperIntelligence", "date": "", "ddg_snippet": "In one paper (accepted to ICML 2025 ), researchers showed over a 20% performance gain by having the model minimize the “surprise” (perplexity) of ...", "subpage_snippet": "", "source": "www.linkedin.com", "link": "https://www.linkedin.com/pulse/digital-superintelligence-bassel-haidar-pxfje", "content": "In one paper (accepted to ICML 2025 ), researchers showed over a 20% performance gain by having the model minimize the “surprise” (perplexity) of ..."}
+{"idx": 7, "title": "Computer Science", "date": "", "ddg_snippet": "Title: A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... In this work, we analyze the strengths and limitations ... Comments: ICML 2025 ...", "subpage_snippet": "", "source": "www.arxiv.org", "link": "http://www.arxiv.org/list/cs/new?skip=1825&show=50", "content": "Title: A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... In this work, we analyze the strengths and limitations ... Comments: ICML 2025 ..."}
+{"idx": 8, "title": "Edward Chang", "date": "", "ddg_snippet": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment · Edward Y Chang. ICML 2025 poster; Readers: Everyone. Position: Limitations ... Submitted ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/profile?id=~Edward_Chang2", "content": "A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment · Edward Y Chang. ICML 2025 poster; Readers: Everyone. Position: Limitations ... Submitted ..."}
+{"idx": 9, "title": "Computer Science", "date": "", "ddg_snippet": "... limitations of LLM-based operators. ... Title: A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... Comments: Accepted at ICML 2025 . This ...", "subpage_snippet": "", "source": "www.arxiv.org", "link": "http://www.arxiv.org/list/cs/new?skip=875&show=500", "content": "... limitations of LLM-based operators. ... Title: A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment ... Comments: Accepted at ICML 2025 . This ..."}
diff --git a/data/sampled_jsons/IMhoJgWANP_Digital_Disparities-_A_Comparative_Web_Measurement_Study_Across_Economic_Boundaries_Figur.jsonl b/data/sampled_jsons/IMhoJgWANP_Digital_Disparities-_A_Comparative_Web_Measurement_Study_Across_Economic_Boundaries_Figur.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..e9f848d2ff0348678e8aea13d73ccc3ee5821096
--- /dev/null
+++ b/data/sampled_jsons/IMhoJgWANP_Digital_Disparities-_A_Comparative_Web_Measurement_Study_Across_Economic_Boundaries_Figur.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Digital Disparities: A Comparative Web Measurement Study Across ...", "date": "", "ddg_snippet": "Concretely, we collect the largest dataset to date that compares web development practices across developed and developing regions -- 200,000 webpages across 20 countries -- which we aim to open source along with this publication.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/10.1145/3696410.3714647", "content": "Concretely, we collect the largest dataset to date that compares web development practices across developed and developing regions -- 200,000 webpages across 20 countries -- which we aim to open source along with this publication."}
+{"idx": 1, "title": "PDF Digital Disparities: A Comparative Web Measurement Study Across ...", "date": "", "ddg_snippet": "Specifically, we select a sample of 10 developing and 10 developed countries for a comparative analysis of their web ecosys-tems. The selection process was guided by two primary criteria: population size and the availability of 10,000 websites per country that met our inclusion criteria (§3.2) within the CrUX database [14].", "subpage_snippet": "", "source": "www.staicu.org", "link": "https://www.staicu.org/publications/www2025.pdf", "content": "Specifically, we select a sample of 10 developing and 10 developed countries for a comparative analysis of their web ecosys-tems. The selection process was guided by two primary criteria: population size and the availability of 10,000 websites per country that met our inclusion criteria (§3.2) within the CrUX database [14]."}
+{"idx": 2, "title": "kal-purush/digital-disparities-www25 - GitHub", "date": "", "ddg_snippet": "Digital Disparities : A Comparative Web Measurement Study This repository contains the dataset, tools, and analysis for our research paper \"Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries \", presented at WWW'25. The study explores differences in web development practices across developed and developing countries , focusing on web size, complexity, security ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/kal-purush/digital-disparities-www25", "content": "Digital Disparities : A Comparative Web Measurement Study This repository contains the dataset, tools, and analysis for our research paper \"Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries \", presented at WWW'25. The study explores differences in web development practices across developed and developing countries , focusing on web size, complexity, security ..."}
+{"idx": 3, "title": "Unequal Internet: study highlights differences between websites from ...", "date": "", "ddg_snippet": "The Internet may be a global phenomenon, but its often-claimed global nature is tempered by the 'digital divide' - digital participation still heavily depends on economic conditions. CISPA researcher Masudul Bhuiyan from CISPA Faculty Dr. Cristian-Alexander Staicu's team explored whether security and data privacy differences can be found on websites as well. His study of 200,000 ...", "subpage_snippet": "", "source": "nachrichten.idw-online.de", "link": "https://nachrichten.idw-online.de/2025/04/29/unequal-internet-study-highlights-differences-between-websites-from-developing-and-developed-countries", "content": "The Internet may be a global phenomenon, but its often-claimed global nature is tempered by the 'digital divide' - digital participation still heavily depends on economic conditions. CISPA researcher Masudul Bhuiyan from CISPA Faculty Dr. Cristian-Alexander Staicu's team explored whether security and data privacy differences can be found on websites as well. His study of 200,000 ..."}
+{"idx": 4, "title": "Digital Disparities: A Comparative Web Measurement Study Across ...", "date": "", "ddg_snippet": "Figure 10 shows the adoption of HTTPS across both developed and developing nations. In developed countries , only 5% of websites still use unencrypted HTTP , a significant improvement from previously-reported mea-surements [35].", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=IMhoJgWANP", "content": "Figure 10 shows the adoption of HTTPS across both developed and developing nations. In developed countries , only 5% of websites still use unencrypted HTTP , a significant improvement from previously-reported mea-surements [35]."}
+{"idx": 5, "title": "\"Digital Disparities: A Comparative Web Measurement Study Across ... - dblp", "date": "", "ddg_snippet": "Bibliographic details on Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries .", "subpage_snippet": "", "source": "dblp.org", "link": "https://dblp.org/rec/conf/www/BhuiyanVSZ25", "content": "Bibliographic details on Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries ."}
+{"idx": 6, "title": "PDF Unequal internet: Study highlights differences between websites from ...", "date": "", "ddg_snippet": "Visualization to the paper \" A Comparative Web Measurement Study Across Economic Boundaries .\" Credit: CISPA Helmholtz Center for Information Security The Internet may be a global phenomenon, but its often-claimed global nature is tempered by the \"digital divide\"—digital participation still heavily depends on economic conditions.", "subpage_snippet": "", "source": "techxplore.com", "link": "https://techxplore.com/news/2025-04-unequal-internet-highlights-differences-websites.pdf", "content": "Visualization to the paper \" A Comparative Web Measurement Study Across Economic Boundaries .\" Credit: CISPA Helmholtz Center for Information Security The Internet may be a global phenomenon, but its often-claimed global nature is tempered by the \"digital divide\"—digital participation still heavily depends on economic conditions."}
+{"idx": 7, "title": "digital-disparities-www25/README.md at main - GitHub", "date": "", "ddg_snippet": "Digital Disparities : A Comparative Web Measurement Study This repository contains the dataset, tools, and analysis for our research paper \"Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries \", presented at WWW'25. The study explores differences in web development practices across developed and developing countries , focusing on web size, complexity, security ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/kal-purush/digital-disparities-www25/blob/main/README.md", "content": "Digital Disparities : A Comparative Web Measurement Study This repository contains the dataset, tools, and analysis for our research paper \"Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries \", presented at WWW'25. The study explores differences in web development practices across developed and developing countries , focusing on web size, complexity, security ..."}
+{"idx": 8, "title": "Digital Disparities: A Comparative Web Measurement Study Across ...", "date": "", "ddg_snippet": "Anecdotal evidence suggests that websites in developing and developed regions differ significantly. In this work, we test this hypothesis by measuring differences in web development practices across the two groups of countries , using multiple dimensions: websites' size, complexity, security, privacy, quality, technology adoption, and accessibility.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=IMhoJgWANP", "content": "Anecdotal evidence suggests that websites in developing and developed regions differ significantly. In this work, we test this hypothesis by measuring differences in web development practices across the two groups of countries , using multiple dimensions: websites' size, complexity, security, privacy, quality, technology adoption, and accessibility."}
+{"idx": 9, "title": "Cristian-Alexandru Staicu", "date": "", "ddg_snippet": "(paper) The Web Conference 2025 Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries Masudul Hasan Masud Bhuiyan, Matteo Varvello, Cristian-Alexandru Staicu, Yasir Zaki (paper) EuroS&P 2025 CHARON: Polyglot Code Analysis for Detecting Vulnerabilities in Scripting Languages Native Extensions", "subpage_snippet": "", "source": "www.staicu.org", "link": "https://www.staicu.org/publications.html", "content": "(paper) The Web Conference 2025 Digital Disparities : A Comparative Web Measurement Study Across Economic Boundaries Masudul Hasan Masud Bhuiyan, Matteo Varvello, Cristian-Alexandru Staicu, Yasir Zaki (paper) EuroS&P 2025 CHARON: Polyglot Code Analysis for Detecting Vulnerabilities in Scripting Languages Native Extensions"}
diff --git a/data/sampled_jsons/IP-Adapter_CLIP_encoding_VAE_latent_space_failure_origin_identification.jsonl b/data/sampled_jsons/IP-Adapter_CLIP_encoding_VAE_latent_space_failure_origin_identification.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..fe6ca458a7da756e41294066acc1d1ef3d3ba619
--- /dev/null
+++ b/data/sampled_jsons/IP-Adapter_CLIP_encoding_VAE_latent_space_failure_origin_identification.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Exploring CLIP Latent Space | tuba", "date": "", "ddg_snippet": "Sep 9, 2024 · One interesting area of research today is the open problem of how to control image models better. For example, it is difficult to tweak an output of an image model through prompting – a small change in prompt usually ends up changing the entire image. New techniques such as ControlNet and IP-Adapter are ways that users can maintain the structure of an image while changing small aspects of it ...", "subpage_snippet": "", "source": "0xtuba.github.io", "link": "https://0xtuba.github.io/posts/clip-latent-space/", "content": "Sep 9, 2024 · One interesting area of research today is the open problem of how to control image models better. For example, it is difficult to tweak an output of an image model through prompting – a small change in prompt usually ends up changing the entire image. New techniques such as ControlNet and IP-Adapter are ways that users can maintain the structure of an image while changing small aspects of it ..."}
+{"idx": 1, "title": "[2503.08455] Controlling Latent Diffusion Using Latent CLIP", "date": "", "ddg_snippet": "Mar 11, 2025 · Instead of performing text-conditioned denoising in the image domain, latent diffusion models (LDMs) operate in latent space of a variational autoencoder ( VAE ), enabling more efficient processing at reduced computational costs. However, while the diffusion process has moved to the latent space , the contrastive language-image pre-training ( CLIP ) models, as used in many image processing tasks ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2503.08455", "content": "Mar 11, 2025 · Instead of performing text-conditioned denoising in the image domain, latent diffusion models (LDMs) operate in latent space of a variational autoencoder ( VAE ), enabling more efficient processing at reduced computational costs. However, while the diffusion process has moved to the latent space , the contrastive language-image pre-training ( CLIP ) models, as used in many image processing tasks ..."}
+{"idx": 2, "title": "Latent Space Representations in Variational Autoencoders ...", "date": "", "ddg_snippet": "Dec 9, 2024 · Extract Latent Vectors: After training the VAE , I extract the latent representations for the dataset. This gives me a clear view of how the encoder maps input data into the latent space .", "subpage_snippet": "", "source": "medium.com", "link": "https://medium.com/@whyamit101/latent-space-representations-in-variational-autoencoders-vaes-e74076eda77b", "content": "Dec 9, 2024 · Extract Latent Vectors: After training the VAE , I extract the latent representations for the dataset. This gives me a clear view of how the encoder maps input data into the latent space ."}
+{"idx": 3, "title": "278-stable-diffusion-ip-adapter.ipynb - Colab - Google Colab", "date": "", "ddg_snippet": "The stable diffusion model with ip-adapter takes a latent image representation, a text prompt is transformed to text embeddings via CLIP text encoder and ip-adapter image is transformed to image embeddings via CLIP Image Encoder.", "subpage_snippet": "", "source": "colab.research.google.com", "link": "https://colab.research.google.com/github/openvinotoolkit/openvino_notebooks/blob/main/notebooks/278-stable-diffusion-ip-adapter/278-stable-diffusion-ip-adapter.ipynb", "content": "The stable diffusion model with ip-adapter takes a latent image representation, a text prompt is transformed to text embeddings via CLIP text encoder and ip-adapter image is transformed to image embeddings via CLIP Image Encoder."}
+{"idx": 4, "title": "GitHub - wendlerc/latent_clip: CLIP trained in SDXL VAE's ...", "date": "", "ddg_snippet": "CLIP trained in SDXL VAE 's latent space . Contribute to wendlerc/ latent _ clip development by creating an account on GitHub .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/wendlerc/latent_clip", "content": "CLIP trained in SDXL VAE 's latent space . Contribute to wendlerc/ latent _ clip development by creating an account on GitHub ."}
+{"idx": 5, "title": "Condition Monitoring Using a Latent Space of Variational ...", "date": "", "ddg_snippet": "Sep 22, 2024 · The VAE encoder maps input data to a continuous probabilistic latent subspace, and the VAE decoder generates output data by randomly sampling in the continuous latent subspace and decoding these samples. This paper introduces a unique approach to fault detection in rotating machines.", "subpage_snippet": "", "source": "www.mdpi.com", "link": "https://www.mdpi.com/1424-8220/24/21/6825", "content": "Sep 22, 2024 · The VAE encoder maps input data to a continuous probabilistic latent subspace, and the VAE decoder generates output data by randomly sampling in the continuous latent subspace and decoding these samples. This paper introduces a unique approach to fault detection in rotating machines."}
+{"idx": 6, "title": "Adaptive Compression of the Latent Space in Variational ...", "date": "", "ddg_snippet": "Sep 17, 2024 · In this paper, we propose ALD- VAE (Adaptive Latent Dimensionality VAE ) - a novel, automated heuristic for determining the optimal latent space size in VAEs. Our approach involves gradually decreasing the latent space size by removing neurons during the training process and observing the reconstruction loss, Fréchet Inception Distance (FID) and ...", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/chapter/10.1007/978-3-031-72332-2_7", "content": "Sep 17, 2024 · In this paper, we propose ALD- VAE (Adaptive Latent Dimensionality VAE ) - a novel, automated heuristic for determining the optimal latent space size in VAEs. Our approach involves gradually decreasing the latent space size by removing neurons during the training process and observing the reconstruction loss, Fréchet Inception Distance (FID) and ..."}
+{"idx": 7, "title": "VAE Decode (Tiled) - ComfyUI Community Manual", "date": "", "ddg_snippet": "VAE Encode (for Inpainting).The VAE Decode (Tiled) node can be used to decode latent space images back into pixel space images, using the provided VAE .", "subpage_snippet": "", "source": "blenderneko.github.io", "link": "https://blenderneko.github.io/ComfyUI-docs/Core+Nodes/Experimental/VAEDecodeTiled/", "content": "VAE Encode (for Inpainting).The VAE Decode (Tiled) node can be used to decode latent space images back into pixel space images, using the provided VAE ."}
+{"idx": 8, "title": "Variational Autoencoder − Dimension of the latent space", "date": "", "ddg_snippet": "After the training of a deep convolutional VAE with a large latent space (8x8x1024) on MNIST, the reconstruction works very well.", "subpage_snippet": "", "source": "stats.stackexchange.com", "link": "https://stats.stackexchange.com/questions/327966/variational-autoencoder-−-dimension-of-the-latent-space", "content": "After the training of a deep convolutional VAE with a large latent space (8x8x1024) on MNIST, the reconstruction works very well."}
+{"idx": 9, "title": "(PDF) Generalizable Origin Identification for Text-Guided...", "date": "", "ddg_snippet": "Generalizable Origin Identification for. Text-Guided Image-to-Image Diffusion Models.on IP - Adapter , which uses CLIP for encoding . We also try. the linear transformed CLIP embedding, but it still fails to. generalize (36.6% mAP and 27.8% Acc).", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/387767437_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Models", "content": "Generalizable Origin Identification for. Text-Guided Image-to-Image Diffusion Models.on IP - Adapter , which uses CLIP for encoding . We also try. the linear transformed CLIP embedding, but it still fails to. generalize (36.6% mAP and 27.8% Acc)."}
diff --git a/data/sampled_jsons/ITBench_Table_4_'Mean_Time_to_Diagnosis'_lowest_model_seconds_year_2024.jsonl b/data/sampled_jsons/ITBench_Table_4_'Mean_Time_to_Diagnosis'_lowest_model_seconds_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..d3e4cc8836b6ffabda1796ae0b55b67b8203d20e
--- /dev/null
+++ b/data/sampled_jsons/ITBench_Table_4_'Mean_Time_to_Diagnosis'_lowest_model_seconds_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "ITBench: Evaluating AI Agents across Diverse Real-World IT Automation Tasks", "date": "", "ddg_snippet": "3. ITBench ITBench is a systematic benchmarking framework and run- time environment designed to evaluate AI agents tasked with automating IT operations, incorporating a robust archi-tecture (see Figure 2) comprising the AI Agent, Scenario Specification and Environment, Evaluator, and Leaderboard to facilitate comprehensive performance assessment.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2502.05352", "content": "3. ITBench ITBench is a systematic benchmarking framework and run- time environment designed to evaluate AI agents tasked with automating IT operations, incorporating a robust archi-tecture (see Figure 2) comprising the AI Agent, Scenario Specification and Environment, Evaluator, and Leaderboard to facilitate comprehensive performance assessment."}
+{"idx": 1, "title": "GitHub - itbench-hub/ITBench: Code repository for ITBench", "date": "", "ddg_snippet": "Code repository for ITBench . Contribute to itbench -hub/ ITBench development by creating an account on GitHub.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/itbench-hub/ITBench", "content": "Code repository for ITBench . Contribute to itbench -hub/ ITBench development by creating an account on GitHub."}
+{"idx": 2, "title": "ITBench: Next-Gen Benchmarking for IT Automation Evaluation", "date": "", "ddg_snippet": "Learn more about ITBench , an open framework to benchmark AI agents for IT automation, focusing on reliability, efficiency, and real-world IT scenarios.", "subpage_snippet": "", "source": "dzone.com", "link": "https://dzone.com/articles/itbench-next-gen-benchmarking-it-automation", "content": "Learn more about ITBench , an open framework to benchmark AI agents for IT automation, focusing on reliability, efficiency, and real-world IT scenarios."}
+{"idx": 3, "title": "(PDF) ITBench: Evaluating AI Agents across Diverse Real-World IT ...", "date": "", "ddg_snippet": "We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/388882803_ITBench_Evaluating_AI_Agents_across_Diverse_Real-World_IT_Automation_Tasks", "content": "We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks."}
+{"idx": 4, "title": "GitHub - IBM/ITBench-Utilities: Code repository for tools as part of ...", "date": "", "ddg_snippet": "Code repository for tools as part of ITBench . Contribute to IBM/ ITBench -Utilities development by creating an account on GitHub.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/IBM/ITBench-Utilities/", "content": "Code repository for tools as part of ITBench . Contribute to IBM/ ITBench -Utilities development by creating an account on GitHub."}
+{"idx": 5, "title": "ITBench User Experience: Democratizing AI Agent Evaluation", "date": "", "ddg_snippet": "Learn how ITBench enables realistic AI agent evaluation in IT with streamlined onboarding, automated benchmarks, and meaningful metrics for real-world impact.", "subpage_snippet": "", "source": "dzone.com", "link": "https://dzone.com/articles/itbench-part-2-ai-agent-evaluation-guide", "content": "Learn how ITBench enables realistic AI agent evaluation in IT with streamlined onboarding, automated benchmarks, and meaningful metrics for real-world impact."}
+{"idx": 6, "title": "ITBench: Evaluating AI Agents across Diverse Real-World IT Automation Tasks", "date": "", "ddg_snippet": "Realizing the vision of using AI agents to automate critical IT tasks depends on the ability to measure and understand effectiveness of proposed solutions. We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2502.05352", "content": "Realizing the vision of using AI agents to automate critical IT tasks depends on the ability to measure and understand effectiveness of proposed solutions. We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and ..."}
+{"idx": 7, "title": "GitHub - itbench-hub/ITBench-Scenarios: Code repository for scenarios ...", "date": "", "ddg_snippet": "ITBench : Central repository providing an overview of the ITBench ecosystem, the ITBench Leaderboard, related announcements, and publications. CISO-CAA Agent: CISO (Chief Information Security Officer) agents that automate compliance assessments by generating policies from natural language, collecting evidence, integrating with GitOps workflows ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/itbench-hub/ITBench-Scenarios", "content": "ITBench : Central repository providing an overview of the ITBench ecosystem, the ITBench Leaderboard, related announcements, and publications. CISO-CAA Agent: CISO (Chief Information Security Officer) agents that automate compliance assessments by generating policies from natural language, collecting evidence, integrating with GitOps workflows ..."}
+{"idx": 8, "title": "Now you can gauge if AI agents are actually useful for work", "date": "", "ddg_snippet": "IBM Research's new ITBench benchmarks aim to bring an objective approach to seeing whether IT agents are actually making work easier for enterprises.", "subpage_snippet": "", "source": "research.ibm.com", "link": "https://research.ibm.com/blog/it-agent-benchmark", "content": "IBM Research's new ITBench benchmarks aim to bring an objective approach to seeing whether IT agents are actually making work easier for enterprises."}
+{"idx": 9, "title": "Publications | Yinfang CHEN", "date": "", "ddg_snippet": "We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and Security Operations (CISO), and Financial Operations (FinOps).", "subpage_snippet": "", "source": "yinfangchen.github.io", "link": "https://yinfangchen.github.io/publications/", "content": "We introduce ITBench , a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and Security Operations (CISO), and Financial Operations (FinOps)."}
diff --git a/data/sampled_jsons/ImageNet-1k_Syn._DP_(ours)_6.5M_IN_real_Val..jsonl b/data/sampled_jsons/ImageNet-1k_Syn._DP_(ours)_6.5M_IN_real_Val..jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..96b09fe2232ba92fb95b5c0fe430f6e8ccc27588
--- /dev/null
+++ b/data/sampled_jsons/ImageNet-1k_Syn._DP_(ours)_6.5M_IN_real_Val..jsonl
@@ -0,0 +1,3 @@
+{"idx": 0, "title": "(PDF) Improving the Scaling Laws of Synthetic Data with Deliberate...", "date": "", "ddg_snippet": "Syn . DP ( ours ) IN-100 100k 1.9 M 74.3 75.0 66.3 52.0 76.6 25.9.On ImageNet -100, DP generated 4.6. million fewer samples and trained for only one-sixth of the iterations compared to previous works, yet achieved. superior performance on the real data.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/389274598_Improving_the_Scaling_Laws_of_Synthetic_Data_with_Deliberate_Practice", "content": "Syn . DP ( ours ) IN-100 100k 1.9 M 74.3 75.0 66.3 52.0 76.6 25.9.On ImageNet -100, DP generated 4.6. million fewer samples and trained for only one-sixth of the iterations compared to previous works, yet achieved. superior performance on the real data."}
+{"idx": 1, "title": "Deliberate Practice", "date": "", "ddg_snippet": "by R Askari-Hemmat · 2025 · Cited by 1 — (Right): Top-1 validation accuracy on ImageNet-1k with models trained solely on synthetic data. ... Syn. DP (ours ). IN-100. 100k. 1.9M. 74.3.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2502.15588", "content": "by R Askari-Hemmat · 2025 · Cited by 1 — (Right): Top-1 validation accuracy on ImageNet-1k with models trained solely on synthetic data. ... Syn. DP (ours ). IN-100. 100k. 1.9M. 74.3."}
+{"idx": 2, "title": "", "date": "", "ddg_snippet": "", "subpage_snippet": "", "source": "", "link": "", "content": ""}
diff --git a/data/sampled_jsons/ImagineFSL_DISEF_Flowers_Table_1_github.jsonl b/data/sampled_jsons/ImagineFSL_DISEF_Flowers_Table_1_github.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..e1ca51bbbf8239ffd56e9cb6b498a9e59833c3f1
--- /dev/null
+++ b/data/sampled_jsons/ImagineFSL_DISEF_Flowers_Table_1_github.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Amazing Table of Flowers and Epoxy resin. - YouTube", "date": "", "ddg_snippet": "How to make an amazing and functional table of flowers , plywood and epoxy with your own hands. #positivecoupleyoutube #art #epoxyOur channel with music:Posit...", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=AsopsTPbnwM", "content": "How to make an amazing and functional table of flowers , plywood and epoxy with your own hands. #positivecoupleyoutube #art #epoxyOur channel with music:Posit..."}
+{"idx": 1, "title": "Markdown table: table - 1 · GitHub", "date": "", "ddg_snippet": "Markdown table: table - 1 . GitHub Gist: instantly share code, notes, and snippets.Save will-edward-moore/0d35453aa7 f 8c01c33b9 f 8d7429b8e3c to your computer and use it in GitHub Desktop. Download ZIP. Markdown table: table - 1 .", "subpage_snippet": "", "source": "gist.github.com", "link": "https://gist.github.com/will-edward-moore/0d35453aa7f8c01c33b9f8d7429b8e3c", "content": "Markdown table: table - 1 . GitHub Gist: instantly share code, notes, and snippets.Save will-edward-moore/0d35453aa7 f 8c01c33b9 f 8d7429b8e3c to your computer and use it in GitHub Desktop. Download ZIP. Markdown table: table - 1 ."}
+{"idx": 2, "title": "How to Create a Folder in Github Repos in 4 Simple Steps - αlphαrithms", "date": "", "ddg_snippet": "github create new folder illustration. Github is an incredible tool for storing, sharing, and learning code. Table of Contents show. 1 Overview. 2 Step 1 : Have a Repository. 3 Step 2: Click “creating a new file”. 4 Step 3: Use the Forward-Slash Character to Create a New Folder.", "subpage_snippet": "", "source": "www.alpharithms.com", "link": "https://www.alpharithms.com/how-to-create-a-folder-in-github-repos-463022/", "content": "github create new folder illustration. Github is an incredible tool for storing, sharing, and learning code. Table of Contents show. 1 Overview. 2 Step 1 : Have a Repository. 3 Step 2: Click “creating a new file”. 4 Step 3: Use the Forward-Slash Character to Create a New Folder."}
+{"idx": 3, "title": "Periodic Table - Ptable - Properties", "date": "", "ddg_snippet": "Interactive periodic table showing names, electrons, and oxidation states.", "subpage_snippet": "", "source": "ptable.com", "link": "https://ptable.com/", "content": "Interactive periodic table showing names, electrons, and oxidation states."}
+{"idx": 4, "title": "Как FSR 4 на RX 6900 XT запускали — Софт на DTF", "date": "", "ddg_snippet": "Как многие знают, относительно недавно AMD на короткое время выложила исходный код FSR 4 на GitHub . И спустя непродолжительное время с помощью OptiScaler его начали запускать на видеокартах rdna 2 и 3, в том числе удачно — и на Steam Deck.", "subpage_snippet": "", "source": "dtf.ru", "link": "https://dtf.ru/software/4035298-fsr-4-na-rx-6900-xt-zapusk-i-testirovanie", "content": "Как многие знают, относительно недавно AMD на короткое время выложила исходный код FSR 4 на GitHub . И спустя непродолжительное время с помощью OptiScaler его начали запускать на видеокартах rdna 2 и 3, в том числе удачно — и на Steam Deck."}
+{"idx": 5, "title": "Profile Readme Generator", "date": "", "ddg_snippet": "Beautify your github profile with this amazing tool, creating the readme your way in a simple and fast way! The best profile readme generator you will find!", "subpage_snippet": "", "source": "profile-readme-generator.com", "link": "https://profile-readme-generator.com/", "content": "Beautify your github profile with this amazing tool, creating the readme your way in a simple and fast way! The best profile readme generator you will find!"}
+{"idx": 6, "title": "Stardew Valley Vanilla IDs | Full 1 .6 Item List", "date": "", "ddg_snippet": "Mixed Flower Seeds. [MixedFlowerSeeds]. Golden Bobber.Ancient Table .", "subpage_snippet": "", "source": "mateusaquino.github.io", "link": "https://mateusaquino.github.io/stardewids/", "content": "Mixed Flower Seeds. [MixedFlowerSeeds]. Golden Bobber.Ancient Table ."}
+{"idx": 7, "title": "Rudalle — Используйте технические возможности моделей для...", "date": "", "ddg_snippet": "Код и веса модели находятся в открытом доступе: GitHub , HuggingFace. Примеры видеороликов можно посмотреть в разделе Галерея, а самостоятельно потестировать генерацию видео на платформе fusionbrain.ai и в Telegram-боте.", "subpage_snippet": "", "source": "rudalle.ru", "link": "https://rudalle.ru/", "content": "Код и веса модели находятся в открытом доступе: GitHub , HuggingFace. Примеры видеороликов можно посмотреть в разделе Галерея, а самостоятельно потестировать генерацию видео на платформе fusionbrain.ai и в Telegram-боте."}
+{"idx": 8, "title": "SSH и удалённые git-репозитории / Хабр", "date": "", "ddg_snippet": "GitHub , GitLab и Yandex рекомендуют использовать ed25519. Название файла, в который будет сохранён ключ, можно оставить дефолтным. Далее по желанию можно задать пароль на генерируемый ключ.", "subpage_snippet": "", "source": "habr.com", "link": "https://habr.com/ru/articles/755036/", "content": "GitHub , GitLab и Yandex рекомендуют использовать ed25519. Название файла, в который будет сохранён ключ, можно оставить дефолтным. Далее по желанию можно задать пароль на генерируемый ключ."}
+{"idx": 9, "title": "SELECT DISTINCT ON in PostgreSQL | Geeky Tidbits", "date": "", "ddg_snippet": "GitHub .That INNER JOIN with a subquery is used to determine the last timestamp for each URL. Then, the outer query is pulling from the log table and using the results of that subquery to limit the results to only the last requests by URL.", "subpage_snippet": "", "source": "www.geekytidbits.com", "link": "https://www.geekytidbits.com/postgres-distinct-on/", "content": "GitHub .That INNER JOIN with a subquery is used to determine the last timestamp for each URL. Then, the outer query is pulling from the log table and using the results of that subquery to limit the results to only the last requests by URL."}
diff --git a/data/sampled_jsons/ImagineFSL_Self-Supervised_Pretraining_Matters_reference_21.jsonl b/data/sampled_jsons/ImagineFSL_Self-Supervised_Pretraining_Matters_reference_21.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..aab0436d8543d1341c5b3ffbcc04b65ae093b3c1
--- /dev/null
+++ b/data/sampled_jsons/ImagineFSL_Self-Supervised_Pretraining_Matters_reference_21.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "ImagineFSL: Self-Supervised Pretraining Matters on Imagined ...", "date": "", "ddg_snippet": "We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter pro-viding better performance. Based on on this finding, we propose an improved self-supervised method tailored for few-shot scenarios, enhancing the transferability of repre-sentations from synthetic to real image domains.", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/papers/Yang_ImagineFSL_Self-Supervised_Pretraining_Matters_on_Imagined_Base_Set_for_VLM-based_CVPR_2025_paper.pdf", "content": "We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter pro-viding better performance. Based on on this finding, we propose an improved self-supervised method tailored for few-shot scenarios, enhancing the transferability of repre-sentations from synthetic to real image domains."}
+{"idx": 1, "title": "ImagineFSL: Self-Supervised Pretraining Matters on Imagined ...", "date": "", "ddg_snippet": "We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter providing better performance. Based on on this finding, we propose an improved self-supervised method tailored for few-shot scenarios, enhancing the transferability of representations from synthetic to real image domains.", "subpage_snippet": "", "source": "ieeexplore.ieee.org", "link": "https://ieeexplore.ieee.org/document/11094502", "content": "We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter providing better performance. Based on on this finding, we propose an improved self-supervised method tailored for few-shot scenarios, enhancing the transferability of representations from synthetic to real image domains."}
+{"idx": 2, "title": "Releases: HaoyuanYang-2023/ImagineFSL_Preview - GitHub", "date": "", "ddg_snippet": "Official implementation of \" ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning\" [CVPR 2025 Highlight] - HaoyuanYang ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/HaoyuanYang-2023/ImagineFSL_Preview/releases", "content": "Official implementation of \" ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning\" [CVPR 2025 Highlight] - HaoyuanYang ..."}
+{"idx": 3, "title": "dblp: ImagineFSL: Self-Supervised Pretraining Matters on ...", "date": "", "ddg_snippet": "Jul 21 , 2025 · Bibliographic details on ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning.", "subpage_snippet": "", "source": "dblp.org", "link": "https://dblp.org/rec/conf/cvpr/YangLLCWL25", "content": "Jul 21 , 2025 · Bibliographic details on ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning."}
+{"idx": 4, "title": "ImagineFSL: Self-Supervised Pretraining Matters on Imagined ...", "date": "", "ddg_snippet": "We introduce a novel CLIP adaptation methodology called * ImagineFSL *, involving pretraining on the imagined base set followed by fine-tuning on downstream few-shot tasks. We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter providing better performance.", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/html/Yang_ImagineFSL_Self-Supervised_Pretraining_Matters_on_Imagined_Base_Set_for_VLM-based_CVPR_2025_paper.html", "content": "We introduce a novel CLIP adaptation methodology called * ImagineFSL *, involving pretraining on the imagined base set followed by fine-tuning on downstream few-shot tasks. We find that, compared to no pretraining , both supervised and self-supervised pretraining are beneficial, with the latter providing better performance."}
+{"idx": 5, "title": "Self-Supervised Pretraining Matters on Imagined Base Set ...", "date": "", "ddg_snippet": "by H Yang — The hyperparameters of fine-tuning for few-shot tasks are presented in Table S-2. For ImagineFSL , we use. AdamW optimizer with a cosine decay of learning rate.", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/supplemental/Yang_ImagineFSL_Self-Supervised_Pretraining_CVPR_2025_supplemental.pdf", "content": "by H Yang — The hyperparameters of fine-tuning for few-shot tasks are presented in Table S-2. For ImagineFSL , we use. AdamW optimizer with a cosine decay of learning rate."}
+{"idx": 6, "title": "Homepage of Qilong Wang - GitHub Pages", "date": "", "ddg_snippet": "Mar 3, 2022 · ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning. 38th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025 [PDF] [Code] (highlight paper (13.5%)) Junjie Wu, Qilong Wang *, Jiangtao Xie, Pengfei Zhu, Qinghua Hu. Asymmetric Factorized Bilinear Operation for Vision Transformer.", "subpage_snippet": "", "source": "csqlwang.github.io", "link": "https://csqlwang.github.io/homepage/", "content": "Mar 3, 2022 · ImagineFSL : Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning. 38th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025 [PDF] [Code] (highlight paper (13.5%)) Junjie Wu, Qilong Wang *, Jiangtao Xie, Pengfei Zhu, Qinghua Hu. Asymmetric Factorized Bilinear Operation for Vision Transformer."}
+{"idx": 7, "title": "52CV/CVPR-2025-Papers", "date": "", "ddg_snippet": "ImagineFSL: Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning ... A Unified Approach to Interpreting Self-supervised Pre- ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/52CV/CVPR-2025-Papers", "content": "ImagineFSL: Self-Supervised Pretraining Matters on Imagined Base Set for VLM-based Few-shot Learning ... A Unified Approach to Interpreting Self-supervised Pre- ..."}
+{"idx": 8, "title": "IEEE/CVF Conference on Computer Vision and Pattern ...", "date": "", "ddg_snippet": "370 ImagineFSL : Self - Supervised Pretraining Matters on Imagined. ❉. Base Set for VLM-based Few-shot Learning, Haoyuan Yang,. ⭐ Xiaoou Li, Jiaming Lv, Xianjun ... 80 pages", "subpage_snippet": "", "source": "media.eventhosts.cc", "link": "https://media.eventhosts.cc/Conferences/CVPR2025/CVPR_main_conf_2025.pdf", "content": "370 ImagineFSL : Self - Supervised Pretraining Matters on Imagined. ❉. Base Set for VLM-based Few-shot Learning, Haoyuan Yang,. ⭐ Xiaoou Li, Jiaming Lv, Xianjun ... 80 pages"}
+{"idx": 9, "title": "Welcome to Vision by Statistical Learning lab - Peihua Li", "date": "", "ddg_snippet": "Representative publications Selected publications by area My google scholar citations", "subpage_snippet": "", "source": "www.peihuali.org", "link": "https://www.peihuali.org/Publications.html", "content": "Representative publications Selected publications by area My google scholar citations"}
diff --git a/data/sampled_jsons/JUICE_dual-run_approach_superposition_effects_attention_heads_year_2024.jsonl b/data/sampled_jsons/JUICE_dual-run_approach_superposition_effects_attention_heads_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..24be90ab3c7c04ea6e3261d87becff7be0748149
--- /dev/null
+++ b/data/sampled_jsons/JUICE_dual-run_approach_superposition_effects_attention_heads_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Massive Values in Self-Attention Modules are the Key to ...", "date": "", "ddg_snippet": "3 Feb 2025 — ... Run Twice ( JuICE ), a test-time attention ... attention heads and leverages a dual - run approach to mitigate the superposition effects .", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Massive-Values-in-Self-Attention-Modules-are-the-to-Jin-Mei/e57e636fe5dfd2d39eafe92448f53336ce6b8911", "content": "3 Feb 2025 — ... Run Twice ( JuICE ), a test-time attention ... attention heads and leverages a dual - run approach to mitigate the superposition effects ."}
+{"idx": 1, "title": "Taming Knowledge Conflicts in Language Models", "date": "", "ddg_snippet": "by G Li · 2025 — JuICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2503.10996", "content": "by G Li · 2025 — JuICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects."}
+{"idx": 2, "title": "Taming Knowledge Conflicts in Language Models", "date": "", "ddg_snippet": "JUICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects.", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46677", "content": "JUICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects."}
+{"idx": 3, "title": "Taming Knowledge Conflicts in Language Models", "date": "", "ddg_snippet": "14 Mar 2025 — JuICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.10996v1", "content": "14 Mar 2025 — JuICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive ..."}
+{"idx": 4, "title": "Taming Knowledge Conflicts in Language Models", "date": "", "ddg_snippet": "by G Li — JuICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=0cEZyhHEks", "content": "by G Li — JuICE identifies a set of reliable attention heads and leverages a dual-run approach to mitigate the superposition effects."}
+{"idx": 5, "title": "Daily Papers", "date": "", "ddg_snippet": "JUICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive experiments across 11 ...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/papers?q=JuICe", "content": "JUICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive experiments across 11 ..."}
+{"idx": 6, "title": "Daily Papers", "date": "", "ddg_snippet": "2 days ago — JUICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive ...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/papers?q=Mapping+Attention+head+ParameterS", "content": "2 days ago — JUICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive ..."}
+{"idx": 7, "title": "Mitigating Knowledge Conflicts in LLM Generation via Identifying ...", "date": "", "ddg_snippet": "This work proposes Just Run Twice ( JuICE ), a test-time attention ... attention heads and leverages a dual - run approach to mitigate the superposition effects .", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/IRCAN:-Mitigating-Knowledge-Conflicts-in-LLM-via-Shi-Jin/f95eb48dd381c04ee797d588c7417d71040c8ac0", "content": "This work proposes Just Run Twice ( JuICE ), a test-time attention ... attention heads and leverages a dual - run approach to mitigate the superposition effects ."}
+{"idx": 8, "title": "[Literature Review] Taming Knowledge Conflicts in ...", "date": "", "ddg_snippet": "14 Mar 2025 — JUICE consistently outperformed baseline methods , including existing model fine-tuning techniques. JUICE demonstrated robustness regarding the ...", "subpage_snippet": "", "source": "www.themoonlight.io", "link": "https://www.themoonlight.io/en/review/taming-knowledge-conflicts-in-language-models", "content": "14 Mar 2025 — JUICE consistently outperformed baseline methods , including existing model fine-tuning techniques. JUICE demonstrated robustness regarding the ..."}
+{"idx": 9, "title": "Spotlight Posters", "date": "", "ddg_snippet": "JuICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive experiments across 11 ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/events/2025SpotlightPosters", "content": "JuICE identifies a set of reliable attention heads and leverages a dual - run approach to mitigate the superposition effects . Extensive experiments across 11 ..."}
diff --git a/data/sampled_jsons/Kerner_2024_research_paper_publication_year_2024.jsonl b/data/sampled_jsons/Kerner_2024_research_paper_publication_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..7fd0fe27e46fc6368c60b963a08259ca19d0343c
--- /dev/null
+++ b/data/sampled_jsons/Kerner_2024_research_paper_publication_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Kerner -Collection, as of February 2024", "date": "", "ddg_snippet": "Gender gaps in publications and citations in gambling studies: Comparisons against addiction science. Sinclair, Eliscia Siu-Lin Liang; Clark", "subpage_snippet": "", "source": "criminologie.org.ro", "link": "https://criminologie.org.ro/wp-content/uploads/2024/02/3-Kerner-Collecton-APA-Journal-Articles_OA-FoC-2024.pdf", "content": "Gender gaps in publications and citations in gambling studies: Comparisons against addiction science. Sinclair, Eliscia Siu-Lin Liang; Clark"}
+{"idx": 1, "title": "About - Hannah Kerner", "date": "", "ddg_snippet": "November 2024 : Our paper , “Hotspotter: A Generalizable Pipeline for Automated Detection of Subtle Volcanic Thermal Features in Satellite Images,” was accepted for publication at AAAI 2025 Innovative Applications of AI! ( public link coming soon).", "subpage_snippet": "", "source": "hannah-rae.github.io", "link": "https://hannah-rae.github.io/", "content": "November 2024 : Our paper , “Hotspotter: A Generalizable Pipeline for Automated Detection of Subtle Volcanic Thermal Features in Satellite Images,” was accepted for publication at AAAI 2025 Innovative Applications of AI! ( public link coming soon)."}
+{"idx": 2, "title": "Kerner 2024 – Weingut Taschlerhof – Peter Wachtler", "date": "", "ddg_snippet": "HomeKerner 2024 . Kerner 2024 . Sort: Cross of Riesling x Trollinger (Vernatsch). Straw-yellow to greenish-yellow in colour, fruity with fresh note, taste of nutmeg, full and minerally.", "subpage_snippet": "", "source": "www.taschlerhof.com", "link": "https://www.taschlerhof.com/en/produkt/kerner/", "content": "HomeKerner 2024 . Kerner 2024 . Sort: Cross of Riesling x Trollinger (Vernatsch). Straw-yellow to greenish-yellow in colour, fruity with fresh note, taste of nutmeg, full and minerally."}
+{"idx": 3, "title": "Mathilde Bernard, prix Kerner 2024 de vulgarisation scientifique...", "date": "", "ddg_snippet": "Elle a remporté le 1er prix Kerner 2024 de vulgarisation scientifique de la Fondation ARC, lors de la 28e édition des Journées Jeunes Chercheurs en cancérologie.", "subpage_snippet": "", "source": "www.iledefrance-villejuif.cnrs.fr", "link": "https://www.iledefrance-villejuif.cnrs.fr/fr/cnrsinfo/mathilde-bernard-prix-kerner-2024-de-vulgarisation-scientifique-fondation-arc", "content": "Elle a remporté le 1er prix Kerner 2024 de vulgarisation scientifique de la Fondation ARC, lors de la 28e édition des Journées Jeunes Chercheurs en cancérologie."}
+{"idx": 4, "title": "Abbazia di Novacella Kerner 2024 | xtraWine", "date": "", "ddg_snippet": "Technical Sheet. Name. Abbazia di Novacella Kerner 2024 . Type. White still.Alto Adige DOC. Vintage. 2024 . Size.", "subpage_snippet": "", "source": "www.xtrawine.com", "link": "https://www.xtrawine.com/en-uk/products/abbazia-di-novacella-kerner-2024", "content": "Technical Sheet. Name. Abbazia di Novacella Kerner 2024 . Type. White still.Alto Adige DOC. Vintage. 2024 . Size."}
+{"idx": 5, "title": "Kerner 2024", "date": "", "ddg_snippet": "An elegant Kerner with a varietally typical aromatic expression and a smooth fruity palate.sandy loess. date of collection. 7.9. 2024 . sugar content at harvest.", "subpage_snippet": "", "source": "en.vinarstvibalaz.cz", "link": "https://en.vinarstvibalaz.cz/vino/kerner-2024-ps-klasik", "content": "An elegant Kerner with a varietally typical aromatic expression and a smooth fruity palate.sandy loess. date of collection. 7.9. 2024 . sugar content at harvest."}
+{"idx": 6, "title": "Ian Kerner - 2024 Annual Conference Speaker", "date": "", "ddg_snippet": "CAMFT's 2024 Annual Conference. May 3 - 4, 2024 at the Los Angeles Airport Marriott.Ian Kerner , PhD, LMFT is the co-Director of the sex therapy program at the Institute for Contemporary Psychotherapy and contributes regularly on the topic of sexual health for CNN.", "subpage_snippet": "", "source": "www.camft.org", "link": "https://www.camft.org/Education/Events/2024-Annual-Conference/Ian-Kerner", "content": "CAMFT's 2024 Annual Conference. May 3 - 4, 2024 at the Los Angeles Airport Marriott.Ian Kerner , PhD, LMFT is the co-Director of the sex therapy program at the Institute for Contemporary Psychotherapy and contributes regularly on the topic of sexual health for CNN."}
+{"idx": 7, "title": "Aduki: un Extraño Planeta by Ivan Kerner ( 2024 ...) | eBay Australia", "date": "", "ddg_snippet": "Publication Year 2024 . Language Spanish. Number of Pages 32 Pages. Author Ivan Kerner . Book Series Aduki Ser.", "subpage_snippet": "", "source": "www.ebay.com.au", "link": "https://www.ebay.com.au/p/28082408567", "content": "Publication Year 2024 . Language Spanish. Number of Pages 32 Pages. Author Ivan Kerner . Book Series Aduki Ser."}
+{"idx": 8, "title": "Information avoidance and testing for COVID‐19 - Kerner - 2024 ...", "date": "", "ddg_snippet": "To limit the spread of the COVID-19 virus, many employers and institutions developed procedures for people who tested positive. We propose that these procedures may have dissuaded people from testing...", "subpage_snippet": "", "source": "compass.onlinelibrary.wiley.com", "link": "https://compass.onlinelibrary.wiley.com/doi/10.1111/spc3.12920", "content": "To limit the spread of the COVID-19 virus, many employers and institutions developed procedures for people who tested positive. We propose that these procedures may have dissuaded people from testing..."}
+{"idx": 9, "title": "AI and Machine Learning Algorithms: Strengths, Weaknesses and Best...", "date": "", "ddg_snippet": "In supervised learning, the supervised learning algorithm learns a mapping between the input features and output labels” (Gupta, 2024 ). Unsupervised Learning Algorithms (ML-specific): K-Means Clustering, Hierarchical Clustering, and Principal Component Analysis (PCA).", "subpage_snippet": "", "source": "www.alexomegapy.com", "link": "https://www.alexomegapy.com/post/ai-and-machine-learning-algorithms-strengths-weaknesses-and-best-use-cases", "content": "In supervised learning, the supervised learning algorithm learns a mapping between the input features and output labels” (Gupta, 2024 ). Unsupervised Learning Algorithms (ML-specific): K-Means Clustering, Hierarchical Clustering, and Principal Component Analysis (PCA)."}
diff --git a/data/sampled_jsons/LLaVA-Next_4x_more_pixels_higher_resolution_672x672_336x1344_vision_encoder_year_2024.jsonl b/data/sampled_jsons/LLaVA-Next_4x_more_pixels_higher_resolution_672x672_336x1344_vision_encoder_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..763ecb530407670fd84f8a21bf26c2009a47dab2
--- /dev/null
+++ b/data/sampled_jsons/LLaVA-Next_4x_more_pixels_higher_resolution_672x672_336x1344_vision_encoder_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "LLaVA-NeXT", "date": "", "ddg_snippet": "20 Mar 2024 — LLaVA ‑ NeXT improves on Llava by increasing the input image resolution by 4x more ... 672x672 , 336x1344 , 1344x336) to better grasp visual details.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/docs/transformers/en/model_doc/llava_next", "content": "20 Mar 2024 — LLaVA ‑ NeXT improves on Llava by increasing the input image resolution by 4x more ... 672x672 , 336x1344 , 1344x336) to better grasp visual details."}
+{"idx": 1, "title": "LLaVA-NeXT: Improved reasoning, OCR, and world knowledge", "date": "", "ddg_snippet": "30 Jan 2024 — It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved ...", "subpage_snippet": "", "source": "llava-vl.github.io", "link": "https://llava-vl.github.io/blog/2024-01-30-llava-next/", "content": "30 Jan 2024 — It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved ..."}
+{"idx": 2, "title": "LLaVA-NeXT", "date": "", "ddg_snippet": "Increasing the input image resolution to 4x more pixels . This allows it to grasp more visual details. It supports three aspect ratios, up to 672x672 ...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/docs/transformers/v4.45.2/en/model_doc/llava_next", "content": "Increasing the input image resolution to 4x more pixels . This allows it to grasp more visual details. It supports three aspect ratios, up to 672x672 ..."}
+{"idx": 3, "title": "llava", "date": "", "ddg_snippet": "Increasing the input image resolution to up to 4x more pixels , supporting 672x672 , 336x1344 , 1344x336 resolutions . · Better visual reasoning and OCR capability ...", "subpage_snippet": "", "source": "ollama.com", "link": "https://ollama.com/library/llava", "content": "Increasing the input image resolution to up to 4x more pixels , supporting 672x672 , 336x1344 , 1344x336 resolutions . · Better visual reasoning and OCR capability ..."}
+{"idx": 4, "title": "LLaVA: Large Language and Vision Assistant Explained", "date": "", "ddg_snippet": "17 Oct 2023 — It supports three aspect ratios, up to 672x672 , 336x1344 , and 1344x336 resolution . Better visual reasoning and zero-shot OCR capability with ...", "subpage_snippet": "", "source": "encord.com", "link": "https://encord.com/blog/llava-large-language-vision-assistant/", "content": "17 Oct 2023 — It supports three aspect ratios, up to 672x672 , 336x1344 , and 1344x336 resolution . Better visual reasoning and zero-shot OCR capability with ..."}
+{"idx": 5, "title": "Large Language and Vision Assistant (LLaVA) — v1.6 vs. ...", "date": "", "ddg_snippet": "By increasing the input image resolution to four times more pixels , as mentioned above, the model now supports images up to 672 x 672 , 336 x ...", "subpage_snippet": "", "source": "medium.com", "link": "https://medium.com/@sulaiman.shamasna/large-language-and-vision-assistant-llava-v1-6-vs-v1-5-ede06b81ab48", "content": "By increasing the input image resolution to four times more pixels , as mentioned above, the model now supports images up to 672 x 672 , 336 x ..."}
+{"idx": 6, "title": "gokayfem/awesome-vlm-architectures: Famous Vision ...", "date": "", "ddg_snippet": "LLaVA seamlessly integrates a pre-trained language model (Vicuna) with a visual encoder (CLIP) using a simple linear layer, creating a robust architecture ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/gokayfem/awesome-vlm-architectures", "content": "LLaVA seamlessly integrates a pre-trained language model (Vicuna) with a visual encoder (CLIP) using a simple linear layer, creating a robust architecture ..."}
+{"idx": 7, "title": "huggingface/transformers v4.39.0 on GitHub", "date": "", "ddg_snippet": "It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved visual instruction ...", "subpage_snippet": "", "source": "newreleases.io", "link": "https://newreleases.io/project/github/huggingface/transformers/release/v4.39.0", "content": "It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved visual instruction ..."}
+{"idx": 8, "title": "Why Dubai Businesses Must Embrace AI for Employee ...", "date": "", "ddg_snippet": "New in LLaVA 1.6: Increasing the input image resolution to up to 4x more pixels , supporting 672x672 , 336x1344 , 1344x336 resolutions . Better ...", "subpage_snippet": "", "source": "www.linkedin.com", "link": "https://www.linkedin.com/pulse/why-dubai-businesses-must-embrace-ai-employee-while-keeping-ibrahim-vh5nf", "content": "New in LLaVA 1.6: Increasing the input image resolution to up to 4x more pixels , supporting 672x672 , 336x1344 , 1344x336 resolutions . Better ..."}
+{"idx": 9, "title": "AI/ML Toolkit", "date": "", "ddg_snippet": "It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved visual instruction ...", "subpage_snippet": "", "source": "gist.github.com", "link": "https://gist.github.com/0xdevalias/09a5c27702cb94f81c9fb4b7434df966", "content": "It supports three aspect ratios, up to 672x672 , 336x1344 , 1344x336 resolution . Better visual reasoning and OCR capability with an improved visual instruction ..."}
diff --git "a/data/sampled_jsons/Linear_convergence_of_Sinkhorn's_algorithm_for_generalized_static_Schr\303\266dinger_bridge_filetypepdf.jsonl" "b/data/sampled_jsons/Linear_convergence_of_Sinkhorn's_algorithm_for_generalized_static_Schr\303\266dinger_bridge_filetypepdf.jsonl"
new file mode 100644
index 0000000000000000000000000000000000000000..316cc2ee83ae83219aa3192f959b26482916a313
--- /dev/null
+++ "b/data/sampled_jsons/Linear_convergence_of_Sinkhorn's_algorithm_for_generalized_static_Schr\303\266dinger_bridge_filetypepdf.jsonl"
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "ICML Poster Linear convergence of Sinkhorn ' s algorithm for ...", "date": "", "ddg_snippet": "We establish Kantorovich duality and linear convergence of Sinkhorn ' s algorithm for the generalized SSB problem under mild conditions. Our results provide a new rigorous foundation for understanding Sinkhorn - type iterative methods in the context of large-scale generalized ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46671", "content": "We establish Kantorovich duality and linear convergence of Sinkhorn ' s algorithm for the generalized SSB problem under mild conditions. Our results provide a new rigorous foundation for understanding Sinkhorn - type iterative methods in the context of large-scale generalized ..."}
+{"idx": 1, "title": "[2201.10059] Stability of Schrödinger Potentials and Convergence of ...", "date": "", "ddg_snippet": "View a PDF of the paper titled Stability of Schr\\\"odinger Potentials and Convergence of Sinkhorn ' s Algorithm , by Marcel Nutz and 1 other authors.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2201.10059", "content": "View a PDF of the paper titled Stability of Schr\\\"odinger Potentials and Convergence of Sinkhorn ' s Algorithm , by Marcel Nutz and 1 other authors."}
+{"idx": 2, "title": "( PDF ) An Optimal Transport Approach for the Schrödinger Bridge ...", "date": "", "ddg_snippet": "Bridge problem, including convergence of Sinkhorn algorithm in the multi-marginal case.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/345547115_An_Optimal_Transport_Approach_for_the_Schrodinger_Bridge_Problem_and_Convergence_of_Sinkhorn_Algorithm", "content": "Bridge problem, including convergence of Sinkhorn algorithm in the multi-marginal case."}
+{"idx": 3, "title": "Quantitative contraction rates for Sinkhorn algorithm", "date": "", "ddg_snippet": "Exponential convergence of sinkhorn algorithm .Theorem 31 (Exponential convergence of Sinkhorn bridges ). Assume the validity of A1 and A2 and (19) for some positive constants A, B > 0. Then, for any n ≥ 1 it holds.", "subpage_snippet": "", "source": "pure.tue.nl", "link": "https://pure.tue.nl/ws/portalfiles/portal/295195017/2304.04451v1.pdf", "content": "Exponential convergence of sinkhorn algorithm .Theorem 31 (Exponential convergence of Sinkhorn bridges ). Assume the validity of A1 and A2 and (19) for some positive constants A, B > 0. Then, for any n ≥ 1 it holds."}
+{"idx": 4, "title": "On the linear convergence of the", "date": "", "ddg_snippet": "Multi-marginal Sinkhorn algorithm . Linear convergence .The aim of this short note is to give an elementary proof of linear convergence of the Sinkhorn algorithm for the entropic regularization of multi-marginal optimal transport.", "subpage_snippet": "", "source": "www.ceremade.dauphine.fr", "link": "https://www.ceremade.dauphine.fr/~carlier/linear-sinkhorn.pdf", "content": "Multi-marginal Sinkhorn algorithm . Linear convergence .The aim of this short note is to give an elementary proof of linear convergence of the Sinkhorn algorithm for the entropic regularization of multi-marginal optimal transport."}
+{"idx": 5, "title": "Optimal transport with f -divergence regularization and generalized", "date": "", "ddg_snippet": "Algorithm 2 Generalized Sinkhorn algorithm for com-puting optimal potentials f, g and optimal coupling π.An optimal transport approach for the schrödinger bridge prob-lem and convergence of sinkhorn algorithm .", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v151/terjek22a/terjek22a.pdf", "content": "Algorithm 2 Generalized Sinkhorn algorithm for com-puting optimal potentials f, g and optimal coupling π.An optimal transport approach for the schrödinger bridge prob-lem and convergence of sinkhorn algorithm ."}
+{"idx": 6, "title": "An Optimal Transport Approach for the Schrödinger Bridge Problem...", "date": "", "ddg_snippet": "3 Convergence of the Sinkhorn /IPFP Algorithm .In this section we generalize the results obtain previously for the Schrödinger problem with more than two marginals, including a proof of convergence of the Sinkhorn algorithm in the several marginals case.", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/article/10.1007/s10915-020-01325-7", "content": "3 Convergence of the Sinkhorn /IPFP Algorithm .In this section we generalize the results obtain previously for the Schrödinger problem with more than two marginals, including a proof of convergence of the Sinkhorn algorithm in the several marginals case."}
+{"idx": 7, "title": "Designing Algorithms for Entropic Optimal Transport", "date": "", "ddg_snippet": "Dynamical Schrödinger bridge problem.Based on this, classical analysis of the Sinkhorn algorithm with Hilbert’ s projective metric and Birkhoff’ s theorem [FL89] renders a linear convergence with contraction rate 1 − Θ(e− c ∞/ε), where ε is the regularisation parameter.", "subpage_snippet": "", "source": "nbviewer.org", "link": "https://nbviewer.org/github/QijiaJ/QijiaJ.github.io/blob/master/_includes/eOT_optimization_2025.pdf", "content": "Dynamical Schrödinger bridge problem.Based on this, classical analysis of the Sinkhorn algorithm with Hilbert’ s projective metric and Birkhoff’ s theorem [FL89] renders a linear convergence with contraction rate 1 − Θ(e− c ∞/ε), where ε is the regularisation parameter."}
+{"idx": 8, "title": "[ PDF ] An Optimal Transport Approach for the Schrödinger Bridge ...", "date": "", "ddg_snippet": "Convergence of the Sinkhorn algorithm when the Schrödinger problem has no solution. Aymeric BaradatE. Ventre. Schrödinger bridge is a stochastic optimal control problem to steer a given initial state density to another, subject to controlled diffusion and deadline constraints.", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/An-Optimal-Transport-Approach-for-the-Schrödinger-Marino-Gerolin/1fa3ad8de61e534b3983553a2af7d629caeb0e1f", "content": "Convergence of the Sinkhorn algorithm when the Schrödinger problem has no solution. Aymeric BaradatE. Ventre. Schrödinger bridge is a stochastic optimal control problem to steer a given initial state density to another, subject to controlled diffusion and deadline constraints."}
+{"idx": 9, "title": "Tree-Based Diffusion Schrödinger Bridge with Applications to...", "date": "", "ddg_snippet": "A general framework for tree-based static Schrödinger Bridges on discrete state -spaces was given in Haasler et al.Carlier, G. On the linear convergence of the multimarginal Sinkhorn algorithm .", "subpage_snippet": "", "source": "hal.science", "link": "https://hal.science/hal-04402356/document", "content": "A general framework for tree-based static Schrödinger Bridges on discrete state -spaces was given in Haasler et al.Carlier, G. On the linear convergence of the multimarginal Sinkhorn algorithm ."}
diff --git a/data/sampled_jsons/Machine_Learning_meets_Algebraic_Combinatorics_mHeight_3412_patterns_Section_4.2.jsonl b/data/sampled_jsons/Machine_Learning_meets_Algebraic_Combinatorics_mHeight_3412_patterns_Section_4.2.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..910f47e48a5a5f30e1ddc7764187d1afd76da63d
--- /dev/null
+++ b/data/sampled_jsons/Machine_Learning_meets_Algebraic_Combinatorics_mHeight_3412_patterns_Section_4.2.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Machine Learning Meets Algebraic Combinatorics", "date": "", "ddg_snippet": "The mHeight of σ is then the minimum height over all 3412 patterns in σ. We provide datasets for n = 8, 9, 10. ML task: Predict the mHeight of a permutation.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/attachment?id=tlniJJFUW2&name=pdf", "content": "The mHeight of σ is then the minimum height over all 3412 patterns in σ. We provide datasets for n = 8, 9, 10. ML task: Predict the mHeight of a permutation."}
+{"idx": 1, "title": "Machine Learning meets Algebraic Combinatorics", "date": "", "ddg_snippet": "by H Chau — In this paper we introduced Algebraic Combinatorics Dataset Repository, a collection of datasets structured for machine learning and designed to facilitate the ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=KQ1gI5qzAf", "content": "by H Chau — In this paper we introduced Algebraic Combinatorics Dataset Repository, a collection of datasets structured for machine learning and designed to facilitate the ..."}
+{"idx": 2, "title": "Machine Learning meets Algebraic Combinatorics", "date": "", "ddg_snippet": "by H Chau · 2025 · Cited by 3 — The mHeight of σ is then the minimum height over all 3412 patterns in σ. We provide datasets for n = 8, 9, 10. ML task: Predict the mHeight ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2503.06366", "content": "by H Chau · 2025 · Cited by 3 — The mHeight of σ is then the minimum height over all 3412 patterns in σ. We provide datasets for n = 8, 9, 10. ML task: Predict the mHeight ..."}
+{"idx": 3, "title": "Machine Learning meets Algebraic Combinatorics: A Suite of ...", "date": "", "ddg_snippet": "To address this, we introduce a new collection of datasets, the Algebraic Combinatorics Dataset Repository (ACD Repo), representing either foundational results or open problems in algebraic combinatorics , a subfield of mathematics that studies discrete structures arising from abstract algebra .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.06366v1", "content": "To address this, we introduce a new collection of datasets, the Algebraic Combinatorics Dataset Repository (ACD Repo), representing either foundational results or open problems in algebraic combinatorics , a subfield of mathematics that studies discrete structures arising from abstract algebra ."}
+{"idx": 4, "title": "Probability and Machine Learning in Combinatorial Commutative ...", "date": "", "ddg_snippet": "elds, un-less explicitly stated otherwise. Although algebraic closure and characteristic are generally crucial to proofs in commutative algebra , the combinatorial nature of monomial ideals ( Sec-tion 1.2.1) often allows us to skirt these considerations.", "subpage_snippet": "", "source": "www.cpp.edu", "link": "https://www.cpp.edu/faculty/lsilverstein/dissertation.pdf", "content": "elds, un-less explicitly stated otherwise. Although algebraic closure and characteristic are generally crucial to proofs in commutative algebra , the combinatorial nature of monomial ideals ( Sec-tion 1.2.1) often allows us to skirt these considerations."}
+{"idx": 5, "title": "Machine Checked Proofs and Programs in Algebraic Combinatorics", "date": "", "ddg_snippet": "A specific feature of algebraic combinatorics is the constant interplay between algorithms and algebraic con-structions: algorithms are not only in computations, but also are key ingredients in definitions and proofs.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2412.04864", "content": "A specific feature of algebraic combinatorics is the constant interplay between algorithms and algebraic con-structions: algorithms are not only in computations, but also are key ingredients in definitions and proofs."}
+{"idx": 6, "title": "An Introduction to Algebraic Combinatorics - LMU", "date": "", "ddg_snippet": "Abstract. This is an introduction to algebraic combinatorics , writ- ten for a quarter-long graduate course. It starts with a rigorous in- troduction to formal power series with some combinatorial applica- tions, then discusses integer partitions (proving Jacobi’s triple prod- uct identity), permutations (Lehmer codes, cycles) and subtractive methods (alternating sums, cancellations and ...", "subpage_snippet": "", "source": "www.cip.ifi.lmu.de", "link": "https://www.cip.ifi.lmu.de/~grinberg/t/21s/lecs.pdf", "content": "Abstract. This is an introduction to algebraic combinatorics , writ- ten for a quarter-long graduate course. It starts with a rigorous in- troduction to formal power series with some combinatorial applica- tions, then discusses integer partitions (proving Jacobi’s triple prod- uct identity), permutations (Lehmer codes, cycles) and subtractive methods (alternating sums, cancellations and ..."}
+{"idx": 7, "title": "Lecture Notes on Algebraic Combinatorics", "date": "", "ddg_snippet": "These lecture notes began as my notes from Vic Reiner’s Algebraic Combinatorics course at the University of Minnesota in Fall 2003. I currently use them for graduate courses at the University of Kansas.", "subpage_snippet": "", "source": "jlmartin.ku.edu", "link": "https://jlmartin.ku.edu/~jlmartin/LectureNotes.pdf", "content": "These lecture notes began as my notes from Vic Reiner’s Algebraic Combinatorics course at the University of Minnesota in Fall 2003. I currently use them for graduate courses at the University of Kansas."}
+{"idx": 8, "title": "Machine Learning meets Algebraic Combinatorics: A Suite of ...", "date": "", "ddg_snippet": "In this paper we introduce the Algebraic Combinatorics Dataset Repository, a collection of research-level mathemat-ics datasets structured for machine learning and designed to accelerate mathematical discovery.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2503.06366v1", "content": "In this paper we introduce the Algebraic Combinatorics Dataset Repository, a collection of research-level mathemat-ics datasets structured for machine learning and designed to accelerate mathematical discovery."}
+{"idx": 9, "title": "(PDF) Machine Learning Algebraic Geometry for Physics", "date": "", "ddg_snippet": "Apr 21, 2022 · PDF | We review some recent applications of machine learning to algebraic geometry and physics. Since problems in algebraic geometry can typically be... | Find, read and cite all the research you ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/360166117_Machine_Learning_Algebraic_Geometry_for_Physics", "content": "Apr 21, 2022 · PDF | We review some recent applications of machine learning to algebraic geometry and physics. Since problems in algebraic geometry can typically be... | Find, read and cite all the research you ..."}
diff --git a/data/sampled_jsons/Machine_Unlearning_A_Survey.jsonl b/data/sampled_jsons/Machine_Unlearning_A_Survey.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..0ea12763aae06d3ef1173c16b383c7a55215257e
--- /dev/null
+++ b/data/sampled_jsons/Machine_Unlearning_A_Survey.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Machine - Wikipedia", "date": "", "ddg_snippet": "A machine is a thermodynamic system that uses power to apply forces and control movement to perform an action. The term is commonly applied to artificial devices, such as those employing engines or motors, but also to natural biological macromolecules, such as molecular machines .", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Machine", "content": "A machine is a thermodynamic system that uses power to apply forces and control movement to perform an action. The term is commonly applied to artificial devices, such as those employing engines or motors, but also to natural biological macromolecules, such as molecular machines ."}
+{"idx": 1, "title": "MACHINE Definition & Meaning | Dictionary .com", "date": "", "ddg_snippet": "Machine definition: an apparatus consisting of interrelated parts with separate functions, used in the performance of some kind of work.. See examples of MACHINE used in a sentence.", "subpage_snippet": "", "source": "www.dictionary.com", "link": "https://www.dictionary.com/browse/machine", "content": "Machine definition: an apparatus consisting of interrelated parts with separate functions, used in the performance of some kind of work.. See examples of MACHINE used in a sentence."}
+{"idx": 2, "title": "MACHINE Definition & Meaning - Merriam-Webster", "date": "", "ddg_snippet": "The meaning of MACHINE is a mechanically, electrically, or electronically operated device for performing a task. How to use machine in a sentence.", "subpage_snippet": "", "source": "www.merriam-webster.com", "link": "https://www.merriam-webster.com/dictionary/machine", "content": "The meaning of MACHINE is a mechanically, electrically, or electronically operated device for performing a task. How to use machine in a sentence."}
+{"idx": 3, "title": "Machine | Definition, Mechanisms & Efficiency | Britannica", "date": "", "ddg_snippet": "machine , device, having a unique purpose, that augments or replaces human or animal effort for the accomplishment of physical tasks.", "subpage_snippet": "", "source": "www.britannica.com", "link": "https://www.britannica.com/technology/machine", "content": "machine , device, having a unique purpose, that augments or replaces human or animal effort for the accomplishment of physical tasks."}
+{"idx": 4, "title": "MACHINE | English meaning - Cambridge Dictionary", "date": "", "ddg_snippet": "MACHINE definition: 1. a piece of equipment with several moving parts that uses power to do a particular type of work…. Learn more.", "subpage_snippet": "", "source": "dictionary.cambridge.org", "link": "https://dictionary.cambridge.org/dictionary/english/machine", "content": "MACHINE definition: 1. a piece of equipment with several moving parts that uses power to do a particular type of work…. Learn more."}
+{"idx": 5, "title": "machine , n. meanings, etymology and more | Oxford English...", "date": "", "ddg_snippet": "machine , n. meanings, etymology, pronunciation and more in the Oxford English Dictionary", "subpage_snippet": "", "source": "www.oed.com", "link": "https://www.oed.com/dictionary/machine_n", "content": "machine , n. meanings, etymology, pronunciation and more in the Oxford English Dictionary"}
+{"idx": 6, "title": "What Is A Machine ? Its Types and How it Works - Mech Lesson", "date": "", "ddg_snippet": "A machine is a mechanical device that uses power to apply force and control motion to perform work efficiently . Machines range from simple tools like pulleys and levers to complex systems like engines, robots, and manufacturing equipment.", "subpage_snippet": "", "source": "mechlesson.com", "link": "https://mechlesson.com/machine/", "content": "A machine is a mechanical device that uses power to apply force and control motion to perform work efficiently . Machines range from simple tools like pulleys and levers to complex systems like engines, robots, and manufacturing equipment."}
+{"idx": 7, "title": "machine - Wiktionary, the free dictionary", "date": "", "ddg_snippet": "3 days ago · (figuratively) A person or organisation that seemingly acts like a machine , being particularly efficient, single-minded, or unemotional. Bruce Campbell was a \"demon-killing machine \" because he made quick work of killing demons. The government has become a money-making machine .", "subpage_snippet": "", "source": "en.wiktionary.org", "link": "https://en.wiktionary.org/wiki/machine", "content": "3 days ago · (figuratively) A person or organisation that seemingly acts like a machine , being particularly efficient, single-minded, or unemotional. Bruce Campbell was a \"demon-killing machine \" because he made quick work of killing demons. The government has become a money-making machine ."}
+{"idx": 8, "title": "Machine - definition of machine by The Free Dictionary", "date": "", "ddg_snippet": "Of, relating to, or felt to resemble a machine : machine repairs; machine politics.", "subpage_snippet": "", "source": "www.thefreedictionary.com", "link": "https://www.thefreedictionary.com/machine", "content": "Of, relating to, or felt to resemble a machine : machine repairs; machine politics."}
+{"idx": 9, "title": "The Smashing Machine | Official Trailer HD | A24 - YouTube", "date": "", "ddg_snippet": "Special Blu-ray & 4K editions of Danny & Michael Philippou's sinister tale of domestic and occult horror. Special features include a commentary with the directors, one deleted scene, and an...", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=aRpnP3LZ99g", "content": "Special Blu-ray & 4K editions of Danny & Michael Philippou's sinister tale of domestic and occult horror. Special features include a commentary with the directors, one deleted scene, and an..."}
diff --git a/data/sampled_jsons/Medusa_Cai_et_al_2024_year_2024.jsonl b/data/sampled_jsons/Medusa_Cai_et_al_2024_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..88705aa753c1683df58b2c4fc89491b638fab076
--- /dev/null
+++ b/data/sampled_jsons/Medusa_Cai_et_al_2024_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Whisper-Medusa: aiOla Achieved a 1.5X Speedup Over OpenAI", "date": "", "ddg_snippet": "Our Medusa model builds on top of ... The results indicate that our Medusa architecture achieves a consistent speedup across all sequence lengths.", "subpage_snippet": "", "source": "aiola.ai", "link": "https://aiola.ai/the-tech-talk/medusa-architect/", "content": "Our Medusa model builds on top of ... The results indicate that our Medusa architecture achieves a consistent speedup across all sequence lengths."}
+{"idx": 1, "title": "Most Influential ArXiv (Machine Learning) Papers (2024-10)", "date": "", "ddg_snippet": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads IF:4 Related Papers Related Patents Related Grants Related Venues ...", "subpage_snippet": "", "source": "www.paperdigest.org", "link": "https://www.paperdigest.org/2024/10/most-influential-arxiv-machine-learning-papers-2024-10/", "content": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads IF:4 Related Papers Related Patents Related Grants Related Venues ..."}
+{"idx": 2, "title": "Reward-Shifted Speculative Sampling Is An Efficient Test-Time", "date": "", "ddg_snippet": "To mitigate these issues, test-time alignment (Khanov et al ., 2024 ; Li et al ., 2024a ; Qiu et al ., 2024 ) has emerged as a training-free ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.15044v1", "content": "To mitigate these issues, test-time alignment (Khanov et al ., 2024 ; Li et al ., 2024a ; Qiu et al ., 2024 ) has emerged as a training-free ..."}
+{"idx": 3, "title": "A1: Asynchronous Test-Time Scaling via Conformal Prediction", "date": "", "ddg_snippet": "Speculative decoding (Li et al ., 2024a ; Leviathan et al ., 2023 ; Kim et al ., 2023 ; Pan et al ., 2025b ; Yang et al ., 2025 ) represents a ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.15148v1", "content": "Speculative decoding (Li et al ., 2024a ; Leviathan et al ., 2023 ; Kim et al ., 2023 ; Pan et al ., 2025b ; Yang et al ., 2025 ) represents a ..."}
+{"idx": 4, "title": "Alignment-Augmented Speculative Decoding with Alignment", "date": "", "ddg_snippet": "Approaches such as EAGLE (Li et al ., 2024 ) strengthen the alignment between the draft and target models through alignment training, causing ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.13204v2", "content": "Approaches such as EAGLE (Li et al ., 2024 ) strengthen the alignment between the draft and target models through alignment training, causing ..."}
+{"idx": 5, "title": "COLING 2025 Tutorial: Speculative Decoding for Efficient LLM", "date": "", "ddg_snippet": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads ( Cai et al ., 2023) ... Speculative Decoding (Sun et al ., 2024 ...", "subpage_snippet": "", "source": "speculative-decoding.github.io", "link": "https://speculative-decoding.github.io/", "content": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads ( Cai et al ., 2023) ... Speculative Decoding (Sun et al ., 2024 ..."}
+{"idx": 6, "title": "transcript", "date": "", "ddg_snippet": "Medusa ’ s authors note it “generate[ s ] consistent responses with the same distribution as the original model,” but “cannot further enhance ...", "subpage_snippet": "", "source": "notes.aarnphm.xyz", "link": "https://notes.aarnphm.xyz/lectures/41/notes", "content": "Medusa ’ s authors note it “generate[ s ] consistent responses with the same distribution as the original model,” but “cannot further enhance ..."}
+{"idx": 7, "title": "Most Influential ArXiv (Machine Learning) Papers (2025-03", "date": "", "ddg_snippet": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads IF:4 Related Papers Related Patents Related Grants Related Venues ...", "subpage_snippet": "", "source": "resources.paperdigest.org", "link": "https://resources.paperdigest.org/2025/03/most-influential-arxiv-machine-learning-papers-2025-03-version/", "content": "Medusa : Simple LLM Inference Acceleration Framework with Multiple Decoding Heads IF:4 Related Papers Related Patents Related Grants Related Venues ..."}
+{"idx": 8, "title": "Pre-Training Curriculum for Multi-Token Prediction in Language", "date": "", "ddg_snippet": "So far, the previous work prioritized mid- and large-sized models with at least 7B parameters, since Gloeckle et al .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.22757v1", "content": "So far, the previous work prioritized mid- and large-sized models with at least 7B parameters, since Gloeckle et al ."}
+{"idx": 9, "title": "SpeLLM: Input Tokens, Output Chars", "date": "", "ddg_snippet": "However, it also raises the costs of generating each token, as the vocabulary embedding table is used both for projecting the output token at the ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.16323v1", "content": "However, it also raises the costs of generating each token, as the vocabulary embedding table is used both for projecting the output token at the ..."}
diff --git a/data/sampled_jsons/Near-Optimal_Online_Learning_for_Multi-Agent_Submodular_Coordination_Section_5_adversarial_target_be_year_2023.jsonl b/data/sampled_jsons/Near-Optimal_Online_Learning_for_Multi-Agent_Submodular_Coordination_Section_5_adversarial_target_be_year_2023.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..356f2c92c40d233d8b6c8e6d615eafaf4c0b7744
--- /dev/null
+++ b/data/sampled_jsons/Near-Optimal_Online_Learning_for_Multi-Agent_Submodular_Coordination_Section_5_adversarial_target_be_year_2023.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Udemy Online Classes - A Leading Learning Destination Ad Viewing ads is privacy protected by DuckDuckGo. Ad clicks are managed by Microsoft's ad network ( more info ).", "date": "", "ddg_snippet": "Study any topic when it suits you. Explore thousands of high-quality courses. Join millions of learners from around the world already learning on Udemy.", "subpage_snippet": "", "source": "duckduckgo.com", "link": "https://duckduckgo.com/y.js?ad_domain=udemy.com&ad_provider=bingv7aa&ad_type=txad&click_metadata=UCPIy_GUbb855RBojMOGC-uDwNBAeG4GMIHfk2L4sl85Cx5p1fXo8HKb_3P5jauZ3-pJUYwqf3qo8ONl7Ogkdb193JPW1at226vc86WzeCb9Dalo8Gzjtv5NyE3uFNoI.rUoH_nWwp_ThukOuUG_O-w&rut=c877445c8dd9084fc6f0ada3e6f289556cbf2ec0041983d1dab57b709e72807b&u3=https://www.bing.com/aclick?ld=e8iAo_DWTNRT-9FuJoZ1BJ4DVUCUxNFBD0Hi3az7bYOCR9sINrU2ZvFWHS-2tlfQ46G4FlJqXRcechjg3bS4EjJm62sfPanrh3f7g_kDdL1rfn12QdP-EBMoBSzQErZILVtO1fIQBp1yJReVJLRlaksrWAEqEr5jri8u4cqj6zqXc1ik-h36HIAkRAcRwj9dmp3Xxo2A&u=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&rlid=876974a832441c07390e1c7aaf273e21&vqd=4-29136137857996143131188833869597288217&iurl={1}IG=86D1EFE663B74D35B8DE312F7109216E&CID=0A565CB0084D62233D344AC0092863EF&ID=DevEx,5035.1", "content": "Study any topic when it suits you. Explore thousands of high-quality courses. Join millions of learners from around the world already learning on Udemy."}
+{"idx": 1, "title": "(PDF) Near - Optimal Online Learning for Multi - Agent Submodular ...", "date": "", "ddg_snippet": "Near - optimal online learning for multi -. Agent submodular coordination : tight ap-. Proximation and communication efficiency.vehicles (UAVs), for multi - target tracking (Zhou et al.,2018;Corah & Michael,2021) as depicted.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/388847678_Near-Optimal_Online_Learning_for_Multi-Agent_Submodular_Coordination_Tight_Approximation_and_Communication_Efficiency", "content": "Near - optimal online learning for multi -. Agent submodular coordination : tight ap-. Proximation and communication efficiency.vehicles (UAVs), for multi - target tracking (Zhou et al.,2018;Corah & Michael,2021) as depicted."}
+{"idx": 2, "title": "Near-Optimal Online Learning for Multi-Agent Submodular ...", "date": "", "ddg_snippet": "7 Feb 2025 — Near-Optimal Online Learning for Multi-Agent Submodular Coordination ... Adversarial' target escapes at a speed of 15 units/s for one ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.05028v1", "content": "7 Feb 2025 — Near-Optimal Online Learning for Multi-Agent Submodular Coordination ... Adversarial' target escapes at a speed of 15 units/s for one ..."}
+{"idx": 3, "title": "Performance-Aware Self-Configurable Multi-Agent Networks: A", "date": "", "ddg_snippet": "In the future, distributed teams of agents will be coordinating via agent -to- agent communication to execute tasks such as target tracking [ 1 ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2409.01411v1", "content": "In the future, distributed teams of agents will be coordinating via agent -to- agent communication to execute tasks such as target tracking [ 1 ..."}
+{"idx": 4, "title": "Inverse Submodular Maximization with Application to", "date": "", "ddg_snippet": "We consider a new type of inverse combinatorial optimization, Inverse Submodular Maximization (ISM), for human-in-the-loop multi -robot coordination .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2403.10991v1", "content": "We consider a new type of inverse combinatorial optimization, Inverse Submodular Maximization (ISM), for human-in-the-loop multi -robot coordination ."}
+{"idx": 5, "title": "Diffusion Models for Influence Maximization on Temporal", "date": "", "ddg_snippet": "... for diverse applications, including enhancing public health outreach, mitigating the spread of harmful rumors, and optimizing marketing campaigns.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.22589v1", "content": "... for diverse applications, including enhancing public health outreach, mitigating the spread of harmful rumors, and optimizing marketing campaigns."}
+{"idx": 6, "title": "N ear -o ptimal", "date": "", "ddg_snippet": "Near - optimal online learning for multi - agent submodular coordination : tight ap-proximation and communication efficiency. Definition 1. For a set function f : 2 V → R+, we define its multi-linear extension as.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=i8dYPGdB1C", "content": "Near - optimal online learning for multi - agent submodular coordination : tight ap-proximation and communication efficiency. Definition 1. For a set function f : 2 V → R+, we define its multi-linear extension as."}
+{"idx": 7, "title": "Combatting Dimensional Collapse in LLM Pre-Training Data via", "date": "", "ddg_snippet": "... analyze its approximation to the optimal solution under the formulation of γ 𝛾 \\gamma italic_γ -weakly submodular optimization problem ( Section ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.20644v1", "content": "... analyze its approximation to the optimal solution under the formulation of γ 𝛾 \\gamma italic_γ -weakly submodular optimization problem ( Section ..."}
+{"idx": 8, "title": "Efficiency of Coordinate Descent Methods on Huge-Scale", "date": "", "ddg_snippet": "Tseng, On the convergence of the coordinate descent method for convex differentiable minimization, J. ... Nesterov, A method for unconstrained convex ...", "subpage_snippet": "", "source": "epubs.siam.org", "link": "https://epubs.siam.org/doi/abs/10.1137/100802001?cookieSet=1", "content": "Tseng, On the convergence of the coordinate descent method for convex differentiable minimization, J. ... Nesterov, A method for unconstrained convex ..."}
+{"idx": 9, "title": "Efficiency of Coordinate Descent Methods on Huge-Scale", "date": "", "ddg_snippet": "Tseng, On the convergence of the coordinate descent method for convex differentiable minimization, J. ... Nesterov, A method for unconstrained convex ...", "subpage_snippet": "", "source": "epubs.siam.org", "link": "https://epubs.siam.org/doi/abs/10.1137/100802001", "content": "Tseng, On the convergence of the coordinate descent method for convex differentiable minimization, J. ... Nesterov, A method for unconstrained convex ..."}
diff --git a/data/sampled_jsons/OmniBench_paper_Section_5.1_experimental_setup_year_2024.jsonl b/data/sampled_jsons/OmniBench_paper_Section_5.1_experimental_setup_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..55f65141e802fa98355e81a80253d2781219a5f1
--- /dev/null
+++ b/data/sampled_jsons/OmniBench_paper_Section_5.1_experimental_setup_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "What Limits Virtual Agent Application? OmniBench: A Scalable Multi ...", "date": "", "ddg_snippet": "In this section , we first introduce the experimental setup ( Sec-tion 5.1 ). Then, we comprehensively compare the differences in capabilities across various models on OmniBench , along with several key findings ( Section 5.2).", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2506.08933", "content": "In this section , we first introduce the experimental setup ( Sec-tion 5.1 ). Then, we comprehensively compare the differences in capabilities across various models on OmniBench , along with several key findings ( Section 5.2)."}
+{"idx": 1, "title": "OmniBench", "date": "", "ddg_snippet": "OmniBench spans five fundamental types of task complexity to construct 10 evaluation dimensions (see the main figure). Test tasks across these dimensions are categorized based on combinations of complexity types. For example, a long-range planning test task typically exhibits higher dependency complexity and hierarchical complexity.", "subpage_snippet": "", "source": "omni-bench.github.io", "link": "https://omni-bench.github.io/", "content": "OmniBench spans five fundamental types of task complexity to construct 10 evaluation dimensions (see the main figure). Test tasks across these dimensions are categorized based on combinations of complexity types. For example, a long-range planning test task typically exhibits higher dependency complexity and hierarchical complexity."}
+{"idx": 2, "title": "OmniBench: Towards The Future of Universal Omni-Language Models", "date": "", "ddg_snippet": "These results underscore the importance of OmniBench as a tool for identifying areas of improvement and guiding research in multimodal systems. In the following sections , we detail the data collection protocol of OmniBench , present our evaluation results on current state-of-the-art MLLMs, and discuss the implications of our findings for the future of research and development. Through OmniBench ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2409.15272v1", "content": "These results underscore the importance of OmniBench as a tool for identifying areas of improvement and guiding research in multimodal systems. In the following sections , we detail the data collection protocol of OmniBench , present our evaluation results on current state-of-the-art MLLMs, and discuss the implications of our findings for the future of research and development. Through OmniBench ..."}
+{"idx": 3, "title": "OmniBench: Towards The Future of Universal Omni-Language Models", "date": "", "ddg_snippet": "Recent advancements in multimodal large language models (MLLMs) have focused on integrating multiple modalities, yet their ability to simultaneously process and reason across different inputs remains underexplored. We introduce OmniBench , a novel benchmark designed to evaluate models' ability to recognize, interpret, and reason across visual, acoustic, and textual inputs simultaneously. We ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2409.15272", "content": "Recent advancements in multimodal large language models (MLLMs) have focused on integrating multiple modalities, yet their ability to simultaneously process and reason across different inputs remains underexplored. We introduce OmniBench , a novel benchmark designed to evaluate models' ability to recognize, interpret, and reason across visual, acoustic, and textual inputs simultaneously. We ..."}
+{"idx": 4, "title": "PDF 7 Detailed Experimental Setup", "date": "", "ddg_snippet": "This document provides additional details, analysis, and experimental results. We begin by discussing the detailed experimental setup and implementation of the methods in Section 7. Then, we provide additional empirical experiments in Section 8. Finally, we discuss the limitation of our work in Section 9.", "subpage_snippet": "", "source": "papers.nips.cc", "link": "https://papers.nips.cc/paper_files/paper/2022/file/fa0126bb7ebad258bf4ffdbbac2dd787-Supplemental-Conference.pdf", "content": "This document provides additional details, analysis, and experimental results. We begin by discussing the detailed experimental setup and implementation of the methods in Section 7. Then, we provide additional empirical experiments in Section 8. Finally, we discuss the limitation of our work in Section 9."}
+{"idx": 5, "title": "OmniBench: Towards The Future of Universal Omni-Language Models", "date": "", "ddg_snippet": "This paper introduces OmniBench , a crucial step towards developing truly multimodal AI. By rigorously evaluating models on their ability to integrate visual, acoustic, and textual information, OmniBench exposes critical limitations in current approaches and highlights the need for dedicated research in tri-modal reasoning.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2409.15272v4", "content": "This paper introduces OmniBench , a crucial step towards developing truly multimodal AI. By rigorously evaluating models on their ability to integrate visual, acoustic, and textual information, OmniBench exposes critical limitations in current approaches and highlights the need for dedicated research in tri-modal reasoning."}
+{"idx": 6, "title": "OmniR: Evaluating Omni-modality Language Models on Reasoning across ...", "date": "", "ddg_snippet": "Video/Audio/Image Evaluation benchmarks. Omnibench specifically aimed at evaluating OLMs' tri-modal, i.e., text, vision, and audio, processing capabilities with human-annotated tasks. Compared to it, OmnixR emphasizes the omni-modality reasoning evaluations with both human-annotated realistic set and scalable synthetic set.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.12219", "content": "Video/Audio/Image Evaluation benchmarks. Omnibench specifically aimed at evaluating OLMs' tri-modal, i.e., text, vision, and audio, processing capabilities with human-annotated tasks. Compared to it, OmnixR emphasizes the omni-modality reasoning evaluations with both human-annotated realistic set and scalable synthetic set."}
+{"idx": 7, "title": "GitHub - opendatalab/OmniDocBench: [CVPR 2025] A Comprehensive ...", "date": "", "ddg_snippet": "OmniDocBench is a benchmark for evaluating diverse document parsing in real-world scenarios, featuring the following characteristics: Diverse Document Types: This benchmark includes 981 PDF pages, covering 9 document types, 4 layout types, and 3 language types. It encompasses a wide range of content, including academic papers , financial reports, newspapers, textbooks, and handwritten notes ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/opendatalab/OmniDocBench", "content": "OmniDocBench is a benchmark for evaluating diverse document parsing in real-world scenarios, featuring the following characteristics: Diverse Document Types: This benchmark includes 981 PDF pages, covering 9 document types, 4 layout types, and 3 language types. It encompasses a wide range of content, including academic papers , financial reports, newspapers, textbooks, and handwritten notes ..."}
+{"idx": 8, "title": "(PDF) Baichuan-Omni Technical Report - ResearchGate", "date": "", "ddg_snippet": "In this paper , we introduce Baichuan-Omni, the first open-source 7B Multimodal Large Language Model (MLLM) adept at concurrently processing and analyzing modalities of image, video, audio, and ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/384887170_Baichuan-Omni_Technical_Report", "content": "In this paper , we introduce Baichuan-Omni, the first open-source 7B Multimodal Large Language Model (MLLM) adept at concurrently processing and analyzing modalities of image, video, audio, and ..."}
+{"idx": 9, "title": "PDF O R: E OMNI MODALITY LANGUAGE M REASONING ACROSS MODALITIES - OpenReview", "date": "", "ddg_snippet": "ABSTRACT We introduce Omni×R, an evaluation suite designed to benchmark state-of-the-art Omni-modality Language Models (OLMs), such as GPT-4o and Gemini. Eval-uating OLMs, which integrate multiple modalities such as text, vision, and audio, presents unique challenges. Particularly, the user message might often consist of multiple modalities, such that OLMs have to establish holistic ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf/a8f5792b167b7d40c9acb6a680d0c8c51c2d13a0.pdf", "content": "ABSTRACT We introduce Omni×R, an evaluation suite designed to benchmark state-of-the-art Omni-modality Language Models (OLMs), such as GPT-4o and Gemini. Eval-uating OLMs, which integrate multiple modalities such as text, vision, and audio, presents unique challenges. Particularly, the user message might often consist of multiple modalities, such that OLMs have to establish holistic ..."}
diff --git a/data/sampled_jsons/On_a_Connection_Between_Imitation_Learning_and_RLHF_LSIF.jsonl b/data/sampled_jsons/On_a_Connection_Between_Imitation_Learning_and_RLHF_LSIF.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..4041b1f657cf9070fb999368d7b0b4a65de1b971
--- /dev/null
+++ b/data/sampled_jsons/On_a_Connection_Between_Imitation_Learning_and_RLHF_LSIF.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "On a Connection Between Imitation Learning and RLHF", "date": "", "ddg_snippet": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback RLHF and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution. Building on this connection , we propose DIL, a ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2503.05079", "content": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback RLHF and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution. Building on this connection , we propose DIL, a ..."}
+{"idx": 1, "title": "Connection Between Imitation Learning and RLHF | Cong's Log", "date": "", "ddg_snippet": "The paper, Xiao, Teng, et al. On a Connection Between Imitation Learning and RLHF . arXiv:2503.05079 1 also expresses that DPO is a form of imitation learning .", "subpage_snippet": "", "source": "congchan.github.io", "link": "https://congchan.github.io/posts/connection-between-imitation-learning-and-rlhf/", "content": "The paper, Xiao, Teng, et al. On a Connection Between Imitation Learning and RLHF . arXiv:2503.05079 1 also expresses that DPO is a form of imitation learning ."}
+{"idx": 2, "title": "On a Connection Between Imitation Learningand RLHF", "date": "", "ddg_snippet": "RLHF Step 2: Reinforcement Learning Reinforcement learning objective: KL-regularized reward maximization Reward model KL regularization (proxy human preference) (closed to reference model) Complexity and poor stability due to the high variance in optimization", "subpage_snippet": "", "source": "aitime-lundao.oss-cn-beijing.aliyuncs.com", "link": "https://aitime-lundao.oss-cn-beijing.aliyuncs.com/AitimeReport/20250312/1741781794566", "content": "RLHF Step 2: Reinforcement Learning Reinforcement learning objective: KL-regularized reward maximization Reward model KL regularization (proxy human preference) (closed to reference model) Complexity and poor stability due to the high variance in optimization"}
+{"idx": 3, "title": "On a Connection Between Imitation Learning and RLHF | AI Research Paper ...", "date": "", "ddg_snippet": "This paper makes a significant contribution by establishing a formal connection between imitation learning and RLHF , offering a simpler alternative for aligning language models with human preferences. The ILHF approach demonstrates that we can achieve results comparable to RLHF through careful selection and imitation of high-quality examples.", "subpage_snippet": "", "source": "www.aimodels.fyi", "link": "https://www.aimodels.fyi/papers/arxiv/connection-between-imitation-learning-rlhf", "content": "This paper makes a significant contribution by establishing a formal connection between imitation learning and RLHF , offering a simpler alternative for aligning language models with human preferences. The ILHF approach demonstrates that we can achieve results comparable to RLHF through careful selection and imitation of high-quality examples."}
+{"idx": 4, "title": "PDF On a Connection Between Imitation Learning and Rlhf", "date": "", "ddg_snippet": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection be-tween reinforcement learning from human feedback ( RLHF ) and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution.", "subpage_snippet": "", "source": "proceedings.iclr.cc", "link": "https://proceedings.iclr.cc/paper_files/paper/2025/file/acf4a08f67724e9d2de34099f57a9c25-Paper-Conference.pdf", "content": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection be-tween reinforcement learning from human feedback ( RLHF ) and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution."}
+{"idx": 5, "title": "On a Connection Between Imitation Learning and RLHF | alphaXiv", "date": "", "ddg_snippet": "View recent discussion. Abstract: This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback RLHF and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution. Building on this ...", "subpage_snippet": "", "source": "www.alphaxiv.org", "link": "https://www.alphaxiv.org/overview/2503.05079v1", "content": "View recent discussion. Abstract: This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback RLHF and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution. Building on this ..."}
+{"idx": 6, "title": "dblp: On a Connection Between Imitation Learning and RLHF.", "date": "", "ddg_snippet": "Bibliographic details on On a Connection Between Imitation Learning and RLHF .", "subpage_snippet": "", "source": "dblp.org", "link": "https://dblp.org/rec/conf/iclr/XiaoYLCH25", "content": "Bibliographic details on On a Connection Between Imitation Learning and RLHF ."}
+{"idx": 7, "title": "On a Connection Between Imitation Learning and RLHF", "date": "", "ddg_snippet": "DIL is proposed, a principled framework that directly optimizes the imitation learning objective and provides a unified imitation learning perspective on alignment, encompassing existing alignment algorithms as special cases while naturally introducing new variants. This work studies the alignment of large language models with preference data from an imitation learning perspective. We ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/On-a-Connection-Between-Imitation-Learning-and-RLHF-Xiao-Yuan/9b4ecc297389ca6753817ce3d3dfb3057c34ae76", "content": "DIL is proposed, a principled framework that directly optimizes the imitation learning objective and provides a unified imitation learning perspective on alignment, encompassing existing alignment algorithms as special cases while naturally introducing new variants. This work studies the alignment of large language models with preference data from an imitation learning perspective. We ..."}
+{"idx": 8, "title": "PDF Lecture 8: Imitation Learning and RLHF - web.stanford.edu", "date": "", "ddg_snippet": "Imitation Learning Summary Imitation learning can greatly reduce the amount of data need to learn a good policy Challenges remain and one exciting area is combining inverse RL / learning from demonstration and online reinforcement learning For a look into some of the theory between imitation learning and RL, see Sun, Venkatraman, Gordon, Boots, Bagnell (ICML 2017)", "subpage_snippet": "", "source": "web.stanford.edu", "link": "https://web.stanford.edu/class/cs234/CS234Spr2024/slides/lecture8post.pdf", "content": "Imitation Learning Summary Imitation learning can greatly reduce the amount of data need to learn a good policy Challenges remain and one exciting area is combining inverse RL / learning from demonstration and online reinforcement learning For a look into some of the theory between imitation learning and RL, see Sun, Venkatraman, Gordon, Boots, Bagnell (ICML 2017)"}
+{"idx": 9, "title": "On a Connection Between Imitation Learning and RLHF", "date": "", "ddg_snippet": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback ( RLHF ) and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.05079v1", "content": "This work studies the alignment of large language models with preference data from an imitation learning perspective. We establish a close theoretical connection between reinforcement learning from human feedback ( RLHF ) and imitation learning (IL), revealing that RLHF implicitly performs imitation learning on the preference data distribution."}
diff --git a/data/sampled_jsons/Origin_Identification_Text-Guided_Image-to-Image_Diffusion_Models_Section_G_IP-Adapter_CLIP.jsonl b/data/sampled_jsons/Origin_Identification_Text-Guided_Image-to-Image_Diffusion_Models_Section_G_IP-Adapter_CLIP.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..bf912641521be82e1f954890199a80e5134e5d12
--- /dev/null
+++ b/data/sampled_jsons/Origin_Identification_Text-Guided_Image-to-Image_Diffusion_Models_Section_G_IP-Adapter_CLIP.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Origin Identification for Text-Guided Image-to-Image Diffusion Models", "date": "", "ddg_snippet": "Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for spreading misinformation, infringing on copyrights, and evading content tracing.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2501.02376", "content": "Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for spreading misinformation, infringing on copyrights, and evading content tracing."}
+{"idx": 1, "title": "[ICML 2025] The official implementation of \"Origin Identification for ...", "date": "", "ddg_snippet": "ID2 [ICML 2025] The official implementation of \" Origin Identification for Text-Guided Image-to-Image Diffusion Models \".", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/WangWenhao0716/ID2", "content": "ID2 [ICML 2025] The official implementation of \" Origin Identification for Text-Guided Image-to-Image Diffusion Models \"."}
+{"idx": 2, "title": "Origin Identification for Text-Guided Image-to-Image Diffusion Models", "date": "", "ddg_snippet": "This motivates us to introduce the task of ori - gin IDentification for text-guided Image-to-image Diffusion models (ID2), aiming to retrieve the original image of a given translated query. A straightforward solution to ID2involves training a specialized deep embedding model to extract and compare features from both query and refer- ence images .", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=46n3izUNiv", "content": "This motivates us to introduce the task of ori - gin IDentification for text-guided Image-to-image Diffusion models (ID2), aiming to retrieve the original image of a given translated query. A straightforward solution to ID2involves training a specialized deep embedding model to extract and compare features from both query and refer- ence images ."}
+{"idx": 3, "title": "PDF Detecting Origin Attribution for Text-to-Image Diffusion Models", "date": "", "ddg_snippet": "Abstract Modern text - to - image (T2I) diffusion models can gener-ate images with remarkable realism and creativity. These advancements have sparked research in fake image detec-tion and attribution, yet prior studies have not fully explored the practical and scientific dimensions of this task. In addi-tion to attributing images to 12 state-of-the-art T2I genera-tors, we provide extensive ...", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/WACV2025/papers/Xu_Detecting_Origin_Attribution_for_Text-to-Image_Diffusion_Models_WACV_2025_paper.pdf", "content": "Abstract Modern text - to - image (T2I) diffusion models can gener-ate images with remarkable realism and creativity. These advancements have sparked research in fake image detec-tion and attribution, yet prior studies have not fully explored the practical and scientific dimensions of this task. In addi-tion to attributing images to 12 state-of-the-art T2I genera-tors, we provide extensive ..."}
+{"idx": 4, "title": "Generalizable Origin Identification for Text-Guided Image-to-Image ...", "date": "", "ddg_snippet": "This motivates us to introduce the task of origin IDentification for text-guided Image-to-image Diffusion models (ID^2), aiming to retrieve the original image of a given translated query. A straightforward solution to ID^2 involves training a specialized deep embedding model to extract and compare features from both query and reference images .", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/papers/2501.02376", "content": "This motivates us to introduce the task of origin IDentification for text-guided Image-to-image Diffusion models (ID^2), aiming to retrieve the original image of a given translated query. A straightforward solution to ID^2 involves training a specialized deep embedding model to extract and compare features from both query and reference images ."}
+{"idx": 5, "title": "Detecting Origin Attribution for Text-to-Image Diffusion Models", "date": "", "ddg_snippet": "Modern text - to - image (T2I) diffusion models can generate images with remarkable realism and creativity. These advancements have sparked research in fake image detection and attribution, yet prior studies have not fully explored the practical and scientific dimensions of this task. In addition to attributing images to 12 state-of-the-art T2I generators, we provide extensive analyses on what ...", "subpage_snippet": "", "source": "ieeexplore.ieee.org", "link": "https://ieeexplore.ieee.org/abstract/document/10944006", "content": "Modern text - to - image (T2I) diffusion models can generate images with remarkable realism and creativity. These advancements have sparked research in fake image detection and attribution, yet prior studies have not fully explored the practical and scientific dimensions of this task. In addition to attributing images to 12 state-of-the-art T2I generators, we provide extensive analyses on what ..."}
+{"idx": 6, "title": "Generalizable Origin Identification for Text-Guided Image-to-Image ...", "date": "", "ddg_snippet": "This motivates us to introduce the task of origin IDentification for text-guided Image-to-image Diffusion models (ID 2 ^2), aiming to retrieve the original image of a given translated query.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/387767437_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Models", "content": "This motivates us to introduce the task of origin IDentification for text-guided Image-to-image Diffusion models (ID 2 ^2), aiming to retrieve the original image of a given translated query."}
+{"idx": 7, "title": "Origin Identification for Text-Guided Image-to-Image Diffusion Models", "date": "", "ddg_snippet": "Text-guided image-to-image diffusion models ex- cel in translating images based on textual prompts, allowing for precise and creative visual modifica- tions. However, such a powerful technique can be misused for spreading misinformation, infring- ing on copyrights, and evading content tracing. This motivates us to introduce the task of ori - gin IDentification for text-guidedImage-to-image ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2501.02376", "content": "Text-guided image-to-image diffusion models ex- cel in translating images based on textual prompts, allowing for precise and creative visual modifica- tions. However, such a powerful technique can be misused for spreading misinformation, infring- ing on copyrights, and evading content tracing. This motivates us to introduce the task of ori - gin IDentification for text-guidedImage-to-image ..."}
+{"idx": 8, "title": "ICML Poster Origin Identification for Text-Guided Image-to-Image ...", "date": "", "ddg_snippet": "Abstract: Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for *spreading misinformation*, *infringing on copyrights*, and *evading content tracing*.", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46505", "content": "Abstract: Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for *spreading misinformation*, *infringing on copyrights*, and *evading content tracing*."}
+{"idx": 9, "title": "Generalizable Origin Identification for Text-Guided Image-to-Image ...", "date": "", "ddg_snippet": "Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for *spreading misinformation*, *infringing on copyrights*, and *evading content tracing*.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=daRu82GAoZ", "content": "Text-guided image-to-image diffusion models excel in translating images based on textual prompts, allowing for precise and creative visual modifications. However, such a powerful technique can be misused for *spreading misinformation*, *infringing on copyrights*, and *evading content tracing*."}
diff --git a/data/sampled_jsons/PS-EIP_EventPS_MAE_13.66_degrees_conclusion_section_8.jsonl b/data/sampled_jsons/PS-EIP_EventPS_MAE_13.66_degrees_conclusion_section_8.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c3965e074a391f4b64100996b2d896f307550a21
--- /dev/null
+++ b/data/sampled_jsons/PS-EIP_EventPS_MAE_13.66_degrees_conclusion_section_8.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Международно-правовой статус Государства Палестина...", "date": "", "ddg_snippet": "При этом 13 сентября 1993 года в результате Соглашений «Осло» между Израилем и Организацией освобождения Палестины (ООП) была подписана Декларация о создании Временной Палестинской самоуправляемой администрации (ПНА).", "subpage_snippet": "", "source": "ru.wikipedia.org", "link": "https://ru.wikipedia.org/wiki/Международно-правовой_статус_Государства_Палестина", "content": "При этом 13 сентября 1993 года в результате Соглашений «Осло» между Израилем и Организацией освобождения Палестины (ООП) была подписана Декларация о создании Временной Палестинской самоуправляемой администрации (ПНА)."}
+{"idx": 1, "title": "PS - EIP : Robust Photometric Stereo Based on Event Interval Profile", "date": "", "ddg_snippet": "8 . Conclusion . This paper presents EIP - PS , a robust event -based photomet-ric stereo method. The average MAEs for all 3D-printed objects was 8 .12 for EIP - PS , in contract to EventPS , which resulted in 13 . 66 .", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/papers/Kitazawa_PS-EIP_Robust_Photometric_Stereo_Based_on_Event_Interval_Profile_CVPR_2025_paper.pdf", "content": "8 . Conclusion . This paper presents EIP - PS , a robust event -based photomet-ric stereo method. The average MAEs for all 3D-printed objects was 8 .12 for EIP - PS , in contract to EventPS , which resulted in 13 . 66 ."}
+{"idx": 2, "title": "Procedure of the proposed method: Events are recorded under moving...", "date": "", "ddg_snippet": "PS - EIP : Robust Photometric Stereo Based on Event Interval Profile.1. Unlike EventPS , which treats each event interval independently, our approach considers a time-series profile formed by event intervals, taking into account the relationships between adjacent intervals.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/figure/Procedure-of-the-proposed-method-Events-are-recorded-under-moving-light-conditions-The_fig1_390142307", "content": "PS - EIP : Robust Photometric Stereo Based on Event Interval Profile.1. Unlike EventPS , which treats each event interval independently, our approach considers a time-series profile formed by event intervals, taking into account the relationships between adjacent intervals."}
+{"idx": 3, "title": "ГДЗ по английскому языку 6 класс (spotlight) Ваулина - рабочая...", "date": "", "ddg_snippet": "Pairwork activities 65. Student A 65 Student B 71. Revision Section 77.", "subpage_snippet": "", "source": "gdz.top", "link": "https://gdz.top/6-klass/english/vaulina-spotlight-rabochaja-tetrad", "content": "Pairwork activities 65. Student A 65 Student B 71. Revision Section 77."}
+{"idx": 4, "title": "Celsius (°C) Conversion Calculator - Inch Calculator", "date": "", "ddg_snippet": "Convert degrees Celsius (°C) to another temperature scale such as fahrenheit, kelvins, or rankine. Learn the formulas for Celsius temp conversions.", "subpage_snippet": "", "source": "www.inchcalculator.com", "link": "https://www.inchcalculator.com/convert/from-celsius/", "content": "Convert degrees Celsius (°C) to another temperature scale such as fahrenheit, kelvins, or rankine. Learn the formulas for Celsius temp conversions."}
+{"idx": 5, "title": "Login | Figma", "date": "", "ddg_snippet": "Create the section heading, paragraph text, and quote block Sign in to view 4m.", "subpage_snippet": "", "source": "www.figma.com", "link": "https://www.figma.com/login", "content": "Create the section heading, paragraph text, and quote block Sign in to view 4m."}
+{"idx": 6, "title": "ESET NOD32 | Свежие Ключи 2025 | Бесплатно | СТЕНА – Telegram", "date": "", "ddg_snippet": "ESET HOME Security Premium | До 15-10-2025 1 | WSFE-XDJT-DAAE-BBSS-MK3K 2 | M5KH-X7 PS -WVWD-4VKF-MHJ 8 3 | HC5B-XH5N-XADK-KXEG-G7GM 4 | FEP6-XPVU-PWSW-N 8 SK-9BM9 5 | 5SUB-X634-SNVV-HABP-4TXS 6 |. 73JE-XPFG-P 8 CM-AE7W-WGNF 7...", "subpage_snippet": "", "source": "t.me", "link": "https://t.me/s/eset_nod32_free", "content": "ESET HOME Security Premium | До 15-10-2025 1 | WSFE-XDJT-DAAE-BBSS-MK3K 2 | M5KH-X7 PS -WVWD-4VKF-MHJ 8 3 | HC5B-XH5N-XADK-KXEG-G7GM 4 | FEP6-XPVU-PWSW-N 8 SK-9BM9 5 | 5SUB-X634-SNVV-HABP-4TXS 6 |. 73JE-XPFG-P 8 CM-AE7W-WGNF 7..."}
+{"idx": 7, "title": "Получить описание изображения онлайн", "date": "", "ddg_snippet": "Опишем ваше фото с помощью нейросети. Сгенерируем текстовое описание к изображению.", "subpage_snippet": "", "source": "facee.ru", "link": "https://facee.ru/describe-image/", "content": "Опишем ваше фото с помощью нейросети. Сгенерируем текстовое описание к изображению."}
+{"idx": 8, "title": "Notices to Mariners - Weekly", "date": "", "ddg_snippet": "Radio Navigation Warnings. Sign in. Notices to Mariners. Value Added Resellers. About.", "subpage_snippet": "", "source": "msi.admiralty.co.uk", "link": "https://msi.admiralty.co.uk/NoticesToMariners/Weekly", "content": "Radio Navigation Warnings. Sign in. Notices to Mariners. Value Added Resellers. About."}
+{"idx": 9, "title": "Циан - база недвижимости в Ленинградской области | Продажа...", "date": "", "ddg_snippet": "от 13 848 445 ₽. Комендантский проспект. 8 мин. ЖК Морская миля.20-я Васильевского острова линия, 13Б.", "subpage_snippet": "", "source": "spb.cian.ru", "link": "https://spb.cian.ru/", "content": "от 13 848 445 ₽. Комендантский проспект. 8 мин. ЖК Морская миля.20-я Васильевского острова линия, 13Б."}
diff --git a/data/sampled_jsons/Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models_Algorithm_1_year_2024.jsonl b/data/sampled_jsons/Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models_Algorithm_1_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..8079b2df3da3b187b9dfd1acf2ac8d54d44b5fba
--- /dev/null
+++ b/data/sampled_jsons/Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models_Algorithm_1_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "PDF Parallel Simulation for Log-concave Sampling and Score-based Diffusion ...", "date": "", "ddg_snippet": "In this section, we present parallel Picard methods for strongly log-concave sampling ( Algorithm 1 ) and show it holds improved convergence rate w.r.t. the KL divergence and total variance (Theorem 4.2 and Corollary 4.3).", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/attachment?id=qtuxDy2qEB&name=pdf", "content": "In this section, we present parallel Picard methods for strongly log-concave sampling ( Algorithm 1 ) and show it holds improved convergence rate w.r.t. the KL divergence and total variance (Theorem 4.2 and Corollary 4.3)."}
+{"idx": 1, "title": "Parallel Simulation for Log-concave Sampling and Score-based Diffusion ...", "date": "", "ddg_snippet": "TL;DR: We propose new parallel algorithms for log-concave sampling and diffusion models , and rigorously prove that our algorithms enjoy $\\ log d$ iteration complexity.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=qtuxDy2qEB", "content": "TL;DR: We propose new parallel algorithms for log-concave sampling and diffusion models , and rigorously prove that our algorithms enjoy $\\ log d$ iteration complexity."}
+{"idx": 2, "title": "Parallel Simulation for Log-concave Sampling and Score-based Diffusion ...", "date": "", "ddg_snippet": "Spotlight Poster Parallel Simulation for Log-concave Sampling and Score-based Diffusion Models Huanjian Zhou · Masashi Sugiyama East Exhibition Hall A-B #E-1103", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/43916", "content": "Spotlight Poster Parallel Simulation for Log-concave Sampling and Score-based Diffusion Models Huanjian Zhou · Masashi Sugiyama East Exhibition Hall A-B #E-1103"}
+{"idx": 3, "title": "Parallel simulation for sampling under isoperimetry and score-based ...", "date": "", "ddg_snippet": "Our work highlights the potential advantages of simulation methods in scientific computation for dynamics- based sampling and diffusion models .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.07435v1", "content": "Our work highlights the potential advantages of simulation methods in scientific computation for dynamics- based sampling and diffusion models ."}
+{"idx": 4, "title": "Parallel Simulation for Log-concave Sampling and Score-based Diffusion ...", "date": "", "ddg_snippet": "This research paper presents a new way to sample from complex probability distributions in machine learning more quickly and efficiently by using parallel computing techniques. The authors compare...", "subpage_snippet": "", "source": "bytez.com", "link": "https://bytez.com/docs/icml/43916/paper", "content": "This research paper presents a new way to sample from complex probability distributions in machine learning more quickly and efficiently by using parallel computing techniques. The authors compare..."}
+{"idx": 5, "title": "Parallel Simulation for Sampling Under Isoperimetry and Score Based ...", "date": "", "ddg_snippet": "In recent years, there has been a surge of interest in proving discretization bounds for sampling under isoperimetry and for diffusion models . As data size grows, reducing the iteration cost becomes an important goal. Inspired by the great success of the parallel simulation of the initial value problem in scientific computation, we propose parallel Picard methods for sampling tasks. Rigorous ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=6Gb7VfTKY7", "content": "In recent years, there has been a surge of interest in proving discretization bounds for sampling under isoperimetry and for diffusion models . As data size grows, reducing the iteration cost becomes an important goal. Inspired by the great success of the parallel simulation of the initial value problem in scientific computation, we propose parallel Picard methods for sampling tasks. Rigorous ..."}
+{"idx": 6, "title": "Parallel simulation for sampling under isoperimetry and score-based ...", "date": "", "ddg_snippet": "This convergence analysis is based on a randomized midpoint method, which is first proposed for log-concave sampling (Shen and Lee, 2019), and then extended to diffusion models by Gupta et al. (2024).", "subpage_snippet": "", "source": "bohrium.dp.tech", "link": "https://bohrium.dp.tech/paper/arxiv/2412.07435", "content": "This convergence analysis is based on a randomized midpoint method, which is first proposed for log-concave sampling (Shen and Lee, 2019), and then extended to diffusion models by Gupta et al. (2024)."}
+{"idx": 7, "title": "PDF The Adaptive Complexity of Parallelized Log Concave Sampling", "date": "", "ddg_snippet": "ABSTRACT In large-data applications, such as the inference process of diffusion models , it is desirable to design sampling algorithms with a high degree of parallelization. In this work, we study the adaptive complexity of sampling , which is the minimum number of sequential rounds required to achieve sampling given polynomially many queries executed in parallel at each round. For unconstrained ...", "subpage_snippet": "", "source": "proceedings.iclr.cc", "link": "https://proceedings.iclr.cc/paper_files/paper/2025/file/297a7a54da76d95f02b04d957a53c7ba-Paper-Conference.pdf", "content": "ABSTRACT In large-data applications, such as the inference process of diffusion models , it is desirable to design sampling algorithms with a high degree of parallelization. In this work, we study the adaptive complexity of sampling , which is the minimum number of sequential rounds required to achieve sampling given polynomially many queries executed in parallel at each round. For unconstrained ..."}
+{"idx": 8, "title": "score-baseddiffusionmodels - arXiv.org", "date": "", "ddg_snippet": "In recent years, there has been a surge of interest in proving discretization bounds for sampling under isoperimetry and for difusion models . As data size grows, reducing the iteration cost becomes an important goal. Inspired by the great success of the parallel simulation of the initial value problem in scientific computation, we propose parallel Picard methods for sampling tasks. Rigorous ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2412.07435v1", "content": "In recent years, there has been a surge of interest in proving discretization bounds for sampling under isoperimetry and for difusion models . As data size grows, reducing the iteration cost becomes an important goal. Inspired by the great success of the parallel simulation of the initial value problem in scientific computation, we propose parallel Picard methods for sampling tasks. Rigorous ..."}
+{"idx": 9, "title": "PDF Accelerating Diffusion Models with Parallel Sampling: Inference at Sub ...", "date": "", "ddg_snippet": "Abstract Diffusion models have become a leading method for generative modeling of both image and scientific data. As these models are costly to train and evaluate, re-ducing the inference cost for diffusion models remains a major goal. Inspired by the recent empirical success in accelerating diffusion models via the parallel sam-pling technique [ 1 ], we propose to divide the sampling process ...", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/f162fa05675e3db4a733aafc081653cf-Paper-Conference.pdf", "content": "Abstract Diffusion models have become a leading method for generative modeling of both image and scientific data. As these models are costly to train and evaluate, re-ducing the inference cost for diffusion models remains a major goal. Inspired by the recent empirical success in accelerating diffusion models via the parallel sam-pling technique [ 1 ], we propose to divide the sampling process ..."}
diff --git a/data/sampled_jsons/Privacy_Loss_vs_C-Proxy_machine_unlearning_hard_samples.jsonl b/data/sampled_jsons/Privacy_Loss_vs_C-Proxy_machine_unlearning_hard_samples.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5bc8dc0b8c5026ced1b9c0f32a4bb76fc7e095ce
--- /dev/null
+++ b/data/sampled_jsons/Privacy_Loss_vs_C-Proxy_machine_unlearning_hard_samples.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Privacy - Wikipedia", "date": "", "ddg_snippet": "There are multiple techniques to invade privacy , which may be employed by corporations or governments for profit or political reasons. Conversely, in order to protect privacy , people may employ encryption or anonymity measures.", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Privacy", "content": "There are multiple techniques to invade privacy , which may be employed by corporations or governments for profit or political reasons. Conversely, in order to protect privacy , people may employ encryption or anonymity measures."}
+{"idx": 1, "title": "Privacy", "date": "", "ddg_snippet": "Privacy makes it easy to share secure payment information with trusted friends, family members, or employees. With one click, you can share a Privacy Card instead of having to share and expose your real debit or credit card details.", "subpage_snippet": "", "source": "www.privacy.com", "link": "https://www.privacy.com/", "content": "Privacy makes it easy to share secure payment information with trusted friends, family members, or employees. With one click, you can share a Privacy Card instead of having to share and expose your real debit or credit card details."}
+{"idx": 2, "title": "What Is Privacy ? | Privacy International", "date": "", "ddg_snippet": "Privacy is essential to who we are as human beings, and we make decisions about it every single day. It gives us a space to be ourselves without judgement, allows us to think freely without discrimination, and is an important element of giving us control over who knows what about us.", "subpage_snippet": "", "source": "privacyinternational.org", "link": "https://privacyinternational.org/explainer/56/what-privacy", "content": "Privacy is essential to who we are as human beings, and we make decisions about it every single day. It gives us a space to be ourselves without judgement, allows us to think freely without discrimination, and is an important element of giving us control over who knows what about us."}
+{"idx": 3, "title": "PRIVACY Definition & Meaning - Merriam-Webster", "date": "", "ddg_snippet": "The meaning of PRIVACY is the quality or state of being apart from company or observation : seclusion. How to use privacy in a sentence.", "subpage_snippet": "", "source": "www.merriam-webster.com", "link": "https://www.merriam-webster.com/dictionary/privacy", "content": "The meaning of PRIVACY is the quality or state of being apart from company or observation : seclusion. How to use privacy in a sentence."}
+{"idx": 4, "title": "Rights of privacy | Definition, Protection & Laws | Britannica", "date": "", "ddg_snippet": "Rights of privacy , in U.S. law, an amalgam of principles embodied in the federal Constitution or recognized by courts or lawmaking bodies concerning what Louis Brandeis, citing Judge Thomas Cooley, described in an 1890 paper (cowritten with Samuel D. Warren) as “the right to be let alone.”", "subpage_snippet": "", "source": "www.britannica.com", "link": "https://www.britannica.com/topic/rights-of-privacy", "content": "Rights of privacy , in U.S. law, an amalgam of principles embodied in the federal Constitution or recognized by courts or lawmaking bodies concerning what Louis Brandeis, citing Judge Thomas Cooley, described in an 1890 paper (cowritten with Samuel D. Warren) as “the right to be let alone.”"}
+{"idx": 5, "title": "Privacy ( Stanford Encyclopedia of Philosophy )", "date": "", "ddg_snippet": "May 14, 2002 · In this article, we will first focus on the histories of privacy in various discourses and spheres of life. We will also discuss the history of legislating privacy protections in different times and (legal) cultures.", "subpage_snippet": "", "source": "plato.stanford.edu", "link": "https://plato.stanford.edu/entries/privacy/", "content": "May 14, 2002 · In this article, we will first focus on the histories of privacy in various discourses and spheres of life. We will also discuss the history of legislating privacy protections in different times and (legal) cultures."}
+{"idx": 6, "title": "Privacy 101", "date": "", "ddg_snippet": "On this page, you’ll find articles and tools to help you get a basic understanding of the job of the privacy pro and data protection laws and practices around the globe.", "subpage_snippet": "", "source": "iapp.org", "link": "https://iapp.org/resources/topics/privacy-101/", "content": "On this page, you’ll find articles and tools to help you get a basic understanding of the job of the privacy pro and data protection laws and practices around the globe."}
+{"idx": 7, "title": "PRIVACY | English meaning - Cambridge Dictionary", "date": "", "ddg_snippet": "PRIVACY definition: 1. someone's right to keep their personal matters and relationships secret: 2. the state of being…. Learn more.", "subpage_snippet": "", "source": "dictionary.cambridge.org", "link": "https://dictionary.cambridge.org/dictionary/english/privacy", "content": "PRIVACY definition: 1. someone's right to keep their personal matters and relationships secret: 2. the state of being…. Learn more."}
+{"idx": 8, "title": "Privacy and why it matters – Information Technology", "date": "", "ddg_snippet": "Sep 27, 2024 · Though privacy concerns are not new, they have evolved with innovations in the use of personal data enabled by technology. The impacts of the intentional and unintentional misuse of personal data can relate to individuals, organizations, distinct communities, and society as a whole.", "subpage_snippet": "", "source": "it.uw.edu", "link": "https://it.uw.edu/guides/privacy/reference-materials/privacy-and-why-it-matters/", "content": "Sep 27, 2024 · Though privacy concerns are not new, they have evolved with innovations in the use of personal data enabled by technology. The impacts of the intentional and unintentional misuse of personal data can relate to individuals, organizations, distinct communities, and society as a whole."}
+{"idx": 9, "title": "Defining Privacy - Markkula Center for Applied Ethics", "date": "", "ddg_snippet": "When we think about privacy , most of us think about particular limits on what we want others to know about us. Those \"others\" include governments, but also corporations, teachers, school administrators, parents, siblings, law enforcement agents, classmates, strangers, neighbors, friends.", "subpage_snippet": "", "source": "www.scu.edu", "link": "https://www.scu.edu/ethics/privacy/defining-privacy/", "content": "When we think about privacy , most of us think about particular limits on what we want others to know about us. Those \"others\" include governments, but also corporations, teachers, school administrators, parents, siblings, law enforcement agents, classmates, strangers, neighbors, friends."}
diff --git a/data/sampled_jsons/Qwen2.5-VL_XLRS-Bench_Table_2_Avg_score_Chinese_English_sitearxiv.org_year_2024.jsonl b/data/sampled_jsons/Qwen2.5-VL_XLRS-Bench_Table_2_Avg_score_Chinese_English_sitearxiv.org_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..de09b874cdc3c451bb1709f31bc8fe1f83ae747d
--- /dev/null
+++ b/data/sampled_jsons/Qwen2.5-VL_XLRS-Bench_Table_2_Avg_score_Chinese_English_sitearxiv.org_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "MME-SCI: A Comprehensive and Challenging Science Benchmark for...", "date": "", "ddg_snippet": "Table 1: Comparison between our MME-SCI and others. ML. denotes Multilingual, CMC. signifies Comprehensive Modality Coverage, MD. represents Multidisciplinary, and FKP. stands for Fine-grained Knowledge Points.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.13938v1", "content": "Table 1: Comparison between our MME-SCI and others. ML. denotes Multilingual, CMC. signifies Comprehensive Modality Coverage, MD. represents Multidisciplinary, and FKP. stands for Fine-grained Knowledge Points."}
+{"idx": 1, "title": "XLRS - Bench : Could Your Multimodal LLMs Understand Extremely...", "date": "", "ddg_snippet": "Qwen 2 - VL excels in both English and Chinese proficiency, outperforming both proprietary and most open-source models. Nevertheless, their performance varies across tasks. From the results, we can draw the following key insights: (1) GPT-4o’s Weakness in Spatiotemporal Reasoning...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.23771v1", "content": "Qwen 2 - VL excels in both English and Chinese proficiency, outperforming both proprietary and most open-source models. Nevertheless, their performance varies across tasks. From the results, we can draw the following key insights: (1) GPT-4o’s Weakness in Spatiotemporal Reasoning..."}
+{"idx": 2, "title": "Multi-TW: Benchmarking Multimodal Models on Traditional Chinese ...", "date": "", "ddg_snippet": "Table 2 . Comparison of Multi-TW with other datasets. For ALM- Bench , we only compare the subset for Traditional Chinese .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.01274v1", "content": "Table 2 . Comparison of Multi-TW with other datasets. For ALM- Bench , we only compare the subset for Traditional Chinese ."}
+{"idx": 3, "title": "MME-SCI: A Comprehensive and Challenging Science Benchmark for...", "date": "", "ddg_snippet": "Table 2 : Results on MME-SCI. We present detailed comparative results for the Dzh and Dizmh g scenarios, including accuracy for four subjects.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2508.13938", "content": "Table 2 : Results on MME-SCI. We present detailed comparative results for the Dzh and Dizmh g scenarios, including accuracy for four subjects."}
+{"idx": 4, "title": "Qwen2.5-VL Technical Report - arXiv.org", "date": "", "ddg_snippet": "In this report, we introduce the latest work Qwen2.5-VL , which continues the open-source philosophy of the Qwen series, achieving and even surpassing top-tier closed-source models on various benchmarks. Technically, our contributions are four-folds: (1) We implement window attention in the visual encoder to optimize inference eficiency; ( 2 ) We introduce dynamic FPS sampling, extending dynamic ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2502.13923v1", "content": "In this report, we introduce the latest work Qwen2.5-VL , which continues the open-source philosophy of the Qwen series, achieving and even surpassing top-tier closed-source models on various benchmarks. Technically, our contributions are four-folds: (1) We implement window attention in the visual encoder to optimize inference eficiency; ( 2 ) We introduce dynamic FPS sampling, extending dynamic ..."}
+{"idx": 5, "title": "[2502.13923] Qwen2.5-VL Technical Report - arXiv.org", "date": "", "ddg_snippet": "We introduce Qwen2.5-VL , the latest flagship model of Qwen vision-language series, which demonstrates significant advancements in both foundational capabilities and innovative functionalities. Qwen2.5-VL achieves a major leap forward in understanding and interacting with the world through enhanced visual recognition, precise object localization, robust document parsing, and long-video ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2502.13923", "content": "We introduce Qwen2.5-VL , the latest flagship model of Qwen vision-language series, which demonstrates significant advancements in both foundational capabilities and innovative functionalities. Qwen2.5-VL achieves a major leap forward in understanding and interacting with the world through enhanced visual recognition, precise object localization, robust document parsing, and long-video ..."}
+{"idx": 6, "title": "[2412.15115] Qwen2.5 Technical Report - arXiv.org", "date": "", "ddg_snippet": "In this report, we introduce Qwen2.5 , a comprehensive series of large language models (LLMs) designed to meet diverse needs. Compared to previous iterations, Qwen 2.5 has been significantly improved during both the pre-training and post-training stages. In terms of pre-training, we have scaled the high-quality pre-training datasets from the previous 7 trillion tokens to 18 trillion tokens ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2412.15115", "content": "In this report, we introduce Qwen2.5 , a comprehensive series of large language models (LLMs) designed to meet diverse needs. Compared to previous iterations, Qwen 2.5 has been significantly improved during both the pre-training and post-training stages. In terms of pre-training, we have scaled the high-quality pre-training datasets from the previous 7 trillion tokens to 18 trillion tokens ..."}
+{"idx": 7, "title": "Qwen2 Technical Report - arXiv.org", "date": "", "ddg_snippet": "The Qwen2 series consists of models of 5 sizes, which are Qwen2-0.5B, Qwen2-1.5B, Qwen2-7B, Qwen2-57B-A14B, and Qwen2-72B. Table 1 lists the hyper-parameters and important information, e.g., the number of pre-trained tokens.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2407.10671v4", "content": "The Qwen2 series consists of models of 5 sizes, which are Qwen2-0.5B, Qwen2-1.5B, Qwen2-7B, Qwen2-57B-A14B, and Qwen2-72B. Table 1 lists the hyper-parameters and important information, e.g., the number of pre-trained tokens."}
+{"idx": 8, "title": "\\thetable Performance comparison across different evidence source on ...", "date": "", "ddg_snippet": "However, Qwen2.5-VL shows an extremely marked improvement over Qwen2- VL on MMLongBench, resulting higher average scores . MMLongBench's greater reliance on image-based questions might explain Qwen2.5-VL's superior performance on this benchmark, possibly indicating that Qwen2.5-VL is better at handling visual question-answering tasks, but ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.13964v1", "content": "However, Qwen2.5-VL shows an extremely marked improvement over Qwen2- VL on MMLongBench, resulting higher average scores . MMLongBench's greater reliance on image-based questions might explain Qwen2.5-VL's superior performance on this benchmark, possibly indicating that Qwen2.5-VL is better at handling visual question-answering tasks, but ..."}
+{"idx": 9, "title": "[2407.10671] Qwen2 Technical Report - arXiv.org", "date": "", "ddg_snippet": "The instruction-tuned variant, Qwen2-72B-Instruct, attains 9.1 on MT- Bench , 48.1 on Arena-Hard, and 35.7 on LiveCodeBench. Moreover, Qwen2 demonstrates robust multilingual capabilities, proficient in approximately 30 languages, spanning English , Chinese , Spanish, French, German, Arabic, Russian, Korean, Japanese, Thai, Vietnamese, and more ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2407.10671", "content": "The instruction-tuned variant, Qwen2-72B-Instruct, attains 9.1 on MT- Bench , 48.1 on Arena-Hard, and 35.7 on LiveCodeBench. Moreover, Qwen2 demonstrates robust multilingual capabilities, proficient in approximately 30 languages, spanning English , Chinese , Spanish, French, German, Arabic, Russian, Korean, Japanese, Thai, Vietnamese, and more ..."}
diff --git a/data/sampled_jsons/RAGGED_Towards_Informed_Design_of_Scalable_and_Stable_RAG_Systems_Section_3.1_retriever_paradigms.jsonl b/data/sampled_jsons/RAGGED_Towards_Informed_Design_of_Scalable_and_Stable_RAG_Systems_Section_3.1_retriever_paradigms.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..801312ba153d92902be87c66cf62479030b67ecc
--- /dev/null
+++ b/data/sampled_jsons/RAGGED_Towards_Informed_Design_of_Scalable_and_Stable_RAG_Systems_Section_3.1_retriever_paradigms.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Gainsborough's House", "date": "", "ddg_snippet": "The museum and art gallery at the birthplace of Thomas Gainsborough in Sudbury, Suffolk, UK. See all news, exhibition updates and more.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/", "content": "The museum and art gallery at the birthplace of Thomas Gainsborough in Sudbury, Suffolk, UK. See all news, exhibition updates and more."}
+{"idx": 1, "title": "Heritage Open Weekend 2025 - gainsborough.org", "date": "", "ddg_snippet": "Explore the very rooms where Thomas Gainsborough lived and worked, discover his story in beautifully restored eighteenth-century surroundings, and enjoy the tranquil historic garden featuring contemporary sculptures by Helaine Blumenfeld OBE.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/event/heritage-open-weekend-2025/", "content": "Explore the very rooms where Thomas Gainsborough lived and worked, discover his story in beautifully restored eighteenth-century surroundings, and enjoy the tranquil historic garden featuring contemporary sculptures by Helaine Blumenfeld OBE."}
+{"idx": 2, "title": "What's On - Gainsborough's House", "date": "", "ddg_snippet": "May 3 , 2025 · All of our current exhibitions and events in one place.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/whats-on/", "content": "May 3 , 2025 · All of our current exhibitions and events in one place."}
+{"idx": 3, "title": "Getting here - Gainsborough's House", "date": "", "ddg_snippet": "Gainsborough’s House is situated in the centre of Sudbury, Suffolk on Gainsborough Street, at the bottom of Market Hill. The cafe and shop are on Weavers Lane which are adjacent to the front of the House.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/your-visit/getting-here/", "content": "Gainsborough’s House is situated in the centre of Sudbury, Suffolk on Gainsborough Street, at the bottom of Market Hill. The cafe and shop are on Weavers Lane which are adjacent to the front of the House."}
+{"idx": 4, "title": "Your visit - Gainsborough's House", "date": "", "ddg_snippet": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/your-visit/", "content": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment."}
+{"idx": 5, "title": "Sudbury - gainsborough.org", "date": "", "ddg_snippet": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment. The international centre for Thomas Gainsborough is now the largest gallery in Suffolk.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/collection/sudbury/", "content": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment. The international centre for Thomas Gainsborough is now the largest gallery in Suffolk."}
+{"idx": 6, "title": "About - Gainsborough's House", "date": "", "ddg_snippet": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/about/", "content": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment."}
+{"idx": 7, "title": "Collection - Gainsborough's House", "date": "", "ddg_snippet": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/collection/", "content": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment."}
+{"idx": 8, "title": "Events from June 7, 2014 – May 26, 2015 - Gainsborough's House", "date": "", "ddg_snippet": "Helaine Blumenfeld: Tree of Life Nov 15 November 15, 2025 - March 22, 2026", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/events/category/exhibitions/", "content": "Helaine Blumenfeld: Tree of Life Nov 15 November 15, 2025 - March 22, 2026"}
+{"idx": 9, "title": "Reviving an Artist’s Birthplace - Gainsborough's House", "date": "", "ddg_snippet": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment.", "subpage_snippet": "", "source": "gainsborough.org", "link": "https://gainsborough.org/about/reviving-an-artists-birthplace/", "content": "The beautiful historic garden at the heart of Gainsborough’s House is maintained by a devoted body of volunteers. In November 2022, the museum re-opened to the public after a transformational refurbishment."}
diff --git a/data/sampled_jsons/RAGGED_paper_Izacard_Grave_2021_retrieval_depth_claim.jsonl b/data/sampled_jsons/RAGGED_paper_Izacard_Grave_2021_retrieval_depth_claim.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c9042e5e2ce32a4b188ff2f2e9340b68aa0c38f6
--- /dev/null
+++ b/data/sampled_jsons/RAGGED_paper_Izacard_Grave_2021_retrieval_depth_claim.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Leveraging Passage Retrieval with Generative Models for Open Domain ...", "date": "", "ddg_snippet": "View a PDF of the paper titled Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering, by Gautier Izacard and Edouard Grave", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2007.01282", "content": "View a PDF of the paper titled Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering, by Gautier Izacard and Edouard Grave"}
+{"idx": 1, "title": "Published as a conference paper at ICLR 2021 - OpenReview", "date": "", "ddg_snippet": "Gautier Izacard and Edouard Grave . Leveraging passage retrieval with generative models for open domain question answering. arXiv preprint arXiv:2007.01282, 2020.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=NTEz-6wysdb", "content": "Gautier Izacard and Edouard Grave . Leveraging passage retrieval with generative models for open domain question answering. arXiv preprint arXiv:2007.01282, 2020."}
+{"idx": 2, "title": "Leveraging Passage Retrieval with Generative Models for Open Domain ...", "date": "", "ddg_snippet": "Gautier Izacard , Edouard Grave . Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume. 2021 .", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/2021.eacl-main.74/", "content": "Gautier Izacard , Edouard Grave . Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume. 2021 ."}
+{"idx": 3, "title": "RAGGED: Towards Informed Design of Scalable and Stable RAG Systems", "date": "", "ddg_snippet": "Retrieval Depth and Performance Prior work offers mixed conclusions on increasing retrieval depth (k). Some studies report consistent improvements ( Izacard & Grave , 2021 ), while others find diminishing returns (Liu et al., 2023) or even performance degradation at high k (Cuconasu et al., 2024; Jiang et al., 2024).", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2403.09040v3", "content": "Retrieval Depth and Performance Prior work offers mixed conclusions on increasing retrieval depth (k). Some studies report consistent improvements ( Izacard & Grave , 2021 ), while others find diminishing returns (Liu et al., 2023) or even performance degradation at high k (Cuconasu et al., 2024; Jiang et al., 2024)."}
+{"idx": 4, "title": "Leveraging Passage Retrieval with Generative Models for Open Domain ...", "date": "", "ddg_snippet": "Generative models for open domain question answering have proven to be competitive, without resorting to external knowledge. While promising, this approach requires to use models with billions of parameters, which are expensive to train and query. In this paper , we investigate how much these models can benefit from retrieving text passages, potentially containing evidence. We obtain state-of ...", "subpage_snippet": "", "source": "hal.science", "link": "https://hal.science/hal-03463108", "content": "Generative models for open domain question answering have proven to be competitive, without resorting to external knowledge. While promising, this approach requires to use models with billions of parameters, which are expensive to train and query. In this paper , we investigate how much these models can benefit from retrieving text passages, potentially containing evidence. We obtain state-of ..."}
+{"idx": 5, "title": "dblp: Leveraging Passage Retrieval with Generative Models for Open ...", "date": "", "ddg_snippet": "Bibliographic details on Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering.", "subpage_snippet": "", "source": "dblp.org", "link": "https://dblp.org/rec/conf/eacl/IzacardG21", "content": "Bibliographic details on Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering."}
+{"idx": 6, "title": "Ragged: Towards Informed Design of Retrieval Augmented Generation ...", "date": "", "ddg_snippet": "provides mixed, even 033 contradictory, suggestions for configuring RAG. While some early works suggest that providing more 034 retrieved passages results in strictly better outputs ( Izacard & Grave , 2021 ), others find there is a 035 limit to that phenomenon as model p", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=KDXj60FpJr", "content": "provides mixed, even 033 contradictory, suggestions for configuring RAG. While some early works suggest that providing more 034 retrieved passages results in strictly better outputs ( Izacard & Grave , 2021 ), others find there is a 035 limit to that phenomenon as model p"}
+{"idx": 7, "title": "Leveraging Passage Retrieval with Generative Models for Open Domain ...", "date": "", "ddg_snippet": "Download Citation | On Jan 1, 2021 , Gautier Izacard and others published Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering | Find, read and cite all the ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/355430399_Leveraging_Passage_Retrieval_with_Generative_Models_for_Open_Domain_Question_Answering", "content": "Download Citation | On Jan 1, 2021 , Gautier Izacard and others published Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering | Find, read and cite all the ..."}
+{"idx": 8, "title": "Enhancing LLMs with Vectorization in RAG for Structured Data", "date": "", "ddg_snippet": "A paper by Izacard & Grave ( 2021 ) demonstrated that combining vector search with RAG improves open-domain question-answering accuracy. These findings confirm that integrating retrieval techniques enhances LLM reliability, particularly for structured data applications.", "subpage_snippet": "", "source": "engineerslog.com", "link": "https://engineerslog.com/enhancing-llms-with-vectorization-in-retrieval-augmented-generation-rag-for-structured-data/", "content": "A paper by Izacard & Grave ( 2021 ) demonstrated that combining vector search with RAG improves open-domain question-answering accuracy. These findings confirm that integrating retrieval techniques enhances LLM reliability, particularly for structured data applications."}
+{"idx": 9, "title": "ACL Anthology", "date": "", "ddg_snippet": "@inproceedings{izacard- grave - 2021 -leveraging, title = \"Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering\", author = \" Izacard , Gautier and Grave , Edouard\", editor = \"Merlo, Paola and Tiedemann, Jorg and Tsarfaty, Reut\",", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/2021.eacl-main.74.bib", "content": "@inproceedings{izacard- grave - 2021 -leveraging, title = \"Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering\", author = \" Izacard , Gautier and Grave , Edouard\", editor = \"Merlo, Paola and Tiedemann, Jorg and Tsarfaty, Reut\","}
diff --git a/data/sampled_jsons/RAP_Reasoning_via_Planning_method_description_tokens.jsonl b/data/sampled_jsons/RAP_Reasoning_via_Planning_method_description_tokens.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..118804e059d0a72dbf5cd9ff247f5a338aef4ddf
--- /dev/null
+++ b/data/sampled_jsons/RAP_Reasoning_via_Planning_method_description_tokens.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Reasoning with Language Model is Planning with World Model", "date": "", "ddg_snippet": "Language Model. Reasoning via Planning ( RAP ).In this paper, we present Reasoning via Planning ( RAP ), a novel LLM reasoning framework that equips LLMs with an ability to reason akin to human-like strategic planning.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=VTWWvYtF1R", "content": "Language Model. Reasoning via Planning ( RAP ).In this paper, we present Reasoning via Planning ( RAP ), a novel LLM reasoning framework that equips LLMs with an ability to reason akin to human-like strategic planning."}
+{"idx": 1, "title": "GitHub - Ber666/ RAP : Reasoning with Language Model is Planning ...", "date": "", "ddg_snippet": "RAP : Reasoning via Planning .For RAP -Aggregation, after running RAP on GSM8k, run python aggregate_gsm8k.py --log-dir .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/Ber666/RAP", "content": "RAP : Reasoning via Planning .For RAP -Aggregation, after running RAP on GSM8k, run python aggregate_gsm8k.py --log-dir ."}
+{"idx": 2, "title": "Advancing Reasoning Strategies in Large Language Models", "date": "", "ddg_snippet": "Reasoning via Planning ( RAP ): Simulating Long-Term Impact. Reasoning via Planning ( RAP ) represents another advanced strategy that leverages LLMs as both the reasoning engine and world model.", "subpage_snippet": "", "source": "www.linkedin.com", "link": "https://www.linkedin.com/pulse/advancing-reasoning-strategies-large-language-models-yerramsetti-irkec", "content": "Reasoning via Planning ( RAP ): Simulating Long-Term Impact. Reasoning via Planning ( RAP ) represents another advanced strategy that leverages LLMs as both the reasoning engine and world model."}
+{"idx": 3, "title": "(PDF) Reasoning with Language Model is Planning with World Model", "date": "", "ddg_snippet": "Reasoning via Planning ( RAP ).Language Model. Figure 1: An overview of Reasoning via Planning ( RAP ). Compared with previous LLM reasoning methods like.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/371009675_Reasoning_with_Language_Model_is_Planning_with_World_Model", "content": "Reasoning via Planning ( RAP ).Language Model. Figure 1: An overview of Reasoning via Planning ( RAP ). Compared with previous LLM reasoning methods like."}
+{"idx": 4, "title": "Reasoning via Planning ( RAP ) the LLM Reasoners - MsTechDiva", "date": "", "ddg_snippet": "Reasoning via Planning ( RAP ) represents an advancement in enhancing the capabilities of Language Models (LLMs) by providing a robust framework for handling complex reasoning tasks.Source code Paper for Reasoning via Planning ( RAP ).", "subpage_snippet": "", "source": "mstechdiva.com", "link": "https://mstechdiva.com/reasoning-via-planning-rap-the-llm-reasoners/", "content": "Reasoning via Planning ( RAP ) represents an advancement in enhancing the capabilities of Language Models (LLMs) by providing a robust framework for handling complex reasoning tasks.Source code Paper for Reasoning via Planning ( RAP )."}
+{"idx": 5, "title": "Prompting LLMs with World Models: A Strategy for Better Reasoning", "date": "", "ddg_snippet": "A recent framework called Reasoning via Planning ( RAP ) suggests using an LLM as both a world model and a reasoning planner to solve problems, which improved performance on complex tasksoutshift.cisco.com.", "subpage_snippet": "", "source": "watchsound.medium.com", "link": "https://watchsound.medium.com/prompting-llms-with-world-models-a-strategy-for-better-reasoning-d5449bb17237", "content": "A recent framework called Reasoning via Planning ( RAP ) suggests using an LLM as both a world model and a reasoning planner to solve problems, which improved performance on complex tasksoutshift.cisco.com."}
+{"idx": 6, "title": "Reasoning with Language Model is Planning with World Model", "date": "", "ddg_snippet": "It incorporates planning algorithms like Monte Carlo Tree Search (MCTS) to navigate through reasoning steps efficiently. RAP 's key strength is its balanced approach to exploring new reasoning paths while also focusing on paths that promise high rewards.", "subpage_snippet": "", "source": "blog.athina.ai", "link": "https://blog.athina.ai/reasoning-with-language-model-is-planning-with-world-model", "content": "It incorporates planning algorithms like Monte Carlo Tree Search (MCTS) to navigate through reasoning steps efficiently. RAP 's key strength is its balanced approach to exploring new reasoning paths while also focusing on paths that promise high rewards."}
+{"idx": 7, "title": "What is a Reasoning Engine and How Does It Work?", "date": "", "ddg_snippet": "Reasoning via Planning ( RAP ). Reasoning via Planning ( RAP ) integrates tree-search algorithms with LLMs to guide multi-step reasoning. This method leverages a learned value function to evaluate different reasoning paths, optimizing decision-making.", "subpage_snippet": "", "source": "coralogix.com", "link": "https://coralogix.com/ai-blog/what-is-a-reasoning-engine/", "content": "Reasoning via Planning ( RAP ). Reasoning via Planning ( RAP ) integrates tree-search algorithms with LLMs to guide multi-step reasoning. This method leverages a learned value function to evaluate different reasoning paths, optimizing decision-making."}
+{"idx": 8, "title": "GameBench: Testing Strategic Reasoning in Language Models", "date": "", "ddg_snippet": "Evaluating LLMs' strategic reasoning capabilities using diverse games.In our tests, we used both GPT-3 and GPT-4 and incorporated two methods to boost their reasoning abilities: Chain-of-Thought (CoT) prompting and Reasoning Via Planning ( RAP ).", "subpage_snippet": "", "source": "scisimple.com", "link": "https://scisimple.com/en/articles/2025-08-01-gamebench-testing-strategic-reasoning-in-language-models--a9np7ol", "content": "Evaluating LLMs' strategic reasoning capabilities using diverse games.In our tests, we used both GPT-3 and GPT-4 and incorporated two methods to boost their reasoning abilities: Chain-of-Thought (CoT) prompting and Reasoning Via Planning ( RAP )."}
+{"idx": 9, "title": "Solving Reasoning Problems with LLMs in 2023", "date": "", "ddg_snippet": "This query finds relevance in two methodologies: reasoning - via - planning ( RAP ) and tree-of-thoughts (ToT). Both empower LLMs to navigate through possible reasoning steps, and to search for the optimal reasoning chain based on specific evaluations.", "subpage_snippet": "", "source": "readmedium.com", "link": "https://readmedium.com/solving-reasoning-problems-with-llms-in-2023-6643bdfd606d", "content": "This query finds relevance in two methodologies: reasoning - via - planning ( RAP ) and tree-of-thoughts (ToT). Both empower LLMs to navigate through possible reasoning steps, and to search for the optimal reasoning chain based on specific evaluations."}
diff --git a/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Figure_5(c)_standard_DPO_per-step_DPO.jsonl b/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Figure_5(c)_standard_DPO_per-step_DPO.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..63479f86916d56b4a941d31a88915ef6ed234691
--- /dev/null
+++ b/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Figure_5(c)_standard_DPO_per-step_DPO.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "PDF RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math ...", "date": "", "ddg_snippet": "e data , attaining performance similar to amplifying the amount of synthetic data by 8×. We show that training on per-step negatives can help to unlearn spurious correlations in the positive data , and is equivalent to advantage-weighted reinforcement learning ( RL )", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/4b77d5b896c321a29277524a98a50215-Paper-Conference.pdf", "content": "e data , attaining performance similar to amplifying the amount of synthetic data by 8×. We show that training on per-step negatives can help to unlearn spurious correlations in the positive data , and is equivalent to advantage-weighted reinforcement learning ( RL )"}
+{"idx": 1, "title": "RLonIncorrectSyntheticDataScalesthe EfficiencyofLLMMathReasoningbyEight ...", "date": "", "ddg_snippet": "RLonIncorrectSyntheticDataScalestheEfficiencyofLLMMathReasoningbyEight-Fold SFTbase Policy ! QnA pairs sampled from GPT/Gemini Synthetic Data Positive Data Correct answers \"! Negative Data Finetune policy RFT : SFT on self-generated correct answers !\" RL with step -level rewards on all answers #!\" e.g., preference - based RL Incorrect answers \"!", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2406.14532", "content": "RLonIncorrectSyntheticDataScalestheEfficiencyofLLMMathReasoningbyEight-Fold SFTbase Policy ! QnA pairs sampled from GPT/Gemini Synthetic Data Positive Data Correct answers \"! Negative Data Finetune policy RFT : SFT on self-generated correct answers !\" RL with step -level rewards on all answers #!\" e.g., preference - based RL Incorrect answers \"!"}
+{"idx": 2, "title": "RL on Incorrect Synthetic Data · MinWoo Park", "date": "", "ddg_snippet": "We show that training on per-step negatives can help to unlearn spurious correlations in the positive data , and is equivalent to advantage-weighted reinforcement learning ( RL ), implying that it inherits robustness benefits of RL over imitating positive data alone.", "subpage_snippet": "", "source": "dsdanielpark.github.io", "link": "https://dsdanielpark.github.io/llm/2024-06-25-RLonIncorrectSyntheticData.html", "content": "We show that training on per-step negatives can help to unlearn spurious correlations in the positive data , and is equivalent to advantage-weighted reinforcement learning ( RL ), implying that it inherits robustness benefits of RL over imitating positive data alone."}
+{"idx": 3, "title": "GitHub - dvlab-research/Step-DPO: Implementation for \"Step-DPO: Step ...", "date": "", "ddg_snippet": "This repo provides the implementation of Step-DPO , a simple, effective, and data -efficient method for boosting the long-chain reasoning ability of LLMs, with a data construction pipeline that yields a high-quality dataset containing 10K step -wise preference pairs.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/dvlab-research/Step-DPO", "content": "This repo provides the implementation of Step-DPO , a simple, effective, and data -efficient method for boosting the long-chain reasoning ability of LLMs, with a data construction pipeline that yields a high-quality dataset containing 10K step -wise preference pairs."}
+{"idx": 4, "title": "xinlai/DeepSeekMath-RL-Step-DPO · Hugging Face", "date": "", "ddg_snippet": "Step-DPO : Step -wise Preference Optimization for Long-chain Reasoning of LLMs 🖥️ Code | 🤗 Data | 📄 Paper This repo contains the DeepSeekMath- RL - Step - DPO model. It is obtained by performing Step-DPO on DeepSeekMath- RL . Step-DPO is a simple, effective, and data -efficient method for boosting the mathematical reasoning ability of LLMs.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/xinlai/DeepSeekMath-RL-Step-DPO", "content": "Step-DPO : Step -wise Preference Optimization for Long-chain Reasoning of LLMs 🖥️ Code | 🤗 Data | 📄 Paper This repo contains the DeepSeekMath- RL - Step - DPO model. It is obtained by performing Step-DPO on DeepSeekMath- RL . Step-DPO is a simple, effective, and data -efficient method for boosting the mathematical reasoning ability of LLMs."}
+{"idx": 5, "title": "RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math ...", "date": "", "ddg_snippet": "The figure demonstrates that per-step DPO improves Q-values at each step , while standard DPO only shows improvement in irrelevant steps . This highlights the advantage of per-step DPO in focusing on critical steps during training.", "subpage_snippet": "", "source": "deep-diver.github.io", "link": "https://deep-diver.github.io/neurips2024/posters/9m87e9keq1/", "content": "The figure demonstrates that per-step DPO improves Q-values at each step , while standard DPO only shows improvement in irrelevant steps . This highlights the advantage of per-step DPO in focusing on critical steps during training."}
+{"idx": 6, "title": "Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of ...", "date": "", "ddg_snippet": "Additionally, we have developed a data construction pipeline for Step-DPO , enabling the creation of a high-quality dataset containing 10K step -wise preference pairs. We also observe that in DPO , self-generated data is more effective than data generated by humans or GPT-4, due to the latter's out-of-distribution nature.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2406.18629", "content": "Additionally, we have developed a data construction pipeline for Step-DPO , enabling the creation of a high-quality dataset containing 10K step -wise preference pairs. We also observe that in DPO , self-generated data is more effective than data generated by humans or GPT-4, due to the latter's out-of-distribution nature."}
+{"idx": 7, "title": "RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math ...", "date": "", "ddg_snippet": "With this per-step scheme, we are able to attain consistent gains over only positive data , attaining performance similar to amplifying the amount of synthetic data by $\\mathbf {8 \\times}$.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/381604579_RL_on_Incorrect_Synthetic_Data_Scales_the_Efficiency_of_LLM_Math_Reasoning_by_Eight-Fold", "content": "With this per-step scheme, we are able to attain consistent gains over only positive data , attaining performance similar to amplifying the amount of synthetic data by $\\mathbf {8 \\times}$."}
+{"idx": 8, "title": "GitHub - TU2021/DPO-VP: Improving Math reasoning through Direct ...", "date": "", "ddg_snippet": "TL;DR: We enhance the mathematical reasoning ability of LLMs solely through Verifiable Reward filtering and the self-improvement training paradigm of DPO . The final model, Qwen2.5-7B- DPO -VP, demonstrates mathematical reasoning capabilities comparable to current RL -based approaches. The entire framework does not require model parallelism, enabling replication on a single A800 GPU.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/TU2021/DPO-VP", "content": "TL;DR: We enhance the mathematical reasoning ability of LLMs solely through Verifiable Reward filtering and the self-improvement training paradigm of DPO . The final model, Qwen2.5-7B- DPO -VP, demonstrates mathematical reasoning capabilities comparable to current RL -based approaches. The entire framework does not require model parallelism, enabling replication on a single A800 GPU."}
+{"idx": 9, "title": "Learning to Reason by Failing: Offline RL on Sub-optimal Rollouts ...", "date": "", "ddg_snippet": "This paper studies the scaling laws of two types of synthetic positive responses for training. It provides a valuable insight into the effectiveness of the self-generated responses and its shortages, which could benefit the scope of scaling synthetic data to improve LLMs. This paper suggests to use per-step DPO to incorporate negative synthetic responses to address the spurious pattern issue ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=v2PV1yCFJk", "content": "This paper studies the scaling laws of two types of synthetic positive responses for training. It provides a valuable insight into the effectiveness of the self-generated responses and its shortages, which could benefit the scope of scaling synthetic data to improve LLMs. This paper suggests to use per-step DPO to incorporate negative synthetic responses to address the spurious pattern issue ..."}
diff --git a/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Llama_DeepSeek_base_model_Section_5_experiments.jsonl b/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Llama_DeepSeek_base_model_Section_5_experiments.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..6f6b224e40048f14a165a7bb82cd60a26a2c9157
--- /dev/null
+++ b/data/sampled_jsons/RL_on_Incorrect_Synthetic_Data_Llama_DeepSeek_base_model_Section_5_experiments.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "ASTRO: Teaching Language Models to Reason by Reflecting and", "date": "", "ddg_snippet": "After RL , our policy based on llama -3.1-70b-instruct achieves 81.8% in MATH-500, 64.4% in AMC 2023 and 30.0% in AIME 2024 (pass@1).", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.00417v1", "content": "After RL , our policy based on llama -3.1-70b-instruct achieves 81.8% in MATH-500, 64.4% in AMC 2023 and 30.0% in AIME 2024 (pass@1)."}
+{"idx": 1, "title": "Improving Value-based Process Verifier via Low-Cost Variance", "date": "", "ddg_snippet": "In beam search experiments , our method trained on Deepseek -math-7b-instruct outperforms other baselines by 1-2 points on MATH-500.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.10539v1", "content": "In beam search experiments , our method trained on Deepseek -math-7b-instruct outperforms other baselines by 1-2 points on MATH-500."}
+{"idx": 2, "title": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for", "date": "", "ddg_snippet": "... on DeepSeekMath- Base with specialization in formal mathematical languages, the model undergoes supervised fine-tuning using an enhanced formal theorem ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2408.08152v1", "content": "... on DeepSeekMath- Base with specialization in formal mathematical languages, the model undergoes supervised fine-tuning using an enhanced formal theorem ..."}
+{"idx": 3, "title": "Deepseek", "date": "", "ddg_snippet": "... is not a super deep dive on Llama 2 (This is just necessary context), so I'll be giving a high-level view which is necessary for what DeepSeek is ...", "subpage_snippet": "", "source": "planetbanatt.net", "link": "https://planetbanatt.net/articles/deepseek.html", "content": "... is not a super deep dive on Llama 2 (This is just necessary context), so I'll be giving a high-level view which is necessary for what DeepSeek is ..."}
+{"idx": 4, "title": "Large Reasoning Models: How o1 Replications Turned into Real", "date": "", "ddg_snippet": "I want to get one thing out of the way first: the amount of press devoted to the $ 5 .5M figure for training DeepSeek -V3 was ridiculous.", "subpage_snippet": "", "source": "synthesis.ai", "link": "https://synthesis.ai/2025/02/25/large-reasoning-models-how-o1-replications-turned-into-real-competition/", "content": "I want to get one thing out of the way first: the amount of press devoted to the $ 5 .5M figure for training DeepSeek -V3 was ridiculous."}
+{"idx": 5, "title": "ExPO: Unlocking Hard Reasoning with Self-Explanation-Guided", "date": "", "ddg_snippet": "On the left, the bar plots of models (with the base model being Qwen2. 5 -3B-Instruct) evaluated on the MATH dataset, highlighting the issue with GRPO ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.02834v1", "content": "On the left, the bar plots of models (with the base model being Qwen2. 5 -3B-Instruct) evaluated on the MATH dataset, highlighting the issue with GRPO ..."}
+{"idx": 6, "title": "Semantically-Aware Rewards for Open-Ended R1 Training in", "date": "", "ddg_snippet": "Identifying the good and bad generations is the key to the success of rlvr to improve llm s’ abilities on structured rule-based tasks such as ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.15068v1", "content": "Identifying the good and bad generations is the key to the success of rlvr to improve llm s’ abilities on structured rule-based tasks such as ..."}
+{"idx": 7, "title": "I think there's two different things going on here:", "date": "", "ddg_snippet": "DeepSeek claims that the cold-start data is from DeepSeekV3, which is the model that has the $ 5 .5M pricetag. ... data , such as creative writing, ...", "subpage_snippet": "", "source": "news.ycombinator.com", "link": "https://news.ycombinator.com/item?id=42867899", "content": "DeepSeek claims that the cold-start data is from DeepSeekV3, which is the model that has the $ 5 .5M pricetag. ... data , such as creative writing, ..."}
+{"idx": 8, "title": "AI Safety II: Goodharting and Reward Hacking - Synthesis AI", "date": "", "ddg_snippet": "And with that, we come to adversarial goodharting examples; the last two sections are the ones that show interesting examples from present-day models .", "subpage_snippet": "", "source": "synthesis.ai", "link": "https://synthesis.ai/2025/05/08/ai-safety-ii-goodharting-and-reward-hacking/", "content": "And with that, we come to adversarial goodharting examples; the last two sections are the ones that show interesting examples from present-day models ."}
+{"idx": 9, "title": "Best Open Source LLMs of 2025 — Klu", "date": "", "ddg_snippet": "Once dominated by Llama 2 and Vicuna, the landscape is now enriched by innovative models emerging from community efforts and new foundation model ...", "subpage_snippet": "", "source": "klu.ai", "link": "https://klu.ai/blog/open-source-llm-models", "content": "Once dominated by Llama 2 and Vicuna, the landscape is now enriched by innovative models emerging from community efforts and new foundation model ..."}
diff --git a/data/sampled_jsons/Self-Refine_Madaan_application_domain_medical_imaging_general_NLP.jsonl b/data/sampled_jsons/Self-Refine_Madaan_application_domain_medical_imaging_general_NLP.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..542b1e7489503f8aec626430289b8d3bae6156a0
--- /dev/null
+++ b/data/sampled_jsons/Self-Refine_Madaan_application_domain_medical_imaging_general_NLP.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "DG-TTA: Out-of- domain medical image segmentation... | Science", "date": "", "ddg_snippet": "Domain Generalization (DG) aims to reduce domain shifts between domains to achieve promising performance on the unseen target domain , which has been widely practiced in medical image segmentation.", "subpage_snippet": "", "source": "science.hiddenlayer.app", "link": "https://science.hiddenlayer.app/c/computer-science/post/dg-tta-out-of-domain-medical-image-segmentation-through-domain-generalization-and-test-time-adaptation-arxiv-2312-06275v2-cs-57ce", "content": "Domain Generalization (DG) aims to reduce domain shifts between domains to achieve promising performance on the unseen target domain , which has been widely practiced in medical image segmentation."}
+{"idx": 1, "title": "Artificial intelligence in healthcare and medicine : clinical applications ...", "date": "", "ddg_snippet": "This methodology has revolutionized medical AI applications owing to its exceptional capacity to manage complexity, especially in image and sequence data analysis [15].", "subpage_snippet": "", "source": "eurjmedres.biomedcentral.com", "link": "https://eurjmedres.biomedcentral.com/articles/10.1186/s40001-025-03196-w", "content": "This methodology has revolutionized medical AI applications owing to its exceptional capacity to manage complexity, especially in image and sequence data analysis [15]."}
+{"idx": 2, "title": "(PDF) Multimodal Auto Validation For Self - Refinement in Web Agents", "date": "", "ddg_snippet": "Utilizing the self - refine method [ Madaan et al., 2023], our approach introduces a self-correcting.open- domain platform for web-based agents. In International Conference on Machine Learning, pages 3135–3144.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/384563089_Multimodal_Auto_Validation_For_Self-Refinement_in_Web_Agents", "content": "Utilizing the self - refine method [ Madaan et al., 2023], our approach introduces a self-correcting.open- domain platform for web-based agents. In International Conference on Machine Learning, pages 3135–3144."}
+{"idx": 3, "title": "Hsinghsudwal/ NLP -Text-Data-Classification | DagsHub", "date": "", "ddg_snippet": "General : natural language processing llm llms Data Domain : nlp Integration: git github.", "subpage_snippet": "", "source": "dagshub.com", "link": "https://dagshub.com/Hsinghsudwal/NLP-Text-Data-Classification", "content": "General : natural language processing llm llms Data Domain : nlp Integration: git github."}
+{"idx": 4, "title": "EMNLP 2021 - sotaro.io", "date": "", "ddg_snippet": "TLDR : We study image translatability of words, which we define as the translatable of words via images , by measuring intra- and inter-cluster ...", "subpage_snippet": "", "source": "sotaro.io", "link": "https://sotaro.io/tldrs/emnlp-2021", "content": "TLDR : We study image translatability of words, which we define as the translatable of words via images , by measuring intra- and inter-cluster ..."}
+{"idx": 5, "title": "Opportunities and challenges for ChatGPT and large language", "date": "", "ddg_snippet": "... domain , we performed a literature survey, exploring their potentials in a wide variety of different applications such as biomedical information ...", "subpage_snippet": "", "source": "academic.oup.com", "link": "https://academic.oup.com/bib/article/25/1/bbad493/7505071", "content": "... domain , we performed a literature survey, exploring their potentials in a wide variety of different applications such as biomedical information ..."}
+{"idx": 6, "title": "Proceedings of the 2022 Conference on Empirical Methods in", "date": "", "ddg_snippet": "... domain ; (2) an interactive online demo for codeless experimentation using our models; and (3) a tutorial covering a wide variety of typical social ...", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/volumes/2022.emnlp-demos/", "content": "... domain ; (2) an interactive online demo for codeless experimentation using our models; and (3) a tutorial covering a wide variety of typical social ..."}
+{"idx": 7, "title": "Wei Xu - ACL Anthology", "date": "", "ddg_snippet": "... deployed in consumer applications , they could lead to new privacy risks due to emergent abilities to identify people in photos, geolocate images , etc.", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/people/w/wei-xu/", "content": "... deployed in consumer applications , they could lead to new privacy risks due to emergent abilities to identify people in photos, geolocate images , etc."}
+{"idx": 8, "title": "A Survey on Data Synthesis and Augmentation for Large Language", "date": "", "ddg_snippet": "... is another concern, as biases and inaccuracies from original datasets can persist in synthetic data, limiting its generalization across domains ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.12896v1", "content": "... is another concern, as biases and inaccuracies from original datasets can persist in synthetic data, limiting its generalization across domains ..."}
+{"idx": 9, "title": "Correcting Hallucinations in News Summaries: Exploration of", "date": "", "ddg_snippet": "... self -correcting LLM methods tend to base the ... Application of iterative self -correcting methods to the news domain is still mostly missing.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.19607v1", "content": "... self -correcting LLM methods tend to base the ... Application of iterative self -correcting methods to the news domain is still mostly missing."}
diff --git a/data/sampled_jsons/Sharpness-Aware_Minimization_for_Efficiently_Improving_Generalization_Foret_et_al._abstract.jsonl b/data/sampled_jsons/Sharpness-Aware_Minimization_for_Efficiently_Improving_Generalization_Foret_et_al._abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5e7eb6130ad13bad2575edd38bee98fd2f8abe1b
--- /dev/null
+++ b/data/sampled_jsons/Sharpness-Aware_Minimization_for_Efficiently_Improving_Generalization_Foret_et_al._abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Sharpness-Aware Minimization for Efficiently Improving Generalization", "date": "", "ddg_snippet": "View a PDF of the paper titled Sharpness - Aware Minimization for Efficiently Improving Generalization , by Pierre Foret and 3 other authors", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2010.01412", "content": "View a PDF of the paper titled Sharpness - Aware Minimization for Efficiently Improving Generalization , by Pierre Foret and 3 other authors"}
+{"idx": 1, "title": "ICLR 2021 Sharpness-aware Minimization for Efficiently Improving ...", "date": "", "ddg_snippet": "Spotlight Sharpness - aware Minimization for Efficiently Improving Generalization Pierre Foret · Ariel Kleiner · Hossein Mobahi · Behnam Neyshabur [ Abstract ] [ Visit Oral Session 2 ] [ Paper ] [ Paper ]", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2021/spotlight/3497", "content": "Spotlight Sharpness - aware Minimization for Efficiently Improving Generalization Pierre Foret · Ariel Kleiner · Hossein Mobahi · Behnam Neyshabur [ Abstract ] [ Visit Oral Session 2 ] [ Paper ] [ Paper ]"}
+{"idx": 2, "title": "[Paper Review] Sharpness-Aware Minimization for Efficiently Improving ...", "date": "", "ddg_snippet": "Outlines References 1. Weak Generalization Power of Sharp Minima 2. Sharpness - Aware Minimization (SAM) 2.1. PAC Bayesian Generalization Bound 2.2. SAM Objective 3. Empirical Evaluation References Sharpness - Aware Minimization for Efficiently Improving Generalization , Foret et al , 2021 Proving that the dual of the lp norm is the lq norm An Introduction to PAC-Bayes 1. Weak Generalization Power ...", "subpage_snippet": "", "source": "suminizz.github.io", "link": "https://suminizz.github.io/sam/", "content": "Outlines References 1. Weak Generalization Power of Sharp Minima 2. Sharpness - Aware Minimization (SAM) 2.1. PAC Bayesian Generalization Bound 2.2. SAM Objective 3. Empirical Evaluation References Sharpness - Aware Minimization for Efficiently Improving Generalization , Foret et al , 2021 Proving that the dual of the lp norm is the lq norm An Introduction to PAC-Bayes 1. Weak Generalization Power ..."}
+{"idx": 3, "title": "Sharpness-aware Minimization for Efficiently Improving Generalization", "date": "", "ddg_snippet": "Indeed, optimizing only the training loss value, as is commonly done, can easily lead to suboptimal model quality. Motivated by prior work connecting the geometry of the loss landscape and generalization , we introduce a novel, effective procedure for instead simultaneously minimizing loss value and loss sharpness .", "subpage_snippet": "", "source": "research.google", "link": "https://research.google/pubs/sharpness-aware-minimization-for-efficiently-improving-generalization/", "content": "Indeed, optimizing only the training loss value, as is commonly done, can easily lead to suboptimal model quality. Motivated by prior work connecting the geometry of the loss landscape and generalization , we introduce a novel, effective procedure for instead simultaneously minimizing loss value and loss sharpness ."}
+{"idx": 4, "title": "Sharpness-Aware Minimization for Efficiently Improving Generalization", "date": "", "ddg_snippet": "This work introduces a novel, effective procedure for simultaneously minimizing loss value and loss sharpness , Sharpness - Aware Minimization (SAM), which improves model generalization across a variety of benchmark datasets and models, yielding novel state-of-the-art performance for several. In today's heavily overparameterized models, the value of the training loss provides few guarantees on ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Sharpness-Aware-Minimization-for-Efficiently-Foret-Kleiner/a2cd073b57be744533152202989228cb4122270a", "content": "This work introduces a novel, effective procedure for simultaneously minimizing loss value and loss sharpness , Sharpness - Aware Minimization (SAM), which improves model generalization across a variety of benchmark datasets and models, yielding novel state-of-the-art performance for several. In today's heavily overparameterized models, the value of the training loss provides few guarantees on ..."}
+{"idx": 5, "title": "Understanding Sharpness-aware Minimization", "date": "", "ddg_snippet": "ABSTRACT Sharpness - Aware Minimization (SAM) is a recent training method that relies on worst-case weight perturbations. SAM significantly improves generalization in various settings, however, existing justifications for its success do not seem conclu-sive. First, we analyze the implicit bias of SAM over diagonal linear networks, and prove that it always chooses a solution that enjoys better ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=qXa0nhTRZGV", "content": "ABSTRACT Sharpness - Aware Minimization (SAM) is a recent training method that relies on worst-case weight perturbations. SAM significantly improves generalization in various settings, however, existing justifications for its success do not seem conclu-sive. First, we analyze the implicit bias of SAM over diagonal linear networks, and prove that it always chooses a solution that enjoys better ..."}
+{"idx": 6, "title": "Sharpness-Aware Minimization for Efficiently Improving Generalization", "date": "", "ddg_snippet": "In particular, our procedure, Sharpness - Aware Minimization (SAM), seeks parameters that lie in neighborhoods having uniformly low loss; this formulation results in a min-max optimization problem on which gradient descent can be performed efficiently .", "subpage_snippet": "", "source": "ui.adsabs.harvard.edu", "link": "https://ui.adsabs.harvard.edu/abs/2020arXiv201001412F/abstract", "content": "In particular, our procedure, Sharpness - Aware Minimization (SAM), seeks parameters that lie in neighborhoods having uniformly low loss; this formulation results in a min-max optimization problem on which gradient descent can be performed efficiently ."}
+{"idx": 7, "title": "Sharpness-Aware Minimization for Efficiently Improving Generalization", "date": "", "ddg_snippet": "The connection between the geometry of the loss landscape—in particular, the flatness of minima—and generalization has been studied extensively from both theoretical and empirical perspectives (Shirish Keskar et al ., 2016; Dziugaite & Roy, 2017; Jiang et al ., 2019). While this connection has held the promise of enabling new approaches to model training that yield better generalization ...", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2010.01412", "content": "The connection between the geometry of the loss landscape—in particular, the flatness of minima—and generalization has been studied extensively from both theoretical and empirical perspectives (Shirish Keskar et al ., 2016; Dziugaite & Roy, 2017; Jiang et al ., 2019). While this connection has held the promise of enabling new approaches to model training that yield better generalization ..."}
+{"idx": 8, "title": "Sharpness-aware minimization and the edge of stability", "date": "", "ddg_snippet": "Pierre Foret , Ariel Kleiner, Hossein Mobahi, and Behnam Neyshabur. Sharpness - aware minimization for efficiently improving generalization . In International Conference on Learning Representations, 2020.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/10.5555/3722577.3722756", "content": "Pierre Foret , Ariel Kleiner, Hossein Mobahi, and Behnam Neyshabur. Sharpness - aware minimization for efficiently improving generalization . In International Conference on Learning Representations, 2020."}
+{"idx": 9, "title": "arXiv:2010.01412v3 [cs.LG] 29 Apr 2021", "date": "", "ddg_snippet": "(Shirish Keskar et al ., 2016; Dziugaite & Roy, 2017; Jiang et al ., 2019). While this connection has held the promise of enabling new approaches to model training that yield better generalization , practical efficient algorithms that specifically seek out flatter minima and furthermore effectively improve generalization on a range of state-of-the-art models have thus far been elusive (e.g., see ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2010.01412", "content": "(Shirish Keskar et al ., 2016; Dziugaite & Roy, 2017; Jiang et al ., 2019). While this connection has held the promise of enabling new approaches to model training that yield better generalization , practical efficient algorithms that specifically seek out flatter minima and furthermore effectively improve generalization on a range of state-of-the-art models have thus far been elusive (e.g., see ..."}
diff --git a/data/sampled_jsons/Stochastic_Gradient_Langevin_Dynamics_SGLD_noise_tensor_shape.jsonl b/data/sampled_jsons/Stochastic_Gradient_Langevin_Dynamics_SGLD_noise_tensor_shape.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..240cb8b4eb2515dce91d01cf5b239a4d18a4abb9
--- /dev/null
+++ b/data/sampled_jsons/Stochastic_Gradient_Langevin_Dynamics_SGLD_noise_tensor_shape.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "「Stochastic」与「Random」有何区别？ - 知乎", "date": "", "ddg_snippet": "With stochastic process, the likelihood or probability of any particular outcome can be specified and not all outcomes are equally likely of occurring. For example, an ornithologist may assign a greater probability that a bird will select a nesting location based on how far it is from the edge of the refuge or whether the location is shielded ...", "subpage_snippet": "", "source": "www.zhihu.com", "link": "https://www.zhihu.com/question/20675303?sort=created", "content": "With stochastic process, the likelihood or probability of any particular outcome can be specified and not all outcomes are equally likely of occurring. For example, an ornithologist may assign a greater probability that a bird will select a nesting location based on how far it is from the edge of the refuge or whether the location is shielded ..."}
+{"idx": 1, "title": "In layman's terms: What is a stochastic process?", "date": "", "ddg_snippet": "Oct 8, 2015 · A stochastic process is a way of representing the evolution of some situation that can be characterized mathematically (by numbers, points in a graph, etc.) over time.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/1470686/in-laymans-terms-what-is-a-stochastic-process", "content": "Oct 8, 2015 · A stochastic process is a way of representing the evolution of some situation that can be characterized mathematically (by numbers, points in a graph, etc.) over time."}
+{"idx": 2, "title": "Books recommendations on stochastic analysis - Mathematics Stack...", "date": "", "ddg_snippet": "Feb 21, 2023 · Stochastic Calculus for Finance I: Binomial asset pricing model and Stochastic Calculus for Finance II: tochastic Calculus for Finance II: Continuous-Time Models. These two books are very good if you want to apply the theory to price derivatives. Stochastic Differential Equations: An Introduction with Applications Bernt Oksanda.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/4643337/books-recommendations-on-stochastic-analysis", "content": "Feb 21, 2023 · Stochastic Calculus for Finance I: Binomial asset pricing model and Stochastic Calculus for Finance II: tochastic Calculus for Finance II: Continuous-Time Models. These two books are very good if you want to apply the theory to price derivatives. Stochastic Differential Equations: An Introduction with Applications Bernt Oksanda."}
+{"idx": 3, "title": "Difference between time series and stochastic process?", "date": "", "ddg_snippet": "Jan 30, 2011 · Stochastic processes are often used in modeling time series data- we assume that the time series we have was produced by a stochastic process, find the parameters of a stochastic process that would be likely to produce that time series, and then use that stochastic process as a model in predicting future values of the time series.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/19568/difference-between-time-series-and-stochastic-process", "content": "Jan 30, 2011 · Stochastic processes are often used in modeling time series data- we assume that the time series we have was produced by a stochastic process, find the parameters of a stochastic process that would be likely to produce that time series, and then use that stochastic process as a model in predicting future values of the time series."}
+{"idx": 4, "title": "What's the difference between stochastic and random?", "date": "", "ddg_snippet": "Feb 28, 2012 · Similarly \" stochastic process\" and \"random process\", but the former is seen more often. Some mathematicians seem to use \"random\" when they mean uniformly distributed, but probabilists and statisticians don't.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/114373/whats-the-difference-between-stochastic-and-random", "content": "Feb 28, 2012 · Similarly \" stochastic process\" and \"random process\", but the former is seen more often. Some mathematicians seem to use \"random\" when they mean uniformly distributed, but probabilists and statisticians don't."}
+{"idx": 5, "title": "What is Ito's formula for a function of two stochastic processes?", "date": "", "ddg_snippet": "Explore related questions reference-request stochastic -processes stochastic -calculus stochastic -differential-equations See similar questions with these tags.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/1351527/what-is-itos-formula-for-a-function-of-two-stochastic-processes", "content": "Explore related questions reference-request stochastic -processes stochastic -calculus stochastic -differential-equations See similar questions with these tags."}
+{"idx": 6, "title": "Fubini's Theorem for Stochastic Integral - Mathematics Stack...", "date": "", "ddg_snippet": "Jul 6, 2015 · In your equation with the question mark, those are not stochastic integrals and Fubini's theorem applies directly.", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/1351342/fubinis-theorem-for-stochastic-integral", "content": "Jul 6, 2015 · In your equation with the question mark, those are not stochastic integrals and Fubini's theorem applies directly."}
+{"idx": 7, "title": "如何理解随机梯度下降（stochastic gradient descent，SGD）？", "date": "", "ddg_snippet": "如何理解随机梯度下降（stochastic gradient descent，SGD）？ 圆桌收录 编程没有那么难 小蓝星 · undefined", "subpage_snippet": "", "source": "www.zhihu.com", "link": "https://www.zhihu.com/question/264189719", "content": "如何理解随机梯度下降（stochastic gradient descent，SGD）？ 圆桌收录 编程没有那么难 小蓝星 · undefined"}
+{"idx": 8, "title": "Where to begin in approaching Stochastic Calculus?", "date": "", "ddg_snippet": "Nov 6, 2012 · 18 I have experience in Abstract algebra (up to Galois theory), Real Analysis (baby Rudin except for the measure integral) and probability theory up to Brownian motion (non-rigorous treatment). Is there a suggested direction I can take in order to begin studying stochastic calculus and stochastic differential equations?", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/231712/where-to-begin-in-approaching-stochastic-calculus", "content": "Nov 6, 2012 · 18 I have experience in Abstract algebra (up to Galois theory), Real Analysis (baby Rudin except for the measure integral) and probability theory up to Brownian motion (non-rigorous treatment). Is there a suggested direction I can take in order to begin studying stochastic calculus and stochastic differential equations?"}
+{"idx": 9, "title": "stochastic processes - Ito's chain rule and multidimensional...", "date": "", "ddg_snippet": "Oct 9, 2020 · In the book \"Quant Job Interview: Questions and Answers\" by M. Joshi in the derivation of the final Black-Scholes formula he makes usage of Ito's chain rule. To get specific, he goes from...", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/3858251/itos-chain-rule-and-multidimensional-derivation", "content": "Oct 9, 2020 · In the book \"Quant Job Interview: Questions and Answers\" by M. Joshi in the derivation of the final Black-Scholes formula he makes usage of Ito's chain rule. To get specific, he goes from..."}
diff --git a/data/sampled_jsons/StreamRF_Streaming_Radiance_Fields_for_Dynamic_Scene_Reconstruction_abstract.jsonl b/data/sampled_jsons/StreamRF_Streaming_Radiance_Fields_for_Dynamic_Scene_Reconstruction_abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..674a7028d6081def770a585609bc084d386a11a4
--- /dev/null
+++ b/data/sampled_jsons/StreamRF_Streaming_Radiance_Fields_for_Dynamic_Scene_Reconstruction_abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "QUEEN: QUantized Efficient ENcoding of Dynamic Gaussians for", "date": "", "ddg_snippet": "NeRFs have also been extended to dynamic 4D scenes [ 1 , 5 , 19 ] providing a powerful tool for reconstructing FVV.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.04469v1", "content": "NeRFs have also been extended to dynamic 4D scenes [ 1 , 5 , 19 ] providing a powerful tool for reconstructing FVV."}
+{"idx": 1, "title": "Real-time Photorealistic Dynamic Scene Representation and", "date": "", "ddg_snippet": "... real-world applications often capture dynamic scenes as monocular videos, making it impractical to train separate static scene representations for ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2310.10642v3", "content": "... real-world applications often capture dynamic scenes as monocular videos, making it impractical to train separate static scene representations for ..."}
+{"idx": 2, "title": "50 is What Percent of 200 ? = 25% | calcforme.com", "date": "", "ddg_snippet": "Online percentage calculator, 50 is What Percent of 200 ? = 25%. Easiest percentage calculator.", "subpage_snippet": "", "source": "calcforme.com", "link": "https://calcforme.com/percentage-calculator/50-is-what-percent-of-200", "content": "Online percentage calculator, 50 is What Percent of 200 ? = 25%. Easiest percentage calculator."}
+{"idx": 3, "title": "What is 50 / 200 as a percent ? - Calculatio", "date": "", "ddg_snippet": "According to 'Fraction to Percentage ' conversion formula if you want to know what percent of 200 is 50 you have to divide 50 by 200 and then multiply the result by 100.", "subpage_snippet": "", "source": "calculat.io", "link": "https://calculat.io/en/number/percentage/50--200", "content": "According to 'Fraction to Percentage ' conversion formula if you want to know what percent of 200 is 50 you have to divide 50 by 200 and then multiply the result by 100."}
+{"idx": 4, "title": "50 / 200 : 50 Out of 200 as a Percentage - getcalc.com", "date": "", "ddg_snippet": "50 out of 200 as a percentage provides the quick answer for what percent is 50 of 200, along with more insight of how to find the percentage and what are all the different variations of real world problems. 25% is the calculated percentage.", "subpage_snippet": "", "source": "getcalc.com", "link": "https://getcalc.com/math-50of200-percentage.htm", "content": "50 out of 200 as a percentage provides the quick answer for what percent is 50 of 200, along with more insight of how to find the percentage and what are all the different variations of real world problems. 25% is the calculated percentage."}
+{"idx": 5, "title": "50 is what percent of 200 ? - ClickCalculators.com", "date": "", "ddg_snippet": "50 is 25 percent of 200. See detailed information with steps. Learn how to calculate percentages with step-by-step solution of example questions.", "subpage_snippet": "", "source": "clickcalculators.com", "link": "https://clickcalculators.com/percentage/50/is-what-percent-of/200", "content": "50 is 25 percent of 200. See detailed information with steps. Learn how to calculate percentages with step-by-step solution of example questions."}
+{"idx": 6, "title": "50 is what percent of 200 ? - Everydaycalculation.com", "date": "", "ddg_snippet": "What percent is 50 of 200? The answer is 25% . Get stepwise instructions to work out \"50 is what percent of 200?\"", "subpage_snippet": "", "source": "answers.everydaycalculation.com", "link": "https://answers.everydaycalculation.com/percent-is/50-200", "content": "What percent is 50 of 200? The answer is 25% . Get stepwise instructions to work out \"50 is what percent of 200?\""}
+{"idx": 7, "title": "50 is What Percent of 200 ? - CalculateMe.com", "date": "", "ddg_snippet": "Use this calculator to find 50 / 200 as a percentage .", "subpage_snippet": "", "source": "www.calculateme.com", "link": "https://www.calculateme.com/math/calculate-a-percent/what-percent-of-200/50", "content": "Use this calculator to find 50 / 200 as a percentage ."}
+{"idx": 8, "title": "50 is what percent of 200 ? - Vocab Dictionary", "date": "", "ddg_snippet": "Percentage = (50 / 200) × 100. Calculating the fraction gives: Percentage = 0.25 × 100. This simplifies to: Percentage = 25. So, 50 is 25% of 200.", "subpage_snippet": "", "source": "vocabdictionary.com", "link": "https://vocabdictionary.com/explained/50-is-what-percent-of-200/", "content": "Percentage = (50 / 200) × 100. Calculating the fraction gives: Percentage = 0.25 × 100. This simplifies to: Percentage = 25. So, 50 is 25% of 200."}
+{"idx": 9, "title": "50 is what % of 200 - Online Calculator", "date": "", "ddg_snippet": "Calculate what percentage 50 is of 200. 50 is 25% of 200. Learn how to find the percent one number is of another with our free calculator and step-by-step guide.", "subpage_snippet": "", "source": "calculatorhistory.net", "link": "https://calculatorhistory.net/en/percent/50-is-what-percent-of-200", "content": "Calculate what percentage 50 is of 200. 50 is 25% of 200. Learn how to find the percent one number is of another with our free calculator and step-by-step guide."}
diff --git a/data/sampled_jsons/StreamRF_paper_abstract_dynamic_scene_reconstruction_year_2023.jsonl b/data/sampled_jsons/StreamRF_paper_abstract_dynamic_scene_reconstruction_year_2023.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5d0152ce880b5441be846b3168170c61b9619ad9
--- /dev/null
+++ b/data/sampled_jsons/StreamRF_paper_abstract_dynamic_scene_reconstruction_year_2023.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Real-time Photorealistic Dynamic Scene Representation and", "date": "", "ddg_snippet": "Reconstructing dynamic 3D scenes from 2D images and generating diverse views over time is challenging due to scene complexity and temporal dynamics .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2310.10642v3", "content": "Reconstructing dynamic 3D scenes from 2D images and generating diverse views over time is challenging due to scene complexity and temporal dynamics ."}
+{"idx": 1, "title": "4D Gaussian Splatting: Modeling Dynamic Scenes with Native 4D", "date": "", "ddg_snippet": "In this paper , we frame dynamic scenes as a spatio-temporal 4D volume learning problem , offering a native explicit reformulation with minimal ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.20720v1", "content": "In this paper , we frame dynamic scenes as a spatio-temporal 4D volume learning problem , offering a native explicit reformulation with minimal ..."}
+{"idx": 2, "title": "QUEEN: QUantized Efficient ENcoding of Dynamic Gaussians for", "date": "", "ddg_snippet": "The underlying problem of reconstructing FVV involves learning a 6D plenoptic function of a dynamic scene P ( 𝒙 , 𝒅 , t ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.04469v1", "content": "The underlying problem of reconstructing FVV involves learning a 6D plenoptic function of a dynamic scene P ( 𝒙 , 𝒅 , t ..."}
+{"idx": 3, "title": "MGStream: Motion-aware 3D Gaussian for Streamable Dynamic ...", "date": "", "ddg_snippet": "In this paper , we introduce MGStream for streamable dynamic scene reconstruction . Specifically, MGStream locates the motion-related 3DGs, and employs them and the vanilla 3DGs for modeling the dynamic and static, respectively.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.13839v1", "content": "In this paper , we introduce MGStream for streamable dynamic scene reconstruction . Specifically, MGStream locates the motion-related 3DGs, and employs them and the vanilla 3DGs for modeling the dynamic and static, respectively."}
+{"idx": 4, "title": "S4D: Streaming 4D Real-World Reconstruction with... | OpenReview", "date": "", "ddg_snippet": "Abstract : Dynamic scene reconstruction using Gaussians has recently attracted increased interest. Mainstream approaches typically employ a global deformation field to warp a 3D scene in canonical space.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=k3Z8CnHdfg", "content": "Abstract : Dynamic scene reconstruction using Gaussians has recently attracted increased interest. Mainstream approaches typically employ a global deformation field to warp a 3D scene in canonical space."}
+{"idx": 5, "title": "GitHub - Lee-JaeWon/2024-Arxiv- Paper -List-Gaussian-Splatting: 2024...", "date": "", "ddg_snippet": "Dynamic scene reconstruction from monocular video is critical for real-world applications. This paper tackles the dual challenges of dynamic novel-view synthesis and 3D geometry reconstruction by introducing a hybrid framework...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/Lee-JaeWon/2024-Arxiv-Paper-List-Gaussian-Splatting", "content": "Dynamic scene reconstruction from monocular video is critical for real-world applications. This paper tackles the dual challenges of dynamic novel-view synthesis and 3D geometry reconstruction by introducing a hybrid framework..."}
+{"idx": 6, "title": "New Method for Dynamic Scene Reconstruction - Simple Science", "date": "", "ddg_snippet": "Abstract : Dynamic scene reconstruction using Gaussians has recently attracted increased interest. Mainstream approaches typically employ a global deformation field to warp a 3D scene in canonical space.", "subpage_snippet": "", "source": "scisimple.com", "link": "https://scisimple.com/en/articles/2025-06-23-new-method-for-dynamic-scene-reconstruction--ake6ldv", "content": "Abstract : Dynamic scene reconstruction using Gaussians has recently attracted increased interest. Mainstream approaches typically employ a global deformation field to warp a 3D scene in canonical space."}
+{"idx": 7, "title": "Daily Papers - Hugging Face", "date": "", "ddg_snippet": "Dynamic scene reconstruction from monocular video is critical for real-world applications. This paper tackles the dual challenges of dynamic novel-view synthesis and 3D geometry reconstruction by introducing a hybrid framework...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/papers?q=dynamic+scene+reconstruction", "content": "Dynamic scene reconstruction from monocular video is critical for real-world applications. This paper tackles the dual challenges of dynamic novel-view synthesis and 3D geometry reconstruction by introducing a hybrid framework..."}
+{"idx": 8, "title": "TransformerFusion: Monocular RGB Scene Reconstruction using...", "date": "", "ddg_snippet": "Abstract : We introduce TransformerFusion, a transformer-based 3D scene reconstruction approach.The feature grid is then decoded to a higher-resolution scene reconstruction , using an MLP-based surface occupancy prediction from interpolated coarse-to-fine 3D features.", "subpage_snippet": "", "source": "papertalk.org", "link": "https://papertalk.org/papertalks/35625", "content": "Abstract : We introduce TransformerFusion, a transformer-based 3D scene reconstruction approach.The feature grid is then decoded to a higher-resolution scene reconstruction , using an MLP-based surface occupancy prediction from interpolated coarse-to-fine 3D features."}
+{"idx": 9, "title": "MGStream: Motion-aware 3D Gaussian for Streamable Dynamic ...", "date": "", "ddg_snippet": "To tackle this, we introduce MGStream which employs the motion-related 3D Gaussians (3DGs) to reconstruct the dynamic and the vanilla 3DGs for the static. The motion-related 3DGs are implemented according to the motion mask and the clustering-based convex hull algorithm.", "subpage_snippet": "", "source": "zhenybao.github.io", "link": "https://zhenybao.github.io/MGStream/?s=09", "content": "To tackle this, we introduce MGStream which employs the motion-related 3D Gaussians (3DGs) to reconstruct the dynamic and the vanilla 3DGs for the static. The motion-related 3DGs are implemented according to the motion mask and the clustering-based convex hull algorithm."}
diff --git a/data/sampled_jsons/TOP-ERL_Transformer-based_Off-Policy_Episodic_Reinforcement_Learning.jsonl b/data/sampled_jsons/TOP-ERL_Transformer-based_Off-Policy_Episodic_Reinforcement_Learning.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c02a388123241a20235685f091685b69cf22cb1d
--- /dev/null
+++ b/data/sampled_jsons/TOP-ERL_Transformer-based_Off-Policy_Episodic_Reinforcement_Learning.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Coordinated management of meaning - Wikipedia", "date": "", "ddg_snippet": "Our social world can be understood through the practice of CMM through \"managing our meanings in our messages based off our values\".", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Coordinated_management_of_meaning", "content": "Our social world can be understood through the practice of CMM through \"managing our meanings in our messages based off our values\"."}
+{"idx": 1, "title": "robot-learning.ml | 7th Robot Learning Workshop: Towards Robots", "date": "", "ddg_snippet": "... Learning Workshop at ICLR 2025” at the top of your camera ready paper. ... TOP - ERL : Transformer - based Off - Policy Episodic Reinforcement Learning", "subpage_snippet": "", "source": "www.robot-learning.ml", "link": "https://www.robot-learning.ml/2025/", "content": "... Learning Workshop at ICLR 2025” at the top of your camera ready paper. ... TOP - ERL : Transformer - based Off - Policy Episodic Reinforcement Learning"}
+{"idx": 2, "title": "GitHub - opendilab/awesome-exploration-rl: A curated list of", "date": "", "ddg_snippet": "TOP - ERL : Transformer - based Off - Policy Episodic Reinforcement Learning ... Key: Reinforcement learning , Exploration in off - policy methods, Continuous ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/opendilab/awesome-exploration-rl", "content": "TOP - ERL : Transformer - based Off - Policy Episodic Reinforcement Learning ... Key: Reinforcement learning , Exploration in off - policy methods, Continuous ..."}
+{"idx": 3, "title": "Expiring contract XI: THREE bonafide Premier League greats, the", "date": "", "ddg_snippet": "Trent has been a topic of discussion for many seasons now, first lauded for his all-action full-back play and then derided for his poor defending.", "subpage_snippet": "", "source": "www.dailymail.co.uk", "link": "https://www.dailymail.co.uk/sport/football/article-13842323/Expiring-contract-XI-Premier-League-Barcelona-free-agents.html", "content": "Trent has been a topic of discussion for many seasons now, first lauded for his all-action full-back play and then derided for his poor defending."}
+{"idx": 4, "title": "Bayern Munich chief hits back at Dietmar Hamann after he", "date": "", "ddg_snippet": "Your details from Facebook will be used to provide you with tailored content, marketing and ads in line with our Privacy Policy .", "subpage_snippet": "", "source": "www.dailymail.co.uk", "link": "https://www.dailymail.co.uk/sport/football/article-13927443/Bayern-Munich-Dietmar-Hamann-Harry-Kane.html", "content": "Your details from Facebook will be used to provide you with tailored content, marketing and ads in line with our Privacy Policy ."}
+{"idx": 5, "title": "Downloads", "date": "", "ddg_snippet": "A Generalized Projected Bellman Error for Off - policy Value Estimation in Reinforcement Learning ... of Shallow Vision Transformers : Learning ...", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/Downloads/2023", "content": "A Generalized Projected Bellman Error for Off - policy Value Estimation in Reinforcement Learning ... of Shallow Vision Transformers : Learning ..."}
+{"idx": 6, "title": "Kristina McElheran on The Effects of AI on Workers and Firms", "date": "", "ddg_snippet": "... grounded way, in a very evidence- based way, this idea that we have to be careful about who gets left behind when all these transformations take place.", "subpage_snippet": "", "source": "www.twothinkminimum.com", "link": "https://www.twothinkminimum.com/kristina-mcelheran-on-automation-and-prediction-analytics/", "content": "... grounded way, in a very evidence- based way, this idea that we have to be careful about who gets left behind when all these transformations take place."}
+{"idx": 7, "title": "PNNL: Highlights: Physical & Computational Sciences", "date": "", "ddg_snippet": "Cooling Off Warming Trends in the Arctic A new study shows that emissions from trans-Arctic shipping could increase clouds, cooling the atmosphere ...", "subpage_snippet": "", "source": "www.pnnl.gov", "link": "https://www.pnnl.gov/science/highlights/highlights.asp?division=979", "content": "Cooling Off Warming Trends in the Arctic A new study shows that emissions from trans-Arctic shipping could increase clouds, cooling the atmosphere ..."}
+{"idx": 8, "title": "Vol. 38 No. 12: AAAI-24 Technical Tracks 12 | Proceedings of", "date": "", "ddg_snippet": "The conference scope included machine learning (deep learning , statistical learning , etc), natural language processing, computer vision, data mining ...", "subpage_snippet": "", "source": "ojs.aaai.org", "link": "https://ojs.aaai.org/index.php/AAAI/issue/view/587", "content": "The conference scope included machine learning (deep learning , statistical learning , etc), natural language processing, computer vision, data mining ..."}
+{"idx": 9, "title": "Hernia Guide: Symptoms, Causes, Treatment Options - Southlake", "date": "", "ddg_snippet": "Compared to other hernias, femoral hernias have a higher risk of becoming strangulated, cutting off blood supply to the intestine.", "subpage_snippet": "", "source": "montereycountyvirtualtours.com", "link": "https://montereycountyvirtualtours.com/article/hernia-guide-symptoms-causes-treatment-options-southlake-general-surgery", "content": "Compared to other hernias, femoral hernias have a higher risk of becoming strangulated, cutting off blood supply to the intestine."}
diff --git a/data/sampled_jsons/The_Value_of_Prediction_in_Identifying_the_Worst-Off_Equation_(2)_policy_value_Gaussian.jsonl b/data/sampled_jsons/The_Value_of_Prediction_in_Identifying_the_Worst-Off_Equation_(2)_policy_value_Gaussian.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..dfe68eb75778bd87ef9c6d178c506e71389c9d4b
--- /dev/null
+++ b/data/sampled_jsons/The_Value_of_Prediction_in_Identifying_the_Worst-Off_Equation_(2)_policy_value_Gaussian.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Selection of ground motion prediction equations for", "date": "", "ddg_snippet": "In the past decades hundreds of GMPEs have been developed for the prediction of PGA, which are gathered in a series of public reports [ 2 -6].", "subpage_snippet": "", "source": "www.extrica.com", "link": "https://www.extrica.com/article/20377", "content": "In the past decades hundreds of GMPEs have been developed for the prediction of PGA, which are gathered in a series of public reports [ 2 -6]."}
+{"idx": 1, "title": "Streaming Generated Gaussian Process Experts for Online", "date": "", "ddg_snippet": "... Gaussian Processes introduce a set of M ≪ N M\\ll N inducing points to efficiently summarize the training data, reducing the complexity to 𝒪 ( ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.03679v1", "content": "... Gaussian Processes introduce a set of M ≪ N M\\ll N inducing points to efficiently summarize the training data, reducing the complexity to 𝒪 ( ..."}
+{"idx": 2, "title": "p value - Inference and predictive models - Cross Validated", "date": "", "ddg_snippet": "Emmanuel Candes gave the 2017 Wald Lectures at the Joint Statistical Meetings entitled What's happening in Selective Inference? , which offers a ...", "subpage_snippet": "", "source": "stats.stackexchange.com", "link": "https://stats.stackexchange.com/questions/322502/inference-and-predictive-models", "content": "Emmanuel Candes gave the 2017 Wald Lectures at the Joint Statistical Meetings entitled What's happening in Selective Inference? , which offers a ..."}
+{"idx": 3, "title": "Testing the local void hypothesis using baryon acoustic", "date": "", "ddg_snippet": "... independent uncertainties, the 42 ... The very pronounced first acoustic peak in the CMB implies an excess clustering of matter on a particular scale.", "subpage_snippet": "", "source": "academic.oup.com", "link": "https://academic.oup.com/mnras/article/540/1/545/8129689", "content": "... independent uncertainties, the 42 ... The very pronounced first acoustic peak in the CMB implies an excess clustering of matter on a particular scale."}
+{"idx": 4, "title": "ICML 2020 Papers", "date": "", "ddg_snippet": "On the ( In )tractability of Computing Normalizing Constants for the Product of Determinantal Point Processes ... the Relation between Quality-Diversity ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2020/papers.html?filter=keywords", "content": "On the ( In )tractability of Computing Normalizing Constants for the Product of Determinantal Point Processes ... the Relation between Quality-Diversity ..."}
+{"idx": 5, "title": "ICML 2020 Papers", "date": "", "ddg_snippet": "On the ( In )tractability of Computing Normalizing Constants for the Product of Determinantal Point Processes ... the Relation between Quality-Diversity ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2020/papers.html", "content": "On the ( In )tractability of Computing Normalizing Constants for the Product of Determinantal Point Processes ... the Relation between Quality-Diversity ..."}
+{"idx": 6, "title": "NeurIPS 2021 Papers", "date": "", "ddg_snippet": "The Limitations of Large Width in Neural Networks: A Deep Gaussian Process Perspective ... The future is log- Gaussian : ResNets and their infinite ...", "subpage_snippet": "", "source": "neurips.cc", "link": "https://neurips.cc/virtual/2021/papers.html", "content": "The Limitations of Large Width in Neural Networks: A Deep Gaussian Process Perspective ... The future is log- Gaussian : ResNets and their infinite ..."}
+{"idx": 7, "title": "Downloads", "date": "", "ddg_snippet": "alpha$-IoU: A Family of Power Intersection ... Adversarial Attacks on Black Box Video Classifiers: Leveraging the Power of Geometric Transformations", "subpage_snippet": "", "source": "neurips.cc", "link": "https://neurips.cc/Downloads/2021", "content": "alpha$-IoU: A Family of Power Intersection ... Adversarial Attacks on Black Box Video Classifiers: Leveraging the Power of Geometric Transformations"}
+{"idx": 8, "title": "ICLR 2021 Schedule", "date": "", "ddg_snippet": "... the Role of Gradient-based Attribution Methods for Model ... Into the Wild with AudioScope: Unsupervised Audio-Visual Separation of On-Screen Sounds", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2021/calendar", "content": "... the Role of Gradient-based Attribution Methods for Model ... Into the Wild with AudioScope: Unsupervised Audio-Visual Separation of On-Screen Sounds"}
+{"idx": 9, "title": "ICLR 2024 Schedule", "date": "", "ddg_snippet": "The Cost of Scaling Down Large Language ... Achieving the Pareto Frontier of Regret Minimization and Best Arm Identification in Multi-Armed Bandits", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2024/calendar", "content": "The Cost of Scaling Down Large Language ... Achieving the Pareto Frontier of Regret Minimization and Best Arm Identification in Multi-Armed Bandits"}
diff --git "a/data/sampled_jsons/Tyurin_Richt\303\241rik_2024_abstract_wasteful_distributed_machine_learning_year_2024.jsonl" "b/data/sampled_jsons/Tyurin_Richt\303\241rik_2024_abstract_wasteful_distributed_machine_learning_year_2024.jsonl"
new file mode 100644
index 0000000000000000000000000000000000000000..29195fad01bd2d3dd223b23675d1a06e5ba47c23
--- /dev/null
+++ "b/data/sampled_jsons/Tyurin_Richt\303\241rik_2024_abstract_wasteful_distributed_machine_learning_year_2024.jsonl"
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Handling Device Heterogeneity in Federated Learning: The ...", "date": "", "ddg_snippet": "Aug 24, 2025 · Peter Richtárik is a professor of Computer Science at KAUST, Saudi Arabia, where he leads the Optimization and Machine Learning Lab. His research interests lie at the intersection of mathematics, computer science, machine learning , optimization, numerical linear algebra, and high-performance computing. Through his work on randomized and distributed optimization algorithms, he has contributed ...", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/full/10.1145/3709023.3737695", "content": "Aug 24, 2025 · Peter Richtárik is a professor of Computer Science at KAUST, Saudi Arabia, where he leads the Optimization and Machine Learning Lab. His research interests lie at the intersection of mathematics, computer science, machine learning , optimization, numerical linear algebra, and high-performance computing. Through his work on randomized and distributed optimization algorithms, he has contributed ..."}
+{"idx": 1, "title": "cherryATA", "date": "", "ddg_snippet": "Abstract . Asynchronous methods are fundamental for par- allelizing computations in distributed machine learning . They aim to accelerate training by fully.", "subpage_snippet": "", "source": "www.arxiv.org", "link": "https://www.arxiv.org/pdf/2502.00775v1", "content": "Abstract . Asynchronous methods are fundamental for par- allelizing computations in distributed machine learning . They aim to accelerate training by fully."}
+{"idx": 2, "title": "Freya PAGE: First Optimal Time Complexity for Large- ...", "date": "", "ddg_snippet": "9 Dec 2024 — In real-world distributed systems used for large-scale machine learning ... Tyurin and Richtárik [2023], Tyurin et al. [ 2024 ] for ...", "subpage_snippet": "", "source": "neurips.cc", "link": "https://neurips.cc/virtual/2024/poster/96254", "content": "9 Dec 2024 — In real-world distributed systems used for large-scale machine learning ... Tyurin and Richtárik [2023], Tyurin et al. [ 2024 ] for ..."}
+{"idx": 3, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "2 Feb 2025 — In a recent breakthrough, Tyurin & Richtárik ( 2024 ) recently developed a parallel SGD method, optimal in terms of a novel notion of complexity ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.00775v1", "content": "2 Feb 2025 — In a recent breakthrough, Tyurin & Richtárik ( 2024 ) recently developed a parallel SGD method, optimal in terms of a novel notion of complexity ..."}
+{"idx": 4, "title": "Enhancing Low-Precision Sampling via Stochastic ...", "date": "", "ddg_snippet": "by Z Wang · Cited by 3 — Our study highlights the potential of low-precision SGHMC as an efficient and accurate sampling method for large-scale and resource-limited machine learning . 1 ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=uSLNzzuiDJ", "content": "by Z Wang · Cited by 3 — Our study highlights the potential of low-precision SGHMC as an efficient and accurate sampling method for large-scale and resource-limited machine learning . 1 ..."}
+{"idx": 5, "title": "Freya PAGE: First Optimal Time Complexity for Large- ...", "date": "", "ddg_snippet": "by A Tyurin · 2024 · Cited by 5 — Tyurin , and P. Richtárik . Permutation compressors for provably faster distributed nonconvex optimization. In International Conference on Learning ... 49 pages", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/618c8af8efd19b4ce90b8864a764d0fa-Paper-Conference.pdf", "content": "by A Tyurin · 2024 · Cited by 5 — Tyurin , and P. Richtárik . Permutation compressors for provably faster distributed nonconvex optimization. In International Conference on Learning ... 49 pages"}
+{"idx": 6, "title": "ATA: Adaptive Task Allocation for Efficient Resource ...", "date": "", "ddg_snippet": "in Distributed Machine Learning Artavazd Maranjyan El Mehdi Saad Peter Richtárik Francesco Orabona Abstract Asynchronous methods are fundamental for parallelizing computations in distributed machine learning . They aim to accelerate training by fully utilizing all available resources.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.00775v2", "content": "in Distributed Machine Learning Artavazd Maranjyan El Mehdi Saad Peter Richtárik Francesco Orabona Abstract Asynchronous methods are fundamental for parallelizing computations in distributed machine learning . They aim to accelerate training by fully utilizing all available resources."}
+{"idx": 7, "title": "Momentum Provably Improves Error Feedback! - Research Collection", "date": "", "ddg_snippet": "Due to the high communication overhead when training machine learning models in a distributed environment, modern algorithms invariably rely on lossy communication compression. However, when untreated, the errors caused by compression propagate, and can lead to severely unstable behavior, including Show more Publication status published ...", "subpage_snippet": "", "source": "www.research-collection.ethz.ch", "link": "https://www.research-collection.ethz.ch/handle/20.500.11850/638302", "content": "Due to the high communication overhead when training machine learning models in a distributed environment, modern algorithms invariably rely on lossy communication compression. However, when untreated, the errors caused by compression propagate, and can lead to severely unstable behavior, including Show more Publication status published ..."}
+{"idx": 8, "title": "Peter Richtarik", "date": "", "ddg_snippet": "Abstract : Asynchronous Stochastic Gradient Descent (Asynchronous SGD) is a cornerstone method for parallelizing learning in distributed machine learning . However, its performance suffers under arbitrarily heterogeneous computation times across workers, leading to suboptimal time complexity and inefficiency as the number of workers scales.", "subpage_snippet": "", "source": "richtarik.org", "link": "https://richtarik.org/", "content": "Abstract : Asynchronous Stochastic Gradient Descent (Asynchronous SGD) is a cornerstone method for parallelizing learning in distributed machine learning . However, its performance suffers under arbitrarily heterogeneous computation times across workers, leading to suboptimal time complexity and inefficiency as the number of workers scales."}
+{"idx": 9, "title": "Freya PAGE: First Optimal Time Complexity for Large-Scale ...", "date": "", "ddg_snippet": "A somewhat similar expression involving minj∈[n] and harmonic means was obtained by Tyurin and Richtárik [2023], Tyurin et al. [ 2024 ] for minimizing expectation under the bounded vari-ance assumption.", "subpage_snippet": "", "source": "papers.nips.cc", "link": "https://papers.nips.cc/paper_files/paper/2024/file/618c8af8efd19b4ce90b8864a764d0fa-Paper-Conference.pdf", "content": "A somewhat similar expression involving minj∈[n] and harmonic means was obtained by Tyurin and Richtárik [2023], Tyurin et al. [ 2024 ] for minimizing expectation under the bounded vari-ance assumption."}
diff --git a/data/sampled_jsons/VRSBench_Table_1_image_size_comparison_year_2024.jsonl b/data/sampled_jsons/VRSBench_Table_1_image_size_comparison_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..69d8dc1d9ee3c45b1b3c44d9c224e5fc156d50c0
--- /dev/null
+++ b/data/sampled_jsons/VRSBench_Table_1_image_size_comparison_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "AeroLite: Tag-Guided Lightweight Generation of Aerial Image", "date": "", "ddg_snippet": "... to equip small-scale language models ( 1 –3B parameters) with robust and interpretable captioning capabilities specifically for remote sensing images ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.09528v1", "content": "... to equip small-scale language models ( 1 –3B parameters) with robust and interpretable captioning capabilities specifically for remote sensing images ..."}
+{"idx": 1, "title": "VRSBench : A Versatile Vision-Language Benchmark", "date": "", "ddg_snippet": "Table 1 : Comparison between existing remote sensing vision-language datasets and our VRSBench dataset.Table 3 shows the comparative performance of different methods in detailed image caption-ing of our VRSBench dataset.", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/05b7f821234f66b78f99e7803fffa78a-Paper-Datasets_and_Benchmarks_Track.pdf", "content": "Table 1 : Comparison between existing remote sensing vision-language datasets and our VRSBench dataset.Table 3 shows the comparative performance of different methods in detailed image caption-ing of our VRSBench dataset."}
+{"idx": 2, "title": "VRSBench", "date": "", "ddg_snippet": "Table 1 . Comparison between existing remote sensing vision-language datasets and our VRSBench dataset.Table 2. Detailed image caption performance on VRSBench dataset.", "subpage_snippet": "", "source": "vrsbench.github.io", "link": "https://vrsbench.github.io/", "content": "Table 1 . Comparison between existing remote sensing vision-language datasets and our VRSBench dataset.Table 2. Detailed image caption performance on VRSBench dataset."}
+{"idx": 3, "title": "xiang709/ VRSBench · Datasets at Hugging Face", "date": "", "ddg_snippet": "return cast_ table _to_schema( table , schema). VRSBench is a Versatile Vision-Language Benchmark for Remote Sensing Image Understanding. It consists of 29,614 remote sensing images with detailed captions, 52,472 object refers, and 3123,221 visual question-answer pairs.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/datasets/xiang709/VRSBench", "content": "return cast_ table _to_schema( table , schema). VRSBench is a Versatile Vision-Language Benchmark for Remote Sensing Image Understanding. It consists of 29,614 remote sensing images with detailed captions, 52,472 object refers, and 3123,221 visual question-answer pairs."}
+{"idx": 4, "title": "GitHub - lx709/ VRSBench", "date": "", "ddg_snippet": "Solutions. By company size . VRSBench is a Versatile Vision-Language Benchmark for Remote Sensing Image Understanding. It consists of 29,614 remote sensing images with detailed captions, 52,472 object refers, and 3123,221 visual question-answer pairs.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/lx709/VRSBench", "content": "Solutions. By company size . VRSBench is a Versatile Vision-Language Benchmark for Remote Sensing Image Understanding. It consists of 29,614 remote sensing images with detailed captions, 52,472 object refers, and 3123,221 visual question-answer pairs."}
+{"idx": 5, "title": "Introducing VRSBench : Advancing Remote Sensing Image Analysis", "date": "", "ddg_snippet": "Table of Contents.Each image in VRSBench is accompanied by a detailed caption. These captions provide a comprehensive description of both general aspects and specific features within the image .", "subpage_snippet": "", "source": "scisimple.com", "link": "https://scisimple.com/en/articles/2025-07-27-introducing-vrsbench-advancing-remote-sensing-image-analysis--ak6l6or", "content": "Table of Contents.Each image in VRSBench is accompanied by a detailed caption. These captions provide a comprehensive description of both general aspects and specific features within the image ."}
+{"idx": 6, "title": "VRSBench : A Versatile Vision-Language Benchmark Dataset for...", "date": "", "ddg_snippet": "• VRSBench contains a large collection of remote sensing images paired with natural language descriptions, providing a versatile benchmark for tasks like image captioning, visual question answering, and multimodal reasoning.", "subpage_snippet": "", "source": "www.aimodels.fyi", "link": "https://www.aimodels.fyi/papers/arxiv/vrsbench-versatile-vision-language-benchmark-dataset-remote", "content": "• VRSBench contains a large collection of remote sensing images paired with natural language descriptions, providing a versatile benchmark for tasks like image captioning, visual question answering, and multimodal reasoning."}
+{"idx": 7, "title": "Transform text into images and explore with endless imagination.", "date": "", "ddg_snippet": "A generated image based on your input prompt. flip_camera_androidFlip card.", "subpage_snippet": "", "source": "labs.google", "link": "https://labs.google/fx/tools/image-fx/unsupported-country", "content": "A generated image based on your input prompt. flip_camera_androidFlip card."}
+{"idx": 8, "title": "AI Image Generator (free, no sign-up, unlimited)", "date": "", "ddg_snippet": "It's an AI-based image generator - i .e. a text-to- image model. No watermark, no account needed, unlimited images . Type words, make pics.", "subpage_snippet": "", "source": "perchance.org", "link": "https://perchance.org/ai-text-to-image-generator", "content": "It's an AI-based image generator - i .e. a text-to- image model. No watermark, no account needed, unlimited images . Type words, make pics."}
+{"idx": 9, "title": "Image Size Finder | Free Tool | PosterBurner", "date": "", "ddg_snippet": "Find the size and pixel depth of your image . This free tool shows you the height, width, and pixel per inch of your image . Works great on mobile and desktop.", "subpage_snippet": "", "source": "www.posterburner.com", "link": "https://www.posterburner.com/Image-Size-Finder.aspx", "content": "Find the size and pixel depth of your image . This free tool shows you the height, width, and pixel per inch of your image . Works great on mobile and desktop."}
diff --git a/data/sampled_jsons/VRSBench_dataset_image_resolution.jsonl b/data/sampled_jsons/VRSBench_dataset_image_resolution.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..68542de2dd6a31df4980a4efad3db0a7eb35e2f5
--- /dev/null
+++ b/data/sampled_jsons/VRSBench_dataset_image_resolution.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "GitHub - lx709/VRSBench", "date": "", "ddg_snippet": "Jun 19, 2024 · Attribute Extraction: we extract image information, including the source and resolution , as well as object information—such as the object category, bounding box, color, position (absolute and relative), and size (absolute and relative)—from existing object detection datasets .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/lx709/VRSBench", "content": "Jun 19, 2024 · Attribute Extraction: we extract image information, including the source and resolution , as well as object information—such as the object category, bounding box, color, position (absolute and relative), and size (absolute and relative)—from existing object detection datasets ."}
+{"idx": 1, "title": "xiang709/VRSBench · Datasets at Hugging Face", "date": "", "ddg_snippet": "There are no explicit details about the resolution of the image . The image , sourced from GoogleEarth, showcases several airplanes and a vehicle situated on a tarmac area. There are four airplanes visible, with varying placements such as to the left, top-right, and bottom section of the image.", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/datasets/xiang709/VRSBench", "content": "There are no explicit details about the resolution of the image . The image , sourced from GoogleEarth, showcases several airplanes and a vehicle situated on a tarmac area. There are four airplanes visible, with varying placements such as to the left, top-right, and bottom section of the image."}
+{"idx": 2, "title": "VRSBench: A Versatile Vision-Language Benchmark Dataset for ... VRSBench: VRSBench: A Versatile Vision-Language Benchmark Dataset for ... VRSBench: A Versatile Vision-Language - nips.cc VRSBench : A Versatile Vision-Language Benchmark Dataset for Rem… VRSBench : A Versatile Vision-Language Benchmark Dataset for Remote GitHub - lx709/ VRSBench xiang709/ VRSBench · Datasets at Hugging Face xiang709/ VRSBench · Datasets at Hugging Face VRSBench : A Versatile Vision-Language Benchmark Dataset for Rem… Introducing VRSBench: Advancing Remote Sensing Image Analysis", "date": "", "ddg_snippet": "Jun 18, 2024 · It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks. We further evaluated state-of-the-art models on this benchmark for three vision-language tasks: image captioning, visual grounding, and visual question answering. This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks. We introduce a new benchmark designed to advance the development of general-purpose, large-scale vision-language models for remote sensing images . We reload the models that are initially trained on large-scale image -text alignment datasets , and then finetune each method using the training set of our VRSBench dataset for 5 epochs. How many images are in vrsbench? VRSBench comprises 29,614 images , each enriched with human-verified detailed captions, complex object referring, and question-answer pairs, check Table 1 for a detailed comparison with existing datasets. This dataset facilitates the training and evaluation of vision-language models across a spectrum of remote sensing image understanding tasks. What is a vrsbench benchmark for remote sensing image understanding? Exploring these improvement opportunities, we present a Versatile vision-language Benchmark for Remote Sensing image understanding, termed VRSBench. This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. How to download vrsbench dataset? VRSBench contains 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. check VRSBench Project Page. The dataset can be downloaded from link and used via the Hugging Face datasets library. To load the dataset , you can use the following code snippet: What data types can be used in vrsbench? Currently, the VRSBench dataset is limited to annotations for RGB images . In future work, we aim to enhance VRSBench by incorporating annotations from a variety of remote sensing data types, including infrared images, multi- and hyperspectral images, Synthetic Aperture Radar (SAR) images, and temporal datasets. What is vrsbench? VRSBench provides a comprehensive benchmark for developing and evaluating generalist vision-language models in both remote sensing and computer vision. This dataset not only supports the training and evaluation of advanced vision-language models but also boosts their ability to tackle complex real-world scenarios in remote sensing. What is the difference between vrsbench-ref and VQA? VRSBench-Ref: The task involves identifying and localizing specific objects from a given remote sensing image based on textual descriptions. VRSBench-VQA : This task aims to answer questions related to visual content in a given remote sensing image. Jul 27, 2025 · To improve how machines understand such images , we introduce a new dataset called VRSBench . This dataset combines images with detailed descriptions, object references, and questions with answers, which help machines learn better.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2406.12384", "content": "Jun 18, 2024 · It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks. We further evaluated state-of-the-art models on this benchmark for three vision-language tasks: image captioning, visual grounding, and visual question answering. This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks. We introduce a new benchmark designed to advance the development of general-purpose, large-scale vision-language models for remote sensing images . We reload the models that are initially trained on large-scale image -text alignment datasets , and then finetune each method using the training set of our VRSBench dataset for 5 epochs. How many images are in vrsbench? VRSBench comprises 29,614 images , each enriched with human-verified detailed captions, complex object referring, and question-answer pairs, check Table 1 for a detailed comparison with existing datasets. This dataset facilitates the training and evaluation of vision-language models across a spectrum of remote sensing image understanding tasks. What is a vrsbench benchmark for remote sensing image understanding? Exploring these improvement opportunities, we present a Versatile vision-language Benchmark for Remote Sensing image understanding, termed VRSBench. This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. How to download vrsbench dataset? VRSBench contains 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. check VRSBench Project Page. The dataset can be downloaded from link and used via the Hugging Face datasets library. To load the dataset , you can use the following code snippet: What data types can be used in vrsbench? Currently, the VRSBench dataset is limited to annotations for RGB images . In future work, we aim to enhance VRSBench by incorporating annotations from a variety of remote sensing data types, including infrared images, multi- and hyperspectral images, Synthetic Aperture Radar (SAR) images, and temporal datasets. What is vrsbench? VRSBench provides a comprehensive benchmark for developing and evaluating generalist vision-language models in both remote sensing and computer vision. This dataset not only supports the training and evaluation of advanced vision-language models but also boosts their ability to tackle complex real-world scenarios in remote sensing. What is the difference between vrsbench-ref and VQA? VRSBench-Ref: The task involves identifying and localizing specific objects from a given remote sensing image based on textual descriptions. VRSBench-VQA : This task aims to answer questions related to visual content in a given remote sensing image. Jul 27, 2025 · To improve how machines understand such images , we introduce a new dataset called VRSBench . This dataset combines images with detailed descriptions, object references, and questions with answers, which help machines learn better."}
+{"idx": 3, "title": "VRSBench:", "date": "", "ddg_snippet": "This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks.", "subpage_snippet": "", "source": "vrsbench.github.io", "link": "https://vrsbench.github.io/", "content": "This benchmark comprises 29,614 images, with 29,614 human-verified detailed captions, 52,472 object references, and 123,221 question-answer pairs. It facilitates the training and evaluation of vision-language models across a broad spectrum of remote sensing image understanding tasks."}
+{"idx": 4, "title": "VRSBench: A Versatile Vision-Language Benchmark Dataset for ...", "date": "", "ddg_snippet": "We introduce a new benchmark designed to advance the development of general-purpose, large-scale vision-language models for remote sensing images .", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/05b7f821234f66b78f99e7803fffa78a-Paper-Datasets_and_Benchmarks_Track.pdf", "content": "We introduce a new benchmark designed to advance the development of general-purpose, large-scale vision-language models for remote sensing images ."}
+{"idx": 5, "title": "VRSBench: A Versatile Vision-Language - nips.cc", "date": "", "ddg_snippet": "We reload the models that are initially trained on large-scale image -text alignment datasets , and then finetune each method using the training set of our VRSBench dataset for 5 epochs.", "subpage_snippet": "", "source": "nips.cc", "link": "https://nips.cc/media/neurips-2024/Slides/97530.pdf", "content": "We reload the models that are initially trained on large-scale image -text alignment datasets , and then finetune each method using the training set of our VRSBench dataset for 5 epochs."}
+{"idx": 6, "title": "Introducing VRSBench: Advancing Remote Sensing Image Analysis", "date": "", "ddg_snippet": "Jul 27, 2025 · To improve how machines understand such images , we introduce a new dataset called VRSBench . This dataset combines images with detailed descriptions, object references, and questions with answers, which help machines learn better.", "subpage_snippet": "", "source": "scisimple.com", "link": "https://scisimple.com/en/articles/2025-07-27-introducing-vrsbench-advancing-remote-sensing-image-analysis--ak6l6or", "content": "Jul 27, 2025 · To improve how machines understand such images , we introduce a new dataset called VRSBench . This dataset combines images with detailed descriptions, object references, and questions with answers, which help machines learn better."}
+{"idx": 7, "title": "Landsat30-AU: A Vision-Language Dataset for Australian Landsat", "date": "", "ddg_snippet": "We address this gap with Landsat30-AU, a large-scale vision-language dataset built from 30-meter resolution imagery collected by four Landsat ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.03127v1", "content": "We address this gap with Landsat30-AU, a large-scale vision-language dataset built from 30-meter resolution imagery collected by four Landsat ..."}
+{"idx": 8, "title": "RSCC: A Large-Scale Remote Sensing Change Caption Dataset for", "date": "", "ddg_snippet": "Large-Scale Event-Driven Dataset : 62,351 pre-/post-disaster image pairs sourced from 31 global events, spanning earthquakes, floods, wildfires, and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.01907v3", "content": "Large-Scale Event-Driven Dataset : 62,351 pre-/post-disaster image pairs sourced from 31 global events, spanning earthquakes, floods, wildfires, and ..."}
+{"idx": 9, "title": "AeroLite: Tag-Guided Lightweight Generation of Aerial Image", "date": "", "ddg_snippet": "These methods typically train on datasets such as MS COCO, Flickr30k (for natural images ), and smaller remote-sensing sets like Sydney/UCM Captions ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.09528v1", "content": "These methods typically train on datasets such as MS COCO, Flickr30k (for natural images ), and smaller remote-sensing sets like Sydney/UCM Captions ..."}
diff --git a/data/sampled_jsons/VRSBench_image_resolution_OR_image_size.jsonl b/data/sampled_jsons/VRSBench_image_resolution_OR_image_size.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..84d6234d956ac912c732a3b54efa4e33e8cda632
--- /dev/null
+++ b/data/sampled_jsons/VRSBench_image_resolution_OR_image_size.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "AeroLite: Tag-Guided Lightweight Generation of Aerial Image", "date": "", "ddg_snippet": "2015a ) ) established a basic framework for image captioning, they were not well-suited to high- resolution aerial imagery (Lu et al .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2504.09528v1", "content": "2015a ) ) established a basic framework for image captioning, they were not well-suited to high- resolution aerial imagery (Lu et al ."}
+{"idx": 1, "title": "Is there an actual minimum input image size for popular ...", "date": "", "ddg_snippet": "Oct 6, 2021 · There is a limitation on your input size which corresponds to the receptive field of the last convolution layer of your network. Intuitively, you can observe the spatial dimensionality decreasing as you progress through the network. At least this is the case for feature extractor CNNs which aim at extracting feature embeddings from the input image . That is most pre-trained models such as ...", "subpage_snippet": "", "source": "stackoverflow.com", "link": "https://stackoverflow.com/questions/69471729/is-there-an-actual-minimum-input-image-size-for-popular-computer-vision-models", "content": "Oct 6, 2021 · There is a limitation on your input size which corresponds to the receptive field of the last convolution layer of your network. Intuitively, you can observe the spatial dimensionality decreasing as you progress through the network. At least this is the case for feature extractor CNNs which aim at extracting feature embeddings from the input image . That is most pre-trained models such as ..."}
+{"idx": 2, "title": "Learning To Resize Images for Computer Vision Tasks", "date": "", "ddg_snippet": "Abstract For all the ways convolutional neural nets have revo-lutionized computer vision in recent years, one important aspect has received surprisingly little attention: the effect of image size on the accuracy of tasks being trained for. Typically, to be efficient, the input images are resized to a relatively small spatial resolution (e.g. 224 224), and both training and inference are ...", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/ICCV2021/papers/Talebi_Learning_To_Resize_Images_for_Computer_Vision_Tasks_ICCV_2021_paper.pdf", "content": "Abstract For all the ways convolutional neural nets have revo-lutionized computer vision in recent years, one important aspect has received surprisingly little attention: the effect of image size on the accuracy of tasks being trained for. Typically, to be efficient, the input images are resized to a relatively small spatial resolution (e.g. 224 224), and both training and inference are ..."}
+{"idx": 3, "title": "Non-Square images (Aspect ratio 16:9) for Training ... - GitHub", "date": "", "ddg_snippet": "Jul 18, 2023 · Also, the same resolution will be used for inference. I cannot possibly convert them to a square aspect ratio without distorting features in the image . I found this issue with training on non-square image size in which it was mentioned to use give the image dimension as imgsz and rect=True. When I am using this command with v8", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/ultralytics/ultralytics/issues/3815", "content": "Jul 18, 2023 · Also, the same resolution will be used for inference. I cannot possibly convert them to a square aspect ratio without distorting features in the image . I found this issue with training on non-square image size in which it was mentioned to use give the image dimension as imgsz and rect=True. When I am using this command with v8"}
+{"idx": 4, "title": "How to Pick the Optimal Image Size for Training Convolution ...", "date": "", "ddg_snippet": "Jun 23, 2021 · From the first plot, it looks like most images are of resolution less than 500 by 500. After zooming in, we can clearly see that images are clustered around either size 300 or 500. My recommendation for this dataset is to start training the neural network with image size 300 and progressively increase it to 400 and finish it with size 500. By this way, the model should be able to generalize ...", "subpage_snippet": "", "source": "www.raravind.com", "link": "https://www.raravind.com/blog/data-science/how-to-pick-the-optimal-image-size-for-training-convolution-neural-network", "content": "Jun 23, 2021 · From the first plot, it looks like most images are of resolution less than 500 by 500. After zooming in, we can clearly see that images are clustered around either size 300 or 500. My recommendation for this dataset is to start training the neural network with image size 300 and progressively increase it to 400 and finish it with size 500. By this way, the model should be able to generalize ..."}
+{"idx": 5, "title": "Image Sizing: Intrinsic vs. Rendered Size | DebugBear", "date": "", "ddg_snippet": "Apr 4, 2025 · Intrinsic and rendered sizes are two metrics of image sizing. Mismatches between them can harm the web performance and aesthetics of your site.", "subpage_snippet": "", "source": "www.debugbear.com", "link": "https://www.debugbear.com/docs/intrinsic-vs-rendered-size", "content": "Apr 4, 2025 · Intrinsic and rendered sizes are two metrics of image sizing. Mismatches between them can harm the web performance and aesthetics of your site."}
+{"idx": 6, "title": "Resize — Torchvision main documentation", "date": "", "ddg_snippet": "Resize class torchvision.transforms.Resize( size , interpolation=InterpolationMode.BILINEAR, max_size=None, antialias=True) [source] Resize the input image to the given size . If the image is torch Tensor, it is expected to have […, H, W] shape, where … means a maximum of two leading dimensions Parameters: size (sequence or int) – Desired output size . If size is a sequence like (h, w ...", "subpage_snippet": "", "source": "docs.pytorch.org", "link": "https://docs.pytorch.org/vision/main/generated/torchvision.transforms.Resize.html", "content": "Resize class torchvision.transforms.Resize( size , interpolation=InterpolationMode.BILINEAR, max_size=None, antialias=True) [source] Resize the input image to the given size . If the image is torch Tensor, it is expected to have […, H, W] shape, where … means a maximum of two leading dimensions Parameters: size (sequence or int) – Desired output size . If size is a sequence like (h, w ..."}
+{"idx": 7, "title": "Landsat30-AU: A Vision-Language Dataset for Australian Landsat", "date": "", "ddg_snippet": "... imagery , which encourages captions centered on fine-grained objects, such as cars, rooftops, or road markings, that are invisible at 30 m resolution ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.03127v1", "content": "... imagery , which encourages captions centered on fine-grained objects, such as cars, rooftops, or road markings, that are invisible at 30 m resolution ..."}
+{"idx": 8, "title": "RSCC: A Large-Scale Remote Sensing Change Caption Dataset for", "date": "", "ddg_snippet": "The images are cropped without overlapping to 512×512 from xBD’s original 1024×1024, while EBD retains its 512×512 resolution .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.01907v3", "content": "The images are cropped without overlapping to 512×512 from xBD’s original 1024×1024, while EBD retains its 512×512 resolution ."}
+{"idx": 9, "title": "Transformer-based Spatial Grounding: A Comprehensive Survey", "date": "", "ddg_snippet": "This process involves understanding both the semantic meaning of the language and the spatial or visual structure of the image .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.12739v1", "content": "This process involves understanding both the semantic meaning of the language and the spatial or visual structure of the image ."}
diff --git a/data/sampled_jsons/Video-ColBERT_Section_5.2_MSVD_frames_sampled_preprocessing.jsonl b/data/sampled_jsons/Video-ColBERT_Section_5.2_MSVD_frames_sampled_preprocessing.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5837bf533b6386a451cd2a3dbc6186ba4b882e4e
--- /dev/null
+++ b/data/sampled_jsons/Video-ColBERT_Section_5.2_MSVD_frames_sampled_preprocessing.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Nvidia Control Panel Multi Frame Sampled AA On or Off - YouTube", "date": "", "ddg_snippet": "This Nvidia Control Panel Multi Frame Sampled AA On or Off | MFAA VS MSAA Benchmark Comparison video covers the best settings to improve image quality and pe...", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=5ivk2N8ewo8", "content": "This Nvidia Control Panel Multi Frame Sampled AA On or Off | MFAA VS MSAA Benchmark Comparison video covers the best settings to improve image quality and pe..."}
+{"idx": 1, "title": "MSVD Dataset Corpus", "date": "", "ddg_snippet": "The Microsoft Research Video Description Corpus ( MSVD ) dataset consists of about 120K sentences collected during the summer of 2010. Workers on Mechanical Turk were paid to watch a short video snippet and then summarize the action in a single sentence.", "subpage_snippet": "", "source": "www.kaggle.com", "link": "https://www.kaggle.com/datasets/vtrnanh/msvd-dataset-corpus", "content": "The Microsoft Research Video Description Corpus ( MSVD ) dataset consists of about 120K sentences collected during the summer of 2010. Workers on Mechanical Turk were paid to watch a short video snippet and then summarize the action in a single sentence."}
+{"idx": 2, "title": "Self-Adaptive Sampling for Efficient Video Question Answering on", "date": "", "ddg_snippet": "Figure 3: Randomly sampled video frames from the msrvtt-qa dataset and two questions. The bracketed timestamps indicate cues for corresponding answers from the video . The QA pair in the red box cannot be grounded from the four sampled frames .", "subpage_snippet": "", "source": "www.cs.jhu.edu", "link": "https://www.cs.jhu.edu/~kevinduh/t/naacl24/final_pdf/paper614.pdf", "content": "Figure 3: Randomly sampled video frames from the msrvtt-qa dataset and two questions. The bracketed timestamps indicate cues for corresponding answers from the video . The QA pair in the red box cannot be grounded from the four sampled frames ."}
+{"idx": 3, "title": "Kvazaar 2.3.2 Download Free - VideoHelp", "date": "", "ddg_snippet": "Video Usability Information: --sar : Specify sample aspect ratio --overscan : Specify crop overscan setting [undef] - undef, show, crop --videoformat : Specify video format [undef] - undef, component, pal, ntsc, secam, mac --range : Specify color range [tv] - tv, pc --colorprim...", "subpage_snippet": "", "source": "www.videohelp.com", "link": "https://www.videohelp.com/software/Kvazaar", "content": "Video Usability Information: --sar : Specify sample aspect ratio --overscan : Specify crop overscan setting [undef] - undef, show, crop --videoformat : Specify video format [undef] - undef, component, pal, ntsc, secam, mac --range : Specify color range [tv] - tv, pc --colorprim..."}
+{"idx": 4, "title": "6 CSS Video Popup Modal - ForFrontend", "date": "", "ddg_snippet": "This project is a great example of combining design and functionality for an interactive video viewing experience. JavaScript Explanation: Hero Section Video Popup (Modal).", "subpage_snippet": "", "source": "forfrontend.com", "link": "https://forfrontend.com/6-css-video-popup-modal/", "content": "This project is a great example of combining design and functionality for an interactive video viewing experience. JavaScript Explanation: Hero Section Video Popup (Modal)."}
+{"idx": 5, "title": "Account 7qEYDbf6cuJ8HmeVbdw9s3DKCvzQvuhv4 mSvD 2iB6AEX", "date": "", "ddg_snippet": "The Address 7qEYDbf6cuJ8HmeVbdw9s3DKCvzQvuhv4 mSvD 2iB6AEX has balance of 6.55 SOL. Account page allows users to view transactions, token holdings, and more on the Solana blockchain.", "subpage_snippet": "", "source": "solscan.io", "link": "https://solscan.io/account/7qEYDbf6cuJ8HmeVbdw9s3DKCvzQvuhv4mSvD2iB6AEX", "content": "The Address 7qEYDbf6cuJ8HmeVbdw9s3DKCvzQvuhv4 mSvD 2iB6AEX has balance of 6.55 SOL. Account page allows users to view transactions, token holdings, and more on the Solana blockchain."}
+{"idx": 6, "title": "Отличия LabelEncoder и OneHotEncoder в SciKit Learn / Хабр", "date": "", "ddg_snippet": "from sklearn. preprocessing import OneHotEncoder onehotencoder = OneHotEncoder(categorical_features = [0]) x = onehotencoder.fit_transform(x).toarray(). В конструкторе мы указываем, какой столбец должен быть обработан OneHotEncoder...", "subpage_snippet": "", "source": "habr.com", "link": "https://habr.com/ru/articles/456294/", "content": "from sklearn. preprocessing import OneHotEncoder onehotencoder = OneHotEncoder(categorical_features = [0]) x = onehotencoder.fit_transform(x).toarray(). В конструкторе мы указываем, какой столбец должен быть обработан OneHotEncoder..."}
+{"idx": 7, "title": "Face Swap Online Free", "date": "", "ddg_snippet": "You can use FaceVary to swap faces in photos. With FaceVary, you can swap your face with your friends, celebrities, or even historical figures. You can create funny videos , memes, and collages, or just have some fun experimenting with your appearance.", "subpage_snippet": "", "source": "remaker.ai", "link": "https://remaker.ai/face-swap-free/", "content": "You can use FaceVary to swap faces in photos. With FaceVary, you can swap your face with your friends, celebrities, or even historical figures. You can create funny videos , memes, and collages, or just have some fun experimenting with your appearance."}
+{"idx": 8, "title": "Как скачать видео с Рутуба: 5 простых способов [2025]", "date": "", "ddg_snippet": "Easy Video Downloader. Утилита от Google Chrome, позволяющее сохранять мультимедиа напрямую с сайта. После установки оно интегрируется в интерфейс веб-браузера, добавляя кнопку скачивания.", "subpage_snippet": "", "source": "video-editor.su", "link": "https://video-editor.su/kak-skachat-video-s-rutube.php", "content": "Easy Video Downloader. Утилита от Google Chrome, позволяющее сохранять мультимедиа напрямую с сайта. После установки оно интегрируется в интерфейс веб-браузера, добавляя кнопку скачивания."}
+{"idx": 9, "title": "WAN 2.2 VACE: Object Swap with an Image in ComfyUI (GGUF Q5)", "date": "", "ddg_snippet": "Face stays the same across frames . In one try, the legs had mixed footwear. I added one small line in the prompt and ran again. Fixed. Workflow Groups. Video section . Upload your clip. In Resolution Master, pick a supported size.", "subpage_snippet": "", "source": "aistudynow.com", "link": "https://aistudynow.com/wan-2-2-vace-object-swap-with-an-image-in-comfyui-gguf-q5/", "content": "Face stays the same across frames . In one try, the legs had mixed footwear. I added one small line in the prompt and ran again. Fixed. Workflow Groups. Video section . Upload your clip. In Resolution Master, pick a supported size."}
diff --git a/data/sampled_jsons/Video-ColBERT_arXiv_equation_5_similarity_score_CVPR_year_2023.jsonl b/data/sampled_jsons/Video-ColBERT_arXiv_equation_5_similarity_score_CVPR_year_2023.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ffdb4675c74aab5e162f5b41682e461d0f1bb6f5
--- /dev/null
+++ b/data/sampled_jsons/Video-ColBERT_arXiv_equation_5_similarity_score_CVPR_year_2023.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "CVPR 2025 Accepted Papers", "date": "", "ddg_snippet": "CVPR 2025 Accepted Papers ; Dual Energy-Based Model with Open-World Uncertainty Estimation for Out-of-distribution Detection Poster Session 5. Qi Chen · Hu Ding.", "subpage_snippet": "", "source": "cvpr.thecvf.com", "link": "https://cvpr.thecvf.com/Conferences/2025/AcceptedPapers", "content": "CVPR 2025 Accepted Papers ; Dual Energy-Based Model with Open-World Uncertainty Estimation for Out-of-distribution Detection Poster Session 5. Qi Chen · Hu Ding."}
+{"idx": 1, "title": "jina-embeddings-v4: Universal Embeddings for Multimodal", "date": "", "ddg_snippet": "Objects with more such features in common will have corresponding vectors that are closer to each other by some metric (typically cosine similarity ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.18902v1", "content": "Objects with more such features in common will have corresponding vectors that are closer to each other by some metric (typically cosine similarity ..."}
+{"idx": 2, "title": "Recurrence-Enhanced Vision-and-Language Transformers for ...", "date": "", "ddg_snippet": "by D Caffagni · 2025 · Cited by 4 — Our model employs a Transformer-based recurrent cell to encode multiple vision-and-language layers into hidden vectors for similarity computation. 3. Proposed ...", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2025/papers/Caffagni_Recurrence-Enhanced_Vision-and-Language_Transformers_for_Robust_Multimodal_Document_Retrieval_CVPR_2025_paper.pdf", "content": "by D Caffagni · 2025 · Cited by 4 — Our model employs a Transformer-based recurrent cell to encode multiple vision-and-language layers into hidden vectors for similarity computation. 3. Proposed ..."}
+{"idx": 3, "title": "Text and Code Embeddings by Contrastive Pre-Training", "date": "", "ddg_snippet": "by A Neelakantan · Cited by 575 — The similarity score between x and y is defined as the cosine similarity between these two embedding vectors. The Transformer encoder maps the input, x and ... 13 pages", "subpage_snippet": "", "source": "cdn.openai.com", "link": "https://cdn.openai.com/papers/Text_and_Code_Embeddings_by_Contrastive_Pre_Training.pdf", "content": "by A Neelakantan · Cited by 575 — The similarity score between x and y is defined as the cosine similarity between these two embedding vectors. The Transformer encoder maps the input, x and ... 13 pages"}
+{"idx": 4, "title": "Generating Action-conditioned Prompts for Open- ...", "date": "", "ddg_snippet": "This is achieved by calculating the similarity between query video and textual embeddings of various categories, with the highest similarity score indicating ... 10 pages", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf/6549ef75c8da495ad410184b87a1f058ad7e02b7.pdf", "content": "This is achieved by calculating the similarity between query video and textual embeddings of various categories, with the highest similarity score indicating ... 10 pages"}
+{"idx": 5, "title": "Ask in Any Modality A Comprehensive Survey on ...", "date": "", "ddg_snippet": "by MM Abootorabi · 2025 · Cited by 22 — By facilitating the joint processing of text, images, audio, and video , multimodal learning is increasingly recognized as a critical enabler of. 34 pages", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/2025.findings-acl.861.pdf", "content": "by MM Abootorabi · 2025 · Cited by 22 — By facilitating the joint processing of text, images, audio, and video , multimodal learning is increasingly recognized as a critical enabler of. 34 pages"}
+{"idx": 6, "title": "What's the Best Way to Retrieve Slides? A Comparative ...", "date": "", "ddg_snippet": "2 days ago — Section 3 details the experimental setup, including datasets, retrieval methods, captioning models, rerankers, baselines, evaluation metrics, ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.15211v1", "content": "2 days ago — Section 3 details the experimental setup, including datasets, retrieval methods, captioning models, rerankers, baselines, evaluation metrics, ..."}
+{"idx": 7, "title": "PRADA: Practical Black-box Adversarial Attacks against ...", "date": "", "ddg_snippet": "by C Wu · 2023 · Cited by 77 — We propose a novel Pseudo Relevance-based ADversarial ranking Attack method (PRADA) that learns a surrogate model based on Pseudo Relevance Feedback (PRF) to ...", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/full/10.1145/3576923", "content": "by C Wu · 2023 · Cited by 77 — We propose a novel Pseudo Relevance-based ADversarial ranking Attack method (PRADA) that learns a surrogate model based on Pseudo Relevance Feedback (PRF) to ..."}
+{"idx": 8, "title": "Recurrence Meets Transformers for Universal Multimodal ...", "date": "", "ddg_snippet": "10 Sept 2025 — ReT [2] is a multimodal retrieval model that introduces a novel Transformer-based recurrent cell designed to fuse multimodal features from both ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.08897v1", "content": "10 Sept 2025 — ReT [2] is a multimodal retrieval model that introduces a novel Transformer-based recurrent cell designed to fuse multimodal features from both ..."}
+{"idx": 9, "title": "FORTIFY: Generative Model Fine-tuning with ORPO for ...", "date": "", "ddg_snippet": "by D DeGenaro · 2025 — In this work, we propose FORTIFY, a generative model fine- tuning approach for noisy document rewriting and summarization, to improve the ... 16 pages", "subpage_snippet": "", "source": "aclanthology.org", "link": "https://aclanthology.org/2025.magmar-1.13.pdf", "content": "by D DeGenaro · 2025 — In this work, we propose FORTIFY, a generative model fine- tuning approach for noisy document rewriting and summarization, to improve the ... 16 pages"}
diff --git a/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_1e-5_5e-5_learning_rate_synthetic_experiments_year_2024.jsonl b/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_1e-5_5e-5_learning_rate_synthetic_experiments_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..271fd5cd936e3c7ce4ee15f9c389ab952d6edc0e
--- /dev/null
+++ b/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_1e-5_5e-5_learning_rate_synthetic_experiments_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "AEGIS: Automated Error Generation and Identification for", "date": "", "ddg_snippet": "... of our dataset by exploring three distinct learning paradigms for the error identification task: Supervised Fine - Tuning , Reinforcement Learning , and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.14295v1", "content": "... of our dataset by exploring three distinct learning paradigms for the error identification task: Supervised Fine - Tuning , Reinforcement Learning , and ..."}
+{"idx": 1, "title": "Introduction to Fine-tuning Large Language Models", "date": "", "ddg_snippet": "I've been wanting to write an introduction on how to fine - tune large language models for a while now because it's such a vast and under-documented ...", "subpage_snippet": "", "source": "www.stephendiehl.com", "link": "https://www.stephendiehl.com/posts/training_llms/", "content": "I've been wanting to write an introduction on how to fine - tune large language models for a while now because it's such a vast and under-documented ..."}
+{"idx": 2, "title": "The Elicitation Game: Evaluating Capability Elicitation", "date": "", "ddg_snippet": "Previous work by (Greenblatt et al., 2024 ) has found that supervised fine - tuning (SFT) and reinforcement learning (RL) are effective at eliciting ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.02180v2", "content": "Previous work by (Greenblatt et al., 2024 ) has found that supervised fine - tuning (SFT) and reinforcement learning (RL) are effective at eliciting ..."}
+{"idx": 3, "title": "Zhongjing: Enhancing the Chinese Medical Capabilities of Large", "date": "", "ddg_snippet": "Existing efforts to incorporate Chinese medicine into LLMs rely on Supervised Fine - Tuning (SFT) with single-turn and distilled dialogue data.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2308.03549v3", "content": "Existing efforts to incorporate Chinese medicine into LLMs rely on Supervised Fine - Tuning (SFT) with single-turn and distilled dialogue data."}
+{"idx": 4, "title": "User-Assistant Bias in LLMs", "date": "", "ddg_snippet": "We further identify sources of user-assistant bias by fine - tuning with different post-training recipes and measuring bias shifts.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.15815v1", "content": "We further identify sources of user-assistant bias by fine - tuning with different post-training recipes and measuring bias shifts."}
+{"idx": 5, "title": "Time-MQA: Time Series Multi-Task Question Answering with", "date": "", "ddg_snippet": "Classical approaches rely on statistical models like z-score analysis and dynamic thresholding, while contemporary methods incorporate deep learning ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.01875v2", "content": "Classical approaches rely on statistical models like z-score analysis and dynamic thresholding, while contemporary methods incorporate deep learning ..."}
+{"idx": 6, "title": "Parameter Reference | PX4 User Guide (v1.12)", "date": "", "ddg_snippet": "For example if ASPD_FS_INNOV is 1 and estimator_status.tas_test_ratio is 2.0, then the exceedance is 1 .0 and the integral will rise at a rate of 1 .0 ...", "subpage_snippet": "", "source": "docs.px4.io", "link": "https://docs.px4.io/v1.12/en/advanced_config/parameter_reference.html", "content": "For example if ASPD_FS_INNOV is 1 and estimator_status.tas_test_ratio is 2.0, then the exceedance is 1 .0 and the integral will rise at a rate of 1 .0 ..."}
+{"idx": 7, "title": "$250 prize for checking Jake Cannell's Brain Efficiency", "date": "", "ddg_snippet": "It would make this whole discussion a lot less frustrating for me ( and probably many others following it) if you would spell out what claims you ...", "subpage_snippet": "", "source": "www.lesswrong.com", "link": "https://www.lesswrong.com/posts/Xo4cqLrAKjdNzpRSa/usd250-prize-for-checking-jake-cannell-s-brain-efficiency-1", "content": "It would make this whole discussion a lot less frustrating for me ( and probably many others following it) if you would spell out what claims you ..."}
+{"idx": 8, "title": "MedGemma Technical Report", "date": "", "ddg_snippet": "Fine - tuning MedGemma further improves performance in subdomains, reducing errors in electronic health record information retrieval by 50% and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.05201v2", "content": "Fine - tuning MedGemma further improves performance in subdomains, reducing errors in electronic health record information retrieval by 50% and ..."}
+{"idx": 9, "title": "Scalpel vs. Hammer: GRPO Amplifies Existing Capabilities, SFT", "date": "", "ddg_snippet": "... has become very popular, and is performed with reinforcement learning (RL) algorithms like GRPO, or with supervised fine - tuning (SFT) on synthetic ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.10616v1", "content": "... has become very popular, and is performed with reinforcement learning (RL) algorithms like GRPO, or with supervised fine - tuning (SFT) on synthetic ..."}
diff --git "a/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_A_Mechanistic_Study_\316\267M_\316\267S_learning_rates_Section_4.jsonl" "b/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_A_Mechanistic_Study_\316\267M_\316\267S_learning_rates_Section_4.jsonl"
new file mode 100644
index 0000000000000000000000000000000000000000..07b4bb93e072516dd93a138996916008f5abbde5
--- /dev/null
+++ "b/data/sampled_jsons/What_Makes_and_Breaks_Safety_Fine-tuning_A_Mechanistic_Study_\316\267M_\316\267S_learning_rates_Section_4.jsonl"
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Can Persuasion Break AI Safety? Exploring the Interplay Between", "date": "", "ddg_snippet": "... methods, experiments, and conclusions presented here are the result of my personal inquiry and are shared in the interest of advancing understanding ...", "subpage_snippet": "", "source": "www.greaterwrong.com", "link": "https://www.greaterwrong.com/posts/NNanHJrMDmHk7sYM7/can-persuasion-break-ai-safety-exploring-the-interplay", "content": "... methods, experiments, and conclusions presented here are the result of my personal inquiry and are shared in the interest of advancing understanding ..."}
+{"idx": 1, "title": "Can Persuasion Break AI Safety? Exploring the Interplay Between", "date": "", "ddg_snippet": "... methods, experiments, and conclusions presented here are the result of my personal inquiry and are shared in the interest of advancing understanding ...", "subpage_snippet": "", "source": "www.lesswrong.com", "link": "https://www.lesswrong.com/posts/NNanHJrMDmHk7sYM7/can-persuasion-break-ai-safety-exploring-the-interplay", "content": "... methods, experiments, and conclusions presented here are the result of my personal inquiry and are shared in the interest of advancing understanding ..."}
+{"idx": 2, "title": "How To Become A Mechanistic Interpretability Researcher —", "date": "", "ddg_snippet": "And in the end, of course, I get much less done ( and learn much less) than I would have if I had just tried something out.", "subpage_snippet": "", "source": "www.lesswrong.com", "link": "https://www.lesswrong.com/posts/jP9KDyMkchuv6tHwm/how-to-become-a-mechanistic-interpretability-researcher", "content": "And in the end, of course, I get much less done ( and learn much less) than I would have if I had just tried something out."}
+{"idx": 3, "title": "Why LLM Safety Guardrails Collapse After Fine-tuning: A", "date": "", "ddg_snippet": "However, once these safety -aligned models undergo further fine - tuning by third parties, their embedded safety guardrails can become compromised.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.05346v1", "content": "However, once these safety -aligned models undergo further fine - tuning by third parties, their embedded safety guardrails can become compromised."}
+{"idx": 4, "title": "Explaining AI through mechanistic interpretability | European", "date": "", "ddg_snippet": "Since the functional organisation of ANNs is both complex and machine-learned, how their internal structure implements the mapping from inputs to ...", "subpage_snippet": "", "source": "link.springer.com", "link": "https://link.springer.com/article/10.1007/s13194-024-00614-4", "content": "Since the functional organisation of ANNs is both complex and machine-learned, how their internal structure implements the mapping from inputs to ..."}
+{"idx": 5, "title": "How To Become A Mechanistic Interpretability Researcher — EA", "date": "", "ddg_snippet": "... applying! The application should ... Mech interp is high-leverage, impactful, and learnable on your own with short feedback loops and modest compute.", "subpage_snippet": "", "source": "forum.effectivealtruism.org", "link": "https://forum.effectivealtruism.org/posts/7mDeuQzJ56wozDAah/how-to-become-a-mechanistic-interpretability-researcher", "content": "... applying! The application should ... Mech interp is high-leverage, impactful, and learnable on your own with short feedback loops and modest compute."}
+{"idx": 6, "title": "How To Become A Mechanistic Interpretability Researcher — AI", "date": "", "ddg_snippet": "... applying! The application should ... Mech interp is high-leverage, impactful, and learnable on your own with short feedback loops and modest compute.", "subpage_snippet": "", "source": "www.alignmentforum.org", "link": "https://www.alignmentforum.org/posts/jP9KDyMkchuv6tHwm/how-to-become-a-mechanistic-interpretability-researcher", "content": "... applying! The application should ... Mech interp is high-leverage, impactful, and learnable on your own with short feedback loops and modest compute."}
+{"idx": 7, "title": "Mechanistically Eliciting Latent Behaviors in Language Models", "date": "", "ddg_snippet": "I apply the method to several alignment-relevant toy examples, and find that the method consistently learns vectors / adapters which encode coherent ...", "subpage_snippet": "", "source": "turntrout.com", "link": "https://turntrout.com/mechanistically-eliciting-latent-behaviors", "content": "I apply the method to several alignment-relevant toy examples, and find that the method consistently learns vectors / adapters which encode coherent ..."}
+{"idx": 8, "title": "‘AI safety’ directory · Gwern.net", "date": "", "ddg_snippet": "Anthropic Figured Out a Way to Look Inside: What Goes on in Artificial Neural Networks Work Is Largely a Mystery, Even to Their Creators.", "subpage_snippet": "", "source": "gwern.net", "link": "https://gwern.net/doc/reinforcement-learning/safe/index", "content": "Anthropic Figured Out a Way to Look Inside: What Goes on in Artificial Neural Networks Work Is Largely a Mystery, Even to Their Creators."}
+{"idx": 9, "title": "Noise Injection Systemically Degrades Large Language Model", "date": "", "ddg_snippet": "Most developers implement these guardrails through an extra safety fine - tuning stage that minimally edits weights to elicit refusals on curated ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.13500v1", "content": "Most developers implement these guardrails through an extra safety fine - tuning stage that minimally edits weights to elicit refusals on curated ..."}
diff --git a/data/sampled_jsons/When_to_Replan_An_Adaptive_Replanning_Strategy_arxiv_2304.12046.jsonl b/data/sampled_jsons/When_to_Replan_An_Adaptive_Replanning_Strategy_arxiv_2304.12046.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..987d03ad320f0ae04b55e8b11033cc0a83c83af0
--- /dev/null
+++ b/data/sampled_jsons/When_to_Replan_An_Adaptive_Replanning_Strategy_arxiv_2304.12046.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[ 2304 . 12046 ] When to Replan ? An Adaptive Replanning Strategy ...", "date": "", "ddg_snippet": "arXiv : 2304 . 12046 (cs). [Submitted on 24 Apr 2023 (v1), last revised 27 Feb 2024 (this version, v3)].To account for unforeseen or dynamic obstacles not present on the pre-built map, `` when to replan '' the reference path is critical for the success of safe and efficient navigation.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2304.12046", "content": "arXiv : 2304 . 12046 (cs). [Submitted on 24 Apr 2023 (v1), last revised 27 Feb 2024 (this version, v3)].To account for unforeseen or dynamic obstacles not present on the pre-built map, `` when to replan '' the reference path is critical for the success of safe and efficient navigation."}
+{"idx": 1, "title": "When to Replan? An Adaptive Replanning Strategy for Autonomous ...", "date": "", "ddg_snippet": "When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation OMRON SINIC X 578 subscribers Subscribe", "subpage_snippet": "", "source": "www.youtube.com", "link": "https://www.youtube.com/watch?v=W8nBFKDxsb0", "content": "When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation OMRON SINIC X 578 subscribers Subscribe"}
+{"idx": 2, "title": "When to Replan? An Adaptive Replanning Strategy for ... - arXiv.org", "date": "", "ddg_snippet": "Based on this insight, we then derive a new adaptive replanning strategy based on deep reinforcement learning, which can learn from experience to decide appropriate replanning timings in the given environment and planning setups.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2304.12046", "content": "Based on this insight, we then derive a new adaptive replanning strategy based on deep reinforcement learning, which can learn from experience to decide appropriate replanning timings in the given environment and planning setups."}
+{"idx": 3, "title": "when2replan/README.md at main · omron-sinicx/when2replan · GitHub", "date": "", "ddg_snippet": "Citation @misc {honda2024replan, title= { When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation using Deep Reinforcement Learning}, author= {Kohei Honda and Ryo Yonetani and Mai Nishimura and Tadashi Kozuno}, year= {2024}, eprint= { 2304.12046 }, archivePrefix= { arXiv }, primaryClass= {cs.RO} }", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/omron-sinicx/when2replan/blob/main/README.md", "content": "Citation @misc {honda2024replan, title= { When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation using Deep Reinforcement Learning}, author= {Kohei Honda and Ryo Yonetani and Mai Nishimura and Tadashi Kozuno}, year= {2024}, eprint= { 2304.12046 }, archivePrefix= { arXiv }, primaryClass= {cs.RO} }"}
+{"idx": 4, "title": "When to Replan? An Adaptive Replanning Strategy for Autonomous ...", "date": "", "ddg_snippet": "We then propose a new adaptive replanning strategy based on deep reinforcement learning, where an agent learns from experiences to decide appropriate replanning timings in the given environment and planning setups.", "subpage_snippet": "", "source": "www.x-mol.com", "link": "https://www.x-mol.com/paper/1650914745020854272?adv", "content": "We then propose a new adaptive replanning strategy based on deep reinforcement learning, where an agent learns from experiences to decide appropriate replanning timings in the given environment and planning setups."}
+{"idx": 5, "title": "什么时候重新计划？使用深度强化学习的自主导航自适应重新规划策略,arXiv - CS - Robotics - X-MOL", "date": "", "ddg_snippet": "什么时候重新计划？使用深度强化学习的自主导航自适应重新规划策略 arXiv - CS - Robotics Pub Date : 2023-04-24 , DOI: arxiv-2304.12046 Kohei Honda , Ryo Yonetani , Mai Nishimura , Tadashi Kozuno", "subpage_snippet": "", "source": "www.x-mol.com", "link": "https://www.x-mol.com/paper/1650914745020854272/t", "content": "什么时候重新计划？使用深度强化学习的自主导航自适应重新规划策略 arXiv - CS - Robotics Pub Date : 2023-04-24 , DOI: arxiv-2304.12046 Kohei Honda , Ryo Yonetani , Mai Nishimura , Tadashi Kozuno"}
+{"idx": 6, "title": "FAPP: Fast and Adaptive Perception and Planning for UAVs in Dynamic ...", "date": "", "ddg_snippet": "V Fast and Adaptive planning In this section, we will present our fast and adaptive planning which contains a trajectory optimization and an adaptive re-planning strategy . Our trajectory optimization does not require a front-end path search and takes into account the uncertainty of dynamic objects' state estimation.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2312.08743v1", "content": "V Fast and Adaptive planning In this section, we will present our fast and adaptive planning which contains a trajectory optimization and an adaptive re-planning strategy . Our trajectory optimization does not require a front-end path search and takes into account the uncertainty of dynamic objects' state estimation."}
+{"idx": 7, "title": "When to Replan ? An Adaptive Replanning Strategy for Autonomous...", "date": "", "ddg_snippet": "License: arXiv .org perpetual non-exclusive license. arXiv : 2304 . 12046 v3 [cs.RO] 27 Feb 2024. When to Replan ?on the existing hierarchical planning frameworks by learning the environment-specific adaptive replanning strategy .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2304.12046v3", "content": "License: arXiv .org perpetual non-exclusive license. arXiv : 2304 . 12046 v3 [cs.RO] 27 Feb 2024. When to Replan ?on the existing hierarchical planning frameworks by learning the environment-specific adaptive replanning strategy ."}
+{"idx": 8, "title": "When to Replan ? An Adaptive Replanning Strategy for Autonomous...", "date": "", "ddg_snippet": "(DOI: 10.48550/ arXiv . 2304 . 12046 ) The hierarchy of global and local planners is one of the most commonly utilized system designs in robot autonomous navigation. While the global planner generates a reference path from the current to goal locations based on the pre-built static map...", "subpage_snippet": "", "source": "scispace.com", "link": "https://scispace.com/papers/when-to-replan-an-adaptive-replanning-strategy-for-16jj6yyn", "content": "(DOI: 10.48550/ arXiv . 2304 . 12046 ) The hierarchy of global and local planners is one of the most commonly utilized system designs in robot autonomous navigation. While the global planner generates a reference path from the current to goal locations based on the pre-built static map..."}
+{"idx": 9, "title": "GitHub - omron-sinicx/ when 2 replan", "date": "", "ddg_snippet": "When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation using Deep Reinforcement Learning.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/omron-sinicx/when2replan", "content": "When to Replan ? An Adaptive Replanning Strategy for Autonomous Navigation using Deep Reinforcement Learning."}
diff --git a/data/sampled_jsons/Witness_complex_vs_Vietoris-Rips_advantage.jsonl b/data/sampled_jsons/Witness_complex_vs_Vietoris-Rips_advantage.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..d7b201329816bdfea023572bed9c4e47699541a8
--- /dev/null
+++ b/data/sampled_jsons/Witness_complex_vs_Vietoris-Rips_advantage.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Vietoris-Rips complex - Wikipedia", "date": "", "ddg_snippet": "A Vietoris - Rips complex of a set of 23 points in the Euclidean plane. This complex has sets of up to four points: the points themselves (shown as red circles), pairs of points (black edges), triples of points (pale blue triangles), and quadruples of points (dark blue tetrahedrons). In topology, the Vietoris - Rips complex , also called the Vietoris complex or Rips complex , is a way of forming ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Vietoris–Rips_complex", "content": "A Vietoris - Rips complex of a set of 23 points in the Euclidean plane. This complex has sets of up to four points: the points themselves (shown as red circles), pairs of points (black edges), triples of points (pale blue triangles), and quadruples of points (dark blue tetrahedrons). In topology, the Vietoris - Rips complex , also called the Vietoris complex or Rips complex , is a way of forming ..."}
+{"idx": 1, "title": "How to compare different type of simplicial complex?", "date": "", "ddg_snippet": "The Vietoris - Rips complex computations are much easier because the ambient dimension does not hinder computations, as the Vietoris - Rips complex depends only on pairwise distances.", "subpage_snippet": "", "source": "mathoverflow.net", "link": "https://mathoverflow.net/questions/331502/how-to-compare-different-type-of-simplicial-complex", "content": "The Vietoris - Rips complex computations are much easier because the ambient dimension does not hinder computations, as the Vietoris - Rips complex depends only on pairwise distances."}
+{"idx": 2, "title": "The Čech Complex and the Vietoris-Rips Complex - Math ∩ Programming", "date": "", "ddg_snippet": "One obvious question is: do we still get the benefits of the nerve theorem with Vietoris - Rips complexes? The answer is no, obviously, because the Vietoris - Rips complex and Čech complex in this triangle example have totally different topology! But everything's not lost.", "subpage_snippet": "", "source": "www.jeremykun.com", "link": "https://www.jeremykun.com/2015/08/06/cech-vietoris-rips-complex/", "content": "One obvious question is: do we still get the benefits of the nerve theorem with Vietoris - Rips complexes? The answer is no, obviously, because the Vietoris - Rips complex and Čech complex in this triangle example have totally different topology! But everything's not lost."}
+{"idx": 3, "title": "Advanced Cell Complex Techniques - numberanalytics.com", "date": "", "ddg_snippet": "Comparison of Construction Methods Different construction methods have varying properties, advantages , and disadvantages. The following table summarizes the key characteristics of alpha complexes, witness complexes, and Vietoris - Rips complexes:", "subpage_snippet": "", "source": "www.numberanalytics.com", "link": "https://www.numberanalytics.com/blog/advanced-cell-complex-techniques-topological-machine-learning", "content": "Comparison of Construction Methods Different construction methods have varying properties, advantages , and disadvantages. The following table summarizes the key characteristics of alpha complexes, witness complexes, and Vietoris - Rips complexes:"}
+{"idx": 4, "title": "Fast construction of the Vietoris-Rips complex - ScienceDirect", "date": "", "ddg_snippet": "In this paper, we present fast algorithms for constructing the filtered Vietoris - Rips complex of a point set. Our software can compute arbitrary dimensional complexes for point sets in arbitrary dimensions, and may be applied toward constructing other clique complexes, such as the weak witness complex . Fig. 1 shows the performance of our fastest algorithm to inspire interest. On this point set ...", "subpage_snippet": "", "source": "www.sciencedirect.com", "link": "https://www.sciencedirect.com/science/article/pii/S0097849310000464", "content": "In this paper, we present fast algorithms for constructing the filtered Vietoris - Rips complex of a point set. Our software can compute arbitrary dimensional complexes for point sets in arbitrary dimensions, and may be applied toward constructing other clique complexes, such as the weak witness complex . Fig. 1 shows the performance of our fastest algorithm to inspire interest. On this point set ..."}
+{"idx": 5, "title": "Vietoris-Rips Complexes: The Backbone of Persistent Homology", "date": "", "ddg_snippet": "Explore the intricacies of Vietoris - Rips complexes and their role in Persistent Homology, providing a deeper understanding of complex data structures.", "subpage_snippet": "", "source": "www.numberanalytics.com", "link": "https://www.numberanalytics.com/blog/vietoris-rips-complexes-persistent-homology-deep-dive", "content": "Explore the intricacies of Vietoris - Rips complexes and their role in Persistent Homology, providing a deeper understanding of complex data structures."}
+{"idx": 6, "title": "PDF Chapter5 SimplicialComplexesonPointClouds", "date": "", "ddg_snippet": "Vietoris - Rips complex easier to compute. Furthermore, we make the following simple observation, showing that the Vietoris - Rips complex still appro imately captures shapes in t", "subpage_snippet": "", "source": "ti.inf.ethz.ch", "link": "https://ti.inf.ethz.ch/ew/courses/TDA24/Chapter5.pdf", "content": "Vietoris - Rips complex easier to compute. Furthermore, we make the following simple observation, showing that the Vietoris - Rips complex still appro imately captures shapes in t"}
+{"idx": 7, "title": "PDF rips shadows final.dvi", "date": "", "ddg_snippet": "The Vietoris - Rips complex is a combinatorial simplicial complex based on proximity of neighbors that serves as an easily-computed but high-dimensional ap-proximation to the homotopy type of D.", "subpage_snippet": "", "source": "jeffe.cs.illinois.edu", "link": "http://jeffe.cs.illinois.edu/pubs/pdf/rips.pdf", "content": "The Vietoris - Rips complex is a combinatorial simplicial complex based on proximity of neighbors that serves as an easily-computed but high-dimensional ap-proximation to the homotopy type of D."}
+{"idx": 8, "title": "A New Construction of the Vietoris-Rips Complex - arXiv.org", "date": "", "ddg_snippet": "Abstract. We present a new, inductive construction of the Vietoris - Rips complex , in which we take advantage of a small amount of unexploited combinatorial structure in the k-skeleton of the complex in order to avoid unnecessary comparisons when identifying its (k + 1)-simplices. In doing so, we achieve a significant reduction in the number of comparisons required to construct the Vietoris - Rips ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2301.07191", "content": "Abstract. We present a new, inductive construction of the Vietoris - Rips complex , in which we take advantage of a small amount of unexploited combinatorial structure in the k-skeleton of the complex in order to avoid unnecessary comparisons when identifying its (k + 1)-simplices. In doing so, we achieve a significant reduction in the number of comparisons required to construct the Vietoris - Rips ..."}
+{"idx": 9, "title": "PDF Computational Topology for Data Analysis: Notes from Book by", "date": "", "ddg_snippet": "Sections 2.2 presents an important construction called the nerve and a complex called the Cech complex whichˇ is defined on this construction. This section also presents a commonly used complex in topolog- ical data analysis called the Vietoris - Rips complex that interleaves with the Cech complexes inˇ terms of containment.", "subpage_snippet": "", "source": "www.cs.purdue.edu", "link": "https://www.cs.purdue.edu/homes/tamaldey/course/531/Simplicial-complex2.pdf", "content": "Sections 2.2 presents an important construction called the nerve and a complex called the Cech complex whichˇ is defined on this construction. This section also presents a commonly used complex in topolog- ical data analysis called the Vietoris - Rips complex that interleaves with the Cech complexes inˇ terms of containment."}
diff --git a/data/sampled_jsons/X-CLIP_Ma_et_al._2022_abstract.jsonl b/data/sampled_jsons/X-CLIP_Ma_et_al._2022_abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..32121658e48e5cfa60091f4c2dcb7e61b94ac94a
--- /dev/null
+++ b/data/sampled_jsons/X-CLIP_Ma_et_al._2022_abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2207.07285] X-CLIP: End-to-End Multi-grained Contrastive ... X-CLIP - Project Page - GitHub Pages X-CLIP: End-to-End Multi-grained Contrastive Learning for ... What is the relation with your X-CLIP and X-CLIP by Yiwei Ma ... [2207.07285] X-CLIP: End-to-End Multi-grained Contrastive ... MedCLIP: Contrastive Learning from Unpaired Medical Images ... (ECCV 2022 Oral) X-CLIP for Video Recognition阅读笔记 - 知乎", "date": "", "ddg_snippet": "Jul 15, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity. Accepted by ACM MM 2022 (Main Track) TL;DR: X-CLIP is a video-text retrieval model with multi-grained contrastive learning based on CLIP . Oct 10, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity. May 13, 2023 · I noticed there is another model called X-CLIP by Yiwei Ma et . al , arXiv:2207.07285. Their paper was submitted on arXiv on July 2022 , while your paper (Bolin Ni et . al , 2208.02816) on August 2022 , one month later. At least, it seems that... CLIP (Radford et al ., 2021) implements the idea of contrastive learning based on a large number of image-text pairs, achieving outstanding performance on several multi-modal downstream tasks (Zhang et al ., 2021; Ji et al ., 2022 ; Ma et al ., 2022 ; Zhu et al ., 2022 ; Ji et al ., 2021; He et al ., 2022 ). Abstract Existing vision-text contrastive learning like CLIP (Radford et al ., 2021) aims to match the paired image and caption embeddings while pushing others apart, which improves representation transferability and supports zero-shot prediction. 论文笔记 会尽量简洁的介绍文章内容+一些个人观点。 Expanding Language-Image Pretrained Models for General Video Recognition (ECCV 2022 Oral)", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2207.07285", "content": "Jul 15, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity. Accepted by ACM MM 2022 (Main Track) TL;DR: X-CLIP is a video-text retrieval model with multi-grained contrastive learning based on CLIP . Oct 10, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity. May 13, 2023 · I noticed there is another model called X-CLIP by Yiwei Ma et . al , arXiv:2207.07285. Their paper was submitted on arXiv on July 2022 , while your paper (Bolin Ni et . al , 2208.02816) on August 2022 , one month later. At least, it seems that... CLIP (Radford et al ., 2021) implements the idea of contrastive learning based on a large number of image-text pairs, achieving outstanding performance on several multi-modal downstream tasks (Zhang et al ., 2021; Ji et al ., 2022 ; Ma et al ., 2022 ; Zhu et al ., 2022 ; Ji et al ., 2021; He et al ., 2022 ). Abstract Existing vision-text contrastive learning like CLIP (Radford et al ., 2021) aims to match the paired image and caption embeddings while pushing others apart, which improves representation transferability and supports zero-shot prediction. 论文笔记 会尽量简洁的介绍文章内容+一些个人观点。 Expanding Language-Image Pretrained Models for General Video Recognition (ECCV 2022 Oral)"}
+{"idx": 1, "title": "X-CLIP - Project Page - GitHub Pages", "date": "", "ddg_snippet": "Accepted by ACM MM 2022 (Main Track) TL;DR: X-CLIP is a video-text retrieval model with multi-grained contrastive learning based on CLIP .", "subpage_snippet": "", "source": "xmu-xiaoma666.github.io", "link": "https://xmu-xiaoma666.github.io/Projects/MM22_XCLIP/", "content": "Accepted by ACM MM 2022 (Main Track) TL;DR: X-CLIP is a video-text retrieval model with multi-grained contrastive learning based on CLIP ."}
+{"idx": 2, "title": "X-CLIP: End-to-End Multi-grained Contrastive Learning for ...", "date": "", "ddg_snippet": "Oct 10, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity.", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/abs/10.1145/3503161.3547910", "content": "Oct 10, 2022 · To this end, this paper presents a novel multi-grained contrastive model, namely X-CLIP , for video-text retrieval. However, another challenge lies in the similarity aggregation problem, which aims to aggregate fine-grained and cross-grained similarity matrices to instance-level similarity."}
+{"idx": 3, "title": "What is the relation with your X-CLIP and X-CLIP by Yiwei Ma ...", "date": "", "ddg_snippet": "May 13, 2023 · I noticed there is another model called X-CLIP by Yiwei Ma et . al , arXiv:2207.07285. Their paper was submitted on arXiv on July 2022 , while your paper (Bolin Ni et . al , 2208.02816) on August 2022 , one month later. At least, it seems that...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/microsoft/VideoX/issues/92", "content": "May 13, 2023 · I noticed there is another model called X-CLIP by Yiwei Ma et . al , arXiv:2207.07285. Their paper was submitted on arXiv on July 2022 , while your paper (Bolin Ni et . al , 2208.02816) on August 2022 , one month later. At least, it seems that..."}
+{"idx": 4, "title": "[2207.07285] X-CLIP: End-to-End Multi-grained Contrastive ...", "date": "", "ddg_snippet": "CLIP (Radford et al ., 2021) implements the idea of contrastive learning based on a large number of image-text pairs, achieving outstanding performance on several multi-modal downstream tasks (Zhang et al ., 2021; Ji et al ., 2022 ; Ma et al ., 2022 ; Zhu et al ., 2022 ; Ji et al ., 2021; He et al ., 2022 ).", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2207.07285", "content": "CLIP (Radford et al ., 2021) implements the idea of contrastive learning based on a large number of image-text pairs, achieving outstanding performance on several multi-modal downstream tasks (Zhang et al ., 2021; Ji et al ., 2022 ; Ma et al ., 2022 ; Zhu et al ., 2022 ; Ji et al ., 2021; He et al ., 2022 )."}
+{"idx": 5, "title": "MedCLIP: Contrastive Learning from Unpaired Medical Images ...", "date": "", "ddg_snippet": "Abstract Existing vision-text contrastive learning like CLIP (Radford et al ., 2021) aims to match the paired image and caption embeddings while pushing others apart, which improves representation transferability and supports zero-shot prediction.", "subpage_snippet": "", "source": "pubmed.ncbi.nlm.nih.gov", "link": "https://pubmed.ncbi.nlm.nih.gov/39144675/", "content": "Abstract Existing vision-text contrastive learning like CLIP (Radford et al ., 2021) aims to match the paired image and caption embeddings while pushing others apart, which improves representation transferability and supports zero-shot prediction."}
+{"idx": 6, "title": "(ECCV 2022 Oral) X-CLIP for Video Recognition阅读笔记 - 知乎", "date": "", "ddg_snippet": "论文笔记 会尽量简洁的介绍文章内容+一些个人观点。 Expanding Language-Image Pretrained Models for General Video Recognition (ECCV 2022 Oral)", "subpage_snippet": "", "source": "zhuanlan.zhihu.com", "link": "https://zhuanlan.zhihu.com/p/563557468", "content": "论文笔记 会尽量简洁的介绍文章内容+一些个人观点。 Expanding Language-Image Pretrained Models for General Video Recognition (ECCV 2022 Oral)"}
+{"idx": 7, "title": "X - CLIP : End-to-End Multi-grained Contrastive Learning for Video-Text...", "date": "", "ddg_snippet": "... University China ( Ma et al ., ( 2022 )) greatly informed how we think about temporally aligning the words someone speaks with the actions appearing on screen; in short, often YouTubers don't speak and act simultaneously. They typically talk first, then show. ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/364484274_X-CLIP_End-to-End_Multi-grained_Contrastive_Learning_for_Video-Text_Retrieval", "content": "... University China ( Ma et al ., ( 2022 )) greatly informed how we think about temporally aligning the words someone speaks with the actions appearing on screen; in short, often YouTubers don't speak and act simultaneously. They typically talk first, then show. ..."}
+{"idx": 8, "title": "Session I Abstracts", "date": "", "ddg_snippet": "Fan et al . 2022 introduced MineDojo: an environment built in Minecraft for training reinforcement learning agents on open-ended tasks.", "subpage_snippet": "", "source": "sfp.caltech.edu", "link": "https://sfp.caltech.edu/documents/25686/2023_Session_I_Abstracts.pdf", "content": "Fan et al . 2022 introduced MineDojo: an environment built in Minecraft for training reinforcement learning agents on open-ended tasks."}
+{"idx": 9, "title": "Towards Efficient and Effective Text-to-Video Retrieval with...", "date": "", "ddg_snippet": "CLIP4Clip† X - CLIP † DRL (Wang et al . 2022 ; Ma et al . 2022 ), the spatial encoder and text encoder of EERCF are initialized by the CLIP checkpoints. We train our model via Adam optimizer and decay the learning rate using a cosine schedule strategy.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2401.00701.pdf", "content": "CLIP4Clip† X - CLIP † DRL (Wang et al . 2022 ; Ma et al . 2022 ), the spatial encoder and text encoder of EERCF are initialized by the CLIP checkpoints. We train our model via Adam optimizer and decay the learning rate using a cosine schedule strategy."}
diff --git a/data/sampled_jsons/Zhao_et_al._2023b_bandit_algorithms_peeling.jsonl b/data/sampled_jsons/Zhao_et_al._2023b_bandit_algorithms_peeling.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..8d3e61f13ccc1fdfa6edc30739b18ddceda73733
--- /dev/null
+++ b/data/sampled_jsons/Zhao_et_al._2023b_bandit_algorithms_peeling.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Catoni Contextual Bandits are Robust to Heavy-tailed ...", "date": "", "ddg_snippet": "by C Ye · 2025 · Cited by 1 — Zhao et al . ( 2023b ) develop a peeling approach for the unknown variance case without variance es- timation in linear settings, and Pacchiano ( ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2502.02486?", "content": "by C Ye · 2025 · Cited by 1 — Zhao et al . ( 2023b ) develop a peeling approach for the unknown variance case without variance es- timation in linear settings, and Pacchiano ( ..."}
+{"idx": 1, "title": "Catoni Contextual Bandits are Robust to Heavy-tailed ...", "date": "", "ddg_snippet": "For the unknown-variance case, we further propose a careful peeling -based algorithm and remove the need for cumbersome variance estimation. With additional ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46438", "content": "For the unknown-variance case, we further propose a careful peeling -based algorithm and remove the need for cumbersome variance estimation. With additional ..."}
+{"idx": 2, "title": "Catoni Contextual Bandits are Robust to Heavy-tailed ...", "date": "", "ddg_snippet": "2 Aug 2025 — The linear analysis for peeling in Zhao . et . al is restricted to the linear vector space structure and special form of uncertainty and thus can not be extended ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=5IpVe9PH14&noteId=J3K6uYfoM5", "content": "2 Aug 2025 — The linear analysis for peeling in Zhao . et . al is restricted to the linear vector space structure and special form of uncertainty and thus can not be extended ..."}
+{"idx": 3, "title": "Unifying (Federated) (Private) High-Dimensional Bandits via ...", "date": "", "ddg_snippet": "... Peeling . 136 algorithms Dwork et al . (2014b) are typically used. The analysis of these algorithms in centralized. 137 and federated, private and non-private ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=icM2HffLwD", "content": "... Peeling . 136 algorithms Dwork et al . (2014b) are typically used. The analysis of these algorithms in centralized. 137 and federated, private and non-private ..."}
+{"idx": 4, "title": "Noise-Adaptive Confidence Sets for Linear Bandits and ...", "date": "", "ddg_snippet": "by KS Jun · 2024 · Cited by 4 — This proof is inspired by Zhao et al . ( 2023b ), but details differ since we leverage the regret equality (Lemma C.7), which helps shorten the proof and ...", "subpage_snippet": "", "source": "par.nsf.gov", "link": "https://par.nsf.gov/servlets/purl/10548723", "content": "by KS Jun · 2024 · Cited by 4 — This proof is inspired by Zhao et al . ( 2023b ), but details differ since we leverage the regret equality (Lemma C.7), which helps shorten the proof and ..."}
+{"idx": 5, "title": "Non-iterative border-peeling clustering algorithm based on swap ...", "date": "", "ddg_snippet": "Article on Non-iterative border- peeling clustering algorithm based on swap strategy, published in Information Sciences 654 on 2023 -11-04 by Hui Tu+5.", "subpage_snippet": "", "source": "discovery.researcher.life", "link": "https://discovery.researcher.life/article/non-iterative-border-peeling-clustering-algorithm-based-on-swap-strategy/fe5c738535113688b0d5ad6b122d5473", "content": "Article on Non-iterative border- peeling clustering algorithm based on swap strategy, published in Information Sciences 654 on 2023 -11-04 by Hui Tu+5."}
+{"idx": 6, "title": "Variance-Dependent Regret Lower Bounds for Contextual ...", "date": "", "ddg_snippet": "by J He · 2025 · Cited by 1 — In this paper, we study variance-dependent lower bounds for linear contextual bandits in different settings. For both prefixed and adaptive ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2503.12020", "content": "by J He · 2025 · Cited by 1 — In this paper, we study variance-dependent lower bounds for linear contextual bandits in different settings. For both prefixed and adaptive ..."}
+{"idx": 7, "title": "Noise-Adaptive Thompson Sampling for Linear Contextual ...", "date": "", "ddg_snippet": "by R Xu · Cited by 11 — (2022b); Zhao et al . ( 2023 ) explored the unknown variance case. Nevertheless, there is a scarcity of results concerning TS algorithms in this context.", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2023/file/4a6824f8f137e78f18e73d9cfc1d22ed-Supplemental-Conference.pdf", "content": "by R Xu · Cited by 11 — (2022b); Zhao et al . ( 2023 ) explored the unknown variance case. Nevertheless, there is a scarcity of results concerning TS algorithms in this context."}
+{"idx": 8, "title": "Revisiting Simple Regret: Fast Rates for Returning a Good Arm", "date": "", "ddg_snippet": "by Y Zhao · 2023 · Cited by 23 — Simple regret is an unverifiable performance measure, mean- ing that the algorithm need not verify its performance to a prescribed level. This is in stark ... 49 pages", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v202/zhao23g/zhao23g.pdf", "content": "by Y Zhao · 2023 · Cited by 23 — Simple regret is an unverifiable performance measure, mean- ing that the algorithm need not verify its performance to a prescribed level. This is in stark ... 49 pages"}
+{"idx": 9, "title": "Noise-Adaptive Thompson Sampling for Linear Contextual ...", "date": "", "ddg_snippet": "Linear contextual bandits represent a fundamental class of models with numerous real-world applications, and it is critical to developing algorithms that ...", "subpage_snippet": "", "source": "neurips.cc", "link": "https://neurips.cc/virtual/2023/poster/72465", "content": "Linear contextual bandits represent a fundamental class of models with numerous real-world applications, and it is critical to developing algorithms that ..."}
diff --git a/data/sampled_jsons/abstract_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera_Yu_et_al._2024.jsonl b/data/sampled_jsons/abstract_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera_Yu_et_al._2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..02b018ba871a59868222361e6dd04dfd3a2fd812
--- /dev/null
+++ b/data/sampled_jsons/abstract_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera_Yu_et_al._2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "EventPS : Real - Time Photometric Stereo Using an Event Camera", "date": "", "ddg_snippet": "Photometric stereo is a well-established technique to estimate the surface normal of an object.This paper introduces EventPS a novel approach to real - time photometric stereo using an event camera .", "subpage_snippet": "", "source": "paperswithcode.com", "link": "https://paperswithcode.com/paper/eventps-real-time-photometric-stereo-using-an", "content": "Photometric stereo is a well-established technique to estimate the surface normal of an object.This paper introduces EventPS a novel approach to real - time photometric stereo using an event camera ."}
+{"idx": 1, "title": "EventPS : Real - Time Photometric Stereo Using an Event Camera", "date": "", "ddg_snippet": "This paper introduces EventPS , a novel approach to real - time photometric stereo using an event camera . Capitalizing on the exceptional temporal resolution, dynamic range, and low bandwidth character-istics of event cameras , EventPS estimates surface...", "subpage_snippet": "", "source": "openaccess.thecvf.com", "link": "https://openaccess.thecvf.com/content/CVPR2024/papers/Yu_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera_CVPR_2024_paper.pdf", "content": "This paper introduces EventPS , a novel approach to real - time photometric stereo using an event camera . Capitalizing on the exceptional temporal resolution, dynamic range, and low bandwidth character-istics of event cameras , EventPS estimates surface..."}
+{"idx": 2, "title": "EventPS : Real - Time Photometric Stereo Using an Event Camera", "date": "", "ddg_snippet": "... While the event camera 's sensitivity to photometric changes could potentially complement traditional cameras in such tasks, it may also be adversely affected. To date, only a few studies have explored using event cameras for photometric stereo of these types of objects [107].", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/390160865_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera", "content": "... While the event camera 's sensitivity to photometric changes could potentially complement traditional cameras in such tasks, it may also be adversely affected. To date, only a few studies have explored using event cameras for photometric stereo of these types of objects [107]."}
+{"idx": 3, "title": "EventPS : Real - Time Photometric Stereo Using an Event Camera", "date": "", "ddg_snippet": "Yu _ EventPS _ Real - Time _ Photometric _ Stereo _ Using _ an _ Event _ Camera @CVPR 2024 @CVF.Authors: Bohan Yu , Jieji Ren, Jin Han, Feishi Wang, Jinxiu Liang, Boxin Shi. Photometric stereo is a well-established technique to estimate the surface normal of an object.", "subpage_snippet": "", "source": "papers.cool", "link": "https://papers.cool/venue/Yu_EventPS_Real-Time_Photometric_Stereo_Using_an_Event_Camera@CVPR2024@CVF", "content": "Yu _ EventPS _ Real - Time _ Photometric _ Stereo _ Using _ an _ Event _ Camera @CVPR 2024 @CVF.Authors: Bohan Yu , Jieji Ren, Jin Han, Feishi Wang, Jinxiu Liang, Boxin Shi. Photometric stereo is a well-established technique to estimate the surface normal of an object."}
+{"idx": 4, "title": "EventPS : Real - Time Photometric Stereo Using an Event Camera", "date": "", "ddg_snippet": "EVK4 HD camera . EventPS capitalizes on the exceptional temporal resolution, dynamic range, and low bandwidth characteristics of event cameras to enhance real - time 3D shape capturing.", "subpage_snippet": "", "source": "www.linkedin.com", "link": "https://www.linkedin.com/posts/chronocam_eventps-real-time-photometric-stereo-using-activity-7209602086332739585-CNFS", "content": "EVK4 HD camera . EventPS capitalizes on the exceptional temporal resolution, dynamic range, and low bandwidth characteristics of event cameras to enhance real - time 3D shape capturing."}
+{"idx": 5, "title": "Yu Bohan from Peking University: EventPS - Real - time Photometric ...", "date": "", "ddg_snippet": "3. Real - time algorithm: Demonstrate the real - time performance of the SVD-based solution algorithm on various objects; 4. Deep learning extension: Extend the existing deep learning photometric stereo vision framework to the EventPS version to adapt to the general reflectance model", "subpage_snippet": "", "source": "inf.news", "link": "https://inf.news/en/tech/4be4c17d4830518b9004d13cb772438e.html", "content": "3. Real - time algorithm: Demonstrate the real - time performance of the SVD-based solution algorithm on various objects; 4. Deep learning extension: Extend the existing deep learning photometric stereo vision framework to the EventPS version to adapt to the general reflectance model"}
+{"idx": 6, "title": "CVPR 2024 Key Research & Dataset Papers - Part 2", "date": "", "ddg_snippet": "EventPS : Real - Time Photometric Stereo Using an Event Camera . Comparing the Decision-Making Mechanisms by Transformers and CNNs via Explanation Methods. LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry.", "subpage_snippet": "", "source": "learnopencv.com", "link": "https://learnopencv.com/cvpr-2024-research-papers/", "content": "EventPS : Real - Time Photometric Stereo Using an Event Camera . Comparing the Decision-Making Mechanisms by Transformers and CNNs via Explanation Methods. LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry."}
+{"idx": 7, "title": "PS -EIP: Robust Photometric Stereo Based on Event Interval Profile", "date": "", "ddg_snippet": "This paper proposes Photometric Stereo based on Event Interval Profile ( PS -EIP), a robust method that recovers pixelwise surface normals from a time -series profile of event intervals.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2503.18341", "content": "This paper proposes Photometric Stereo based on Event Interval Profile ( PS -EIP), a robust method that recovers pixelwise surface normals from a time -series profile of event intervals."}
+{"idx": 8, "title": "CVPR 2024 Announces Best Paper Award Winners", "date": "", "ddg_snippet": "Honorable mention papers included, “ EventPS : Real - Time Photometric Stereo Using an Event Camera ” and “ pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction. ”", "subpage_snippet": "", "source": "www.computer.org", "link": "https://www.computer.org/press-room/cvpr-2024-announces-best-paper-award-winners", "content": "Honorable mention papers included, “ EventPS : Real - Time Photometric Stereo Using an Event Camera ” and “ pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction. ”"}
+{"idx": 9, "title": "Bohan Yu - Google Scholar", "date": "", "ddg_snippet": "EventPS : Real - time photometric stereo using an event camera .Releaps: Reinforcement learning-based illumination planning for generalized photometric stereo .", "subpage_snippet": "", "source": "scholar.google.com", "link": "https://scholar.google.com/citations?user=pagJhBAAAAAJ&hl=en", "content": "EventPS : Real - time photometric stereo using an event camera .Releaps: Reinforcement learning-based illumination planning for generalized photometric stereo ."}
diff --git a/data/sampled_jsons/adaptive_regression_differentially_private_median_condition_number_blowup.jsonl b/data/sampled_jsons/adaptive_regression_differentially_private_median_condition_number_blowup.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5961a95b44309fe77e7a8ed3985829cec477ef34
--- /dev/null
+++ b/data/sampled_jsons/adaptive_regression_differentially_private_median_condition_number_blowup.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "On Differential Privacy for Adaptively Solving Search ...", "date": "", "ddg_snippet": "by S Feng — In this paper we investigate the use of differential privacy for adaptive queries to {\\it search} problems, which are significantly more ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=kEn7Wt6Yj2", "content": "by S Feng — In this paper we investigate the use of differential privacy for adaptive queries to {\\it search} problems, which are significantly more ..."}
+{"idx": 1, "title": "On Differential Privacy for Adaptively Solving Search ...", "date": "", "ddg_snippet": "by S Feng · 2025 — For adaptive regression , we show how to upgrade the private median framework of [BKM+22] to output the solution vector, and how to obtain ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2506.05503", "content": "by S Feng · 2025 — For adaptive regression , we show how to upgrade the private median framework of [BKM+22] to output the solution vector, and how to obtain ..."}
+{"idx": 2, "title": "On Differential Privacy for Adaptively Solving Search ...", "date": "", "ddg_snippet": "5 Jun 2025 — In this paper, we investigate the use of differential privacy for adaptive queries to search problems, which are significantly more challenging ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.05503v1", "content": "5 Jun 2025 — In this paper, we investigate the use of differential privacy for adaptive queries to search problems, which are significantly more challenging ..."}
+{"idx": 3, "title": "ICML Poster On Differential Privacy for Adaptively Solving ...", "date": "", "ddg_snippet": "Adaptive Regression via Differentially Private Median and ℓ∞ Guarantee ... To offset the blowup in condition number , we scale down α by a factor of κ , and ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/44265", "content": "Adaptive Regression via Differentially Private Median and ℓ∞ Guarantee ... To offset the blowup in condition number , we scale down α by a factor of κ , and ..."}
+{"idx": 4, "title": "primo: private regression in multiple outcomes", "date": "", "ddg_snippet": "by S Neel · 2023 — We introduce a new differentially private regression setting we call Private Regression in Multiple. Outcomes (PRIMO), inspired the common situation where a ... 21 pages", "subpage_snippet": "", "source": "www.hbs.edu", "link": "https://www.hbs.edu/ris/Publication+Files/2303.04195_0ecfb110-7ab1-48c3-91dc-7e9e05db56d9.pdf", "content": "by S Neel · 2023 — We introduce a new differentially private regression setting we call Private Regression in Multiple. Outcomes (PRIMO), inspired the common situation where a ... 21 pages"}
+{"idx": 5, "title": "Private Regression In Multiple Outcomes (PRIMO)", "date": "", "ddg_snippet": "by S Neel — We introduce a new differentially private regression setting we call Private Regres- sion in Multiple Outcomes (PRIMO) inspired the common situation in the ...", "subpage_snippet": "", "source": "tpdp.journalprivacyconfidentiality.org", "link": "https://tpdp.journalprivacyconfidentiality.org/2022/papers/primo.pdf", "content": "by S Neel — We introduce a new differentially private regression setting we call Private Regres- sion in Multiple Outcomes (PRIMO) inspired the common situation in the ..."}
+{"idx": 6, "title": "Differentially private statistical estimation", "date": "", "ddg_snippet": "Differential Privacy is now a gold standard for data privacy in many learning and statistical tasks. It has enjoyed over a decade of intense study, ... 245 pages", "subpage_snippet": "", "source": "repository.library.northeastern.edu", "link": "https://repository.library.northeastern.edu/files/neu:bz613973x/fulltext.pdf", "content": "Differential Privacy is now a gold standard for data privacy in many learning and statistical tasks. It has enjoyed over a decade of intense study, ... 245 pages"}
+{"idx": 7, "title": "Upper and Lower Bounds for Privacy and Adaptivity in ...", "date": "", "ddg_snippet": "by TA Steinke · 2016 · Cited by 10 — We prove lower bounds in the adaptive data analysis setting that nearly match the upper bounds given by differential privacy . Namely, we show that, given n ...", "subpage_snippet": "", "source": "projects.iq.harvard.edu", "link": "https://projects.iq.harvard.edu/files/privacytools/files/steinke-dissertation-2016.pdf", "content": "by TA Steinke · 2016 · Cited by 10 — We prove lower bounds in the adaptive data analysis setting that nearly match the upper bounds given by differential privacy . Namely, we show that, given n ..."}
+{"idx": 8, "title": "Fast Algorithm for Dynamic Kronecker Projection Maintenance", "date": "", "ddg_snippet": "by Z Song · 2023 · Cited by 33 — For each coordinate, we use a differentially private median procedure together with Chernoff bound. However, note that we have to estimate k entries and we ...", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v202/song23i/song23i.pdf", "content": "by Z Song · 2023 · Cited by 33 — For each coordinate, we use a differentially private median procedure together with Chernoff bound. However, note that we have to estimate k entries and we ..."}
+{"idx": 9, "title": "Robustness Implies Privacy in Statistical Estimation", "date": "", "ddg_snippet": "by SB Hopkins · 2023 · Cited by 79 — The first polynomial-time algorithms to learn high-dimensional Gauss- ian distributions with nearly-optimal sample complexity subject to differential privacy : ...", "subpage_snippet": "", "source": "dspace.mit.edu", "link": "https://dspace.mit.edu/bitstream/handle/1721.1/150986/3564246.3585115.pdf?sequence=1&isAllowed=y", "content": "by SB Hopkins · 2023 · Cited by 79 — The first polynomial-time algorithms to learn high-dimensional Gauss- ian distributions with nearly-optimal sample complexity subject to differential privacy : ..."}
diff --git a/data/sampled_jsons/arXiv2404.08819_abstract_constant_depth_finite_state_machines.jsonl b/data/sampled_jsons/arXiv2404.08819_abstract_constant_depth_finite_state_machines.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..7340d4a9e124f4d1bc4754ffeb7938ca2985dc77
--- /dev/null
+++ b/data/sampled_jsons/arXiv2404.08819_abstract_constant_depth_finite_state_machines.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Extracting Finite State Machines from Transformers", "date": "", "ddg_snippet": "Abstract . Fueled by the popularity of the transformer architecture in deep learning, several works have investigated what formal languages a transformer can learn.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/384769312_Extracting_Finite_State_Machines_from_Transformers", "content": "Abstract . Fueled by the popularity of the transformer architecture in deep learning, several works have investigated what formal languages a transformer can learn."}
+{"idx": 1, "title": "Illusion of State in SSMs like Mamba · Data Artificer and code:Breaker", "date": "", "ddg_snippet": "Abstract Link to heading Last time we looked into the weak points of State Space Models (Mamba, Mamba2), especially when compared with Attention-Based models (LLama, GPT-like).", "subpage_snippet": "", "source": "n1o.github.io", "link": "https://n1o.github.io/posts/ssm-the-illusion/", "content": "Abstract Link to heading Last time we looked into the weak points of State Space Models (Mamba, Mamba2), especially when compared with Attention-Based models (LLama, GPT-like)."}
+{"idx": 2, "title": "Understanding the Differences in Foundation Models: Attention,", "date": "", "ddg_snippet": "Why does state expansion help to improve performance of RNNs and SSMs? TL;DR: This is related to the second question: state expansion increases the ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2405.15731v3", "content": "Why does state expansion help to improve performance of RNNs and SSMs? TL;DR: This is related to the second question: state expansion increases the ..."}
+{"idx": 3, "title": "Yuri Gurevich: Annotated works", "date": "", "ddg_snippet": "[255] Yuri Gurevich The umbilical cord of finite model theory Bulletin of the EATCS 139, February 2023 Journal of Logic and Computation September 2023 DOI arXiv :2301.09145. ABSTRACT . Model theory was born and developed as a part of mathematical logic.", "subpage_snippet": "", "source": "web.eecs.umich.edu", "link": "https://web.eecs.umich.edu/~gurevich/annotated.htm", "content": "[255] Yuri Gurevich The umbilical cord of finite model theory Bulletin of the EATCS 139, February 2023 Journal of Logic and Computation September 2023 DOI arXiv :2301.09145. ABSTRACT . Model theory was born and developed as a part of mathematical logic."}
+{"idx": 4, "title": "Math and Theory Proof Methods: A Comparison | Gödel's Lost Letter...", "date": "", "ddg_snippet": "For a relatively recent paper on constant depth arithmetic circuits take a look at the paper of Maurice Jansen and Ken Regan. They use a discrete uncertainty principle to prove their lower bound.", "subpage_snippet": "", "source": "rjlipton.com", "link": "https://rjlipton.com/2010/10/08/math-and-theory-proof-methods-a-comparison/", "content": "For a relatively recent paper on constant depth arithmetic circuits take a look at the paper of Maurice Jansen and Ken Regan. They use a discrete uncertainty principle to prove their lower bound."}
+{"idx": 5, "title": "Through the L ens of C omplexity t heory", "date": "", "ddg_snippet": "• Regular Languages: The class of languages recognizable by finite automata or equivalently, by regular expressions. They are the simplest class in the Chomsky hierarchy and can be recognized using constant memory.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=DhdqML3FdM", "content": "• Regular Languages: The class of languages recognizable by finite automata or equivalently, by regular expressions. They are the simplest class in the Chomsky hierarchy and can be recognized using constant memory."}
+{"idx": 6, "title": "Communications of the Association for Computing Machinery", "date": "", "ddg_snippet": "Webb Miller Toward Abstract Numerical Analysis . . . 399--408. G. A. Watson An Algorithm for the Inversion of Block.", "subpage_snippet": "", "source": "ftp.math.utah.edu", "link": "https://ftp.math.utah.edu/pub/tex/bib/toc/jacm.html", "content": "Webb Miller Toward Abstract Numerical Analysis . . . 399--408. G. A. Watson An Algorithm for the Inversion of Block."}
+{"idx": 7, "title": "Full text of \"Models of Computation: Exploring the Power of Computing\"", "date": "", "ddg_snippet": "In Chapters 4 and 5 the finite - state machine, pushdown automaton, and Turing machine are characterized by their language recognition capability.", "subpage_snippet": "", "source": "archive.org", "link": "https://archive.org/stream/ost-computer-science-modelsofcomputation/ModelsOfComputation_djvu.txt", "content": "In Chapters 4 and 5 the finite - state machine, pushdown automaton, and Turing machine are characterized by their language recognition capability."}
+{"idx": 8, "title": "International Journal of Engineering Research and Applications(IJERA)", "date": "", "ddg_snippet": "Abstract . Hard turning is a machining process defined of hardness higher than 45 HRC under appropriate cutting tools and cutting speed. The objective of this paper is to investigate the optimum process parameters for a particular work piece-tool material combination. .", "subpage_snippet": "", "source": "www.academia.edu", "link": "https://www.academia.edu/3248836/International_Journal_of_Engineering_Research_and_Applications_IJERA_", "content": "Abstract . Hard turning is a machining process defined of hardness higher than 45 HRC under appropriate cutting tools and cutting speed. The objective of this paper is to investigate the optimum process parameters for a particular work piece-tool material combination. ."}
+{"idx": 9, "title": "DESIGN CONSIDERATIONS", "date": "", "ddg_snippet": "Figure 7 : The annotated 512-point fast Fourier transform program. To simplify the presentation, the length of the data-list to be transformed has been xed to 512 elements, which explains the origin of the constant 128 in the program.", "subpage_snippet": "", "source": "wim.vree.org", "link": "https://wim.vree.org/pubs/thesis89.pdf", "content": "Figure 7 : The annotated 512-point fast Fourier transform program. To simplify the presentation, the length of the data-list to be transformed has been xed to 512 elements, which explains the origin of the constant 128 in the program."}
diff --git a/data/sampled_jsons/arxiv_2305.09828_abstract_Mimetic_Initialization_of_Self-Attention_Layers_year_2023.jsonl b/data/sampled_jsons/arxiv_2305.09828_abstract_Mimetic_Initialization_of_Self-Attention_Layers_year_2023.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b714a822711e76f242e307fb561d12e77f57ff85
--- /dev/null
+++ b/data/sampled_jsons/arxiv_2305.09828_abstract_Mimetic_Initialization_of_Self-Attention_Layers_year_2023.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2305.09828] Mimetic Initialization of Self-Attention Layers", "date": "", "ddg_snippet": "May 16, 2023 · It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2305.09828", "content": "May 16, 2023 · It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their ..."}
+{"idx": 1, "title": "Mimetic Initialization of Self-Attention Layers - PMLR", "date": "", "ddg_snippet": "Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their pre-trained counterparts allows us to train vanilla Transformers faster and to higher final accuracies, particularly on vision tasks such as CIFAR-10 and ImageNet classification, where we see gains in accuracy of over 5% and 4 ...", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v202/trockman23a.html", "content": "Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their pre-trained counterparts allows us to train vanilla Transformers faster and to higher final accuracies, particularly on vision tasks such as CIFAR-10 and ImageNet classification, where we see gains in accuracy of over 5% and 4 ..."}
+{"idx": 2, "title": "Mimetic Initialization of Self-Attention Layers - NASA/ADS", "date": "", "ddg_snippet": "Abstract It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more ...", "subpage_snippet": "", "source": "ui.adsabs.harvard.edu", "link": "https://ui.adsabs.harvard.edu/abs/2023arXiv230509828T/abstract", "content": "Abstract It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more ..."}
+{"idx": 3, "title": "Mimetic Initialization of Self-Attention Layers - LOVO AI", "date": "", "ddg_snippet": "Aug 23, 2023 · Figure 3. Attention maps from one CIFAR-10 batch for 1, 4, and 11th transformer layers of ViT-Tiny (a) untrained, (b) CIFAR-10 trained (c) ImageNet pretrained (d) Mimetic initialization (e) Mimetic initialization + CIFAR-10 trained. Mimetic initialization was shown to also work with language models. While the improvements in the metrics are not as obvious as those in vision models, results ...", "subpage_snippet": "", "source": "lovo.ai", "link": "https://lovo.ai/post/mimetic-initialization-of-self-attention-layers", "content": "Aug 23, 2023 · Figure 3. Attention maps from one CIFAR-10 batch for 1, 4, and 11th transformer layers of ViT-Tiny (a) untrained, (b) CIFAR-10 trained (c) ImageNet pretrained (d) Mimetic initialization (e) Mimetic initialization + CIFAR-10 trained. Mimetic initialization was shown to also work with language models. While the improvements in the metrics are not as obvious as those in vision models, results ..."}
+{"idx": 4, "title": "\"Mimetic Initialization of Self-Attention Layers.\" - dblp", "date": "", "ddg_snippet": "Bibliographic details on Mimetic Initialization of Self-Attention Layers .", "subpage_snippet": "", "source": "dblp.org", "link": "https://dblp.org/rec/journals/corr/abs-2305-09828", "content": "Bibliographic details on Mimetic Initialization of Self-Attention Layers ."}
+{"idx": 5, "title": "Paper page - Mimetic Initialization of Self-Attention Layers", "date": "", "ddg_snippet": "May 16, 2023 · Abstract It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so ...", "subpage_snippet": "", "source": "huggingface.co", "link": "https://huggingface.co/papers/2305.09828", "content": "May 16, 2023 · Abstract It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so ..."}
+{"idx": 6, "title": "Mimetic initialization of self-attention layers | Proceedings ...", "date": "", "ddg_snippet": "Jul 23, 2023 · It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their ...", "subpage_snippet": "", "source": "dl.acm.org", "link": "https://dl.acm.org/doi/10.5555/3618408.3619843", "content": "Jul 23, 2023 · It is notoriously difficult to train Transformers on small datasets; typically, large pre-trained models are instead used as the starting point. We explore the weights of such pre-trained Transformers (particularly for vision) to attempt to find reasons for this discrepancy. Surprisingly, we find that simply initializing the weights of self-attention layers so that they \"look\" more like their ..."}
+{"idx": 7, "title": "‘self-attention’ directory · Gwern.net", "date": "", "ddg_snippet": "... Abstraction and Reasoning Corpus With Vision Transformers: the ... Mimetic Initialization of Self - Attention Layers ”, Trockman & Kolter 2023", "subpage_snippet": "", "source": "gwern.net", "link": "https://gwern.net/doc/ai/nn/transformer/attention/index", "content": "... Abstraction and Reasoning Corpus With Vision Transformers: the ... Mimetic Initialization of Self - Attention Layers ”, Trockman & Kolter 2023"}
+{"idx": 8, "title": "On the Surprising Effectiveness of Attention Transfer for", "date": "", "ddg_snippet": "They are i) partial transfer with a subset of layers or heads; ii) variants of our method that transfer other attention -related activations; and ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.09702v1", "content": "They are i) partial transfer with a subset of layers or heads; ii) variants of our method that transfer other attention -related activations; and ..."}
+{"idx": 9, "title": "Characterizing Large Language Model Geometry Helps Solve", "date": "", "ddg_snippet": "In particular, each LLM layer employs a multi-head self - attention block (MHA), and a multilayer perceptron block (MLP) (Vaswani et al., 2017 ) .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2312.01648v3", "content": "In particular, each LLM layer employs a multi-head self - attention block (MHA), and a multilayer perceptron block (MLP) (Vaswani et al., 2017 ) ."}
diff --git a/data/sampled_jsons/arxiv_2411.13653_No_Free_Delivery_Service_abstract.jsonl b/data/sampled_jsons/arxiv_2411.13653_No_Free_Delivery_Service_abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..f3a6f882a4e33efb1689d9924446abdf833c2c31
--- /dev/null
+++ b/data/sampled_jsons/arxiv_2411.13653_No_Free_Delivery_Service_abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2411.13653] No Free Delivery Service: Epistemic limits of ... No Free Delivery Service: Epistemic limits of passive data ... [2411.13653] Untitled Document [ar5iv.labs.arxiv.org] GitHub - Zhangpengfeiscu/zotero-arxiv-daily: Recommend new ... vocab.txt · unitary/toxic-bert at ... U.S. Patent for Visual detection and localization of package ... Transformer Led Policing model: a framework for applying ...", "date": "", "ddg_snippet": "Nov 20, 2024 · Abstract page for arXiv paper 2411.13653 : No Free Delivery Service : Epistemic limits of passive data collection in complex social systems Nov 20, 2024 · Rapid model validation via the train-test paradigm has been a key driver for the breathtaking progress in machine learning and AI. However, modern AI systems often depend on a combination of tasks and data collection practices that violate all assumptions ensuring test validity. Yet, without rigorous model validation we cannot ensure the intended outcomes of deployed AI systems, including ... Dec 5, 2024 · Generated on Thu Dec 5 16:20:01 2024 by L aT eXML Zotero- arXiv -Daily finds arxiv papers that may attract you based on the context of your Zotero library, and then sends the result to your mailbox📮. It can be deployed as Github Action Workflow with zero cost, no installation, and few configuration of Github Action environment variables for daily automatic delivery . Use this modela447313 Jun 20, 2023 · Unmanned aerial vehicles (also referred to as drones) can be adapted for package delivery missions to provide an aerial delivery service . One type of unmanned aerial vehicle (UAV) is a vertical takeoff and landing (VTOL) UAV. VTOL UAVs are particularly well-suited for package delivery missions. 3 days ago · Abstract This article provides two outputs. First, it introduces the Transformer Led Policing (TLP) model, which is a structured framework for integrating Generative Artificial Intelligence (GenAI) in policing. The purpose of developing the framework is to provide a model for police practitioners and researchers considering implementing or studying GenAI to conduct policing functions. Doing so ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2411.13653", "content": "Nov 20, 2024 · Abstract page for arXiv paper 2411.13653 : No Free Delivery Service : Epistemic limits of passive data collection in complex social systems Nov 20, 2024 · Rapid model validation via the train-test paradigm has been a key driver for the breathtaking progress in machine learning and AI. However, modern AI systems often depend on a combination of tasks and data collection practices that violate all assumptions ensuring test validity. Yet, without rigorous model validation we cannot ensure the intended outcomes of deployed AI systems, including ... Dec 5, 2024 · Generated on Thu Dec 5 16:20:01 2024 by L aT eXML Zotero- arXiv -Daily finds arxiv papers that may attract you based on the context of your Zotero library, and then sends the result to your mailbox📮. It can be deployed as Github Action Workflow with zero cost, no installation, and few configuration of Github Action environment variables for daily automatic delivery . Use this modela447313 Jun 20, 2023 · Unmanned aerial vehicles (also referred to as drones) can be adapted for package delivery missions to provide an aerial delivery service . One type of unmanned aerial vehicle (UAV) is a vertical takeoff and landing (VTOL) UAV. VTOL UAVs are particularly well-suited for package delivery missions. 3 days ago · Abstract This article provides two outputs. First, it introduces the Transformer Led Policing (TLP) model, which is a structured framework for integrating Generative Artificial Intelligence (GenAI) in policing. The purpose of developing the framework is to provide a model for police practitioners and researchers considering implementing or studying GenAI to conduct policing functions. Doing so ..."}
+{"idx": 1, "title": "GitHub - Zhangpengfeiscu/zotero-arxiv-daily: Recommend new ...", "date": "", "ddg_snippet": "Zotero- arXiv -Daily finds arxiv papers that may attract you based on the context of your Zotero library, and then sends the result to your mailbox📮. It can be deployed as Github Action Workflow with zero cost, no installation, and few configuration of Github Action environment variables for daily automatic delivery .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/Zhangpengfeiscu/zotero-arxiv-daily", "content": "Zotero- arXiv -Daily finds arxiv papers that may attract you based on the context of your Zotero library, and then sends the result to your mailbox📮. It can be deployed as Github Action Workflow with zero cost, no installation, and few configuration of Github Action environment variables for daily automatic delivery ."}
+{"idx": 2, "title": "TempGmailer – Free Temp Gmail & Disposable Emails", "date": "", "ddg_snippet": "TempGmailer - Free temporary Gmail accounts and disposable email service . Free disposable emails that auto-expire in 24 hours. No signup needed.", "subpage_snippet": "", "source": "tempgmailer.com", "link": "https://tempgmailer.com/", "content": "TempGmailer - Free temporary Gmail accounts and disposable email service . Free disposable emails that auto-expire in 24 hours. No signup needed."}
+{"idx": 3, "title": "Driver Booster 12 Free – Скачать", "date": "", "ddg_snippet": "Бесплатно. Windows. Категория: Программы для драйверов. Driver Booster Free - полезная программа, автоматически сканирующая и определяющая драйвера на ПК. После нахождения устаревших драйверов программа предлагает загрузить и установить обновления...", "subpage_snippet": "", "source": "www.SoftPortal.com", "link": "https://www.SoftPortal.com/software-31816-driver-booster-free.html", "content": "Бесплатно. Windows. Категория: Программы для драйверов. Driver Booster Free - полезная программа, автоматически сканирующая и определяющая драйвера на ПК. После нахождения устаревших драйверов программа предлагает загрузить и установить обновления..."}
+{"idx": 4, "title": "Free Image Background Remover | Adobe Express", "date": "", "ddg_snippet": "This free background remover app allows you to highlight the subject of your photo and create a clear background, so you can place your new image into a variety of new designs and destinations.", "subpage_snippet": "", "source": "www.adobe.com", "link": "https://www.adobe.com/express/feature/image/remove-background", "content": "This free background remover app allows you to highlight the subject of your photo and create a clear background, so you can place your new image into a variety of new designs and destinations."}
+{"idx": 5, "title": "Free Online Excel to CSV Converter | No Excel Needed", "date": "", "ddg_snippet": "This free online service allows you to convert Excel files to CSV without opening them. Easily convert multiple Excel files and sheets into CSV format.", "subpage_snippet": "", "source": "boost-tool.com", "link": "https://boost-tool.com/en/tools/excel_to_csv", "content": "This free online service allows you to convert Excel files to CSV without opening them. Easily convert multiple Excel files and sheets into CSV format."}
+{"idx": 6, "title": "Humanize AI Text Free - Get 100% Human Score", "date": "", "ddg_snippet": "Free AI humanizer is an accurate and efficient ai to human text converter. Make your content more human-like and undetectable, easily bypasses AI detection.", "subpage_snippet": "", "source": "decopy.ai", "link": "https://decopy.ai/ai-humanizer/", "content": "Free AI humanizer is an accurate and efficient ai to human text converter. Make your content more human-like and undetectable, easily bypasses AI detection."}
+{"idx": 7, "title": "Incredibox Blinding Lights 9.0 APK MOD Game Free Download for...", "date": "", "ddg_snippet": "Use Automatic (Shuffle) Mode if Unsure: If I’m lacking ideas, I can let the game pick icons randomly in Automatic or Shuffle Mode to generate a base, which I can then tweak. Conclusion – Free Download Incredibox Blinding Lights APK Latest Version for Android.", "subpage_snippet": "", "source": "modlegen.com", "link": "https://modlegen.com/incredibox-blinding-lights/", "content": "Use Automatic (Shuffle) Mode if Unsure: If I’m lacking ideas, I can let the game pick icons randomly in Automatic or Shuffle Mode to generate a base, which I can then tweak. Conclusion – Free Download Incredibox Blinding Lights APK Latest Version for Android."}
+{"idx": 8, "title": "No Free Delivery Service: Epistemic limits of passive data ...", "date": "", "ddg_snippet": "Nov 20, 2024 · Rapid model validation via the train-test paradigm has been a key driver for the breathtaking progress in machine learning and AI. However, modern AI systems often depend on a combination of tasks and data collection practices that violate all assumptions ensuring test validity. Yet, without rigorous model validation we cannot ensure the intended outcomes of deployed AI systems, including ...", "subpage_snippet": "", "source": "researchtrend.ai", "link": "https://researchtrend.ai/papers/2411.13653", "content": "Nov 20, 2024 · Rapid model validation via the train-test paradigm has been a key driver for the breathtaking progress in machine learning and AI. However, modern AI systems often depend on a combination of tasks and data collection practices that violate all assumptions ensuring test validity. Yet, without rigorous model validation we cannot ensure the intended outcomes of deployed AI systems, including ..."}
+{"idx": 9, "title": "[2411.13653] Untitled Document [ar5iv.labs.arxiv.org]", "date": "", "ddg_snippet": "Dec 5, 2024 · Generated on Thu Dec 5 16:20:01 2024 by L aT eXML", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2411.13653", "content": "Dec 5, 2024 · Generated on Thu Dec 5 16:20:01 2024 by L aT eXML"}
diff --git a/data/sampled_jsons/causal_representation_learning_variational_autoencoder_noise_robust_2024.jsonl b/data/sampled_jsons/causal_representation_learning_variational_autoencoder_noise_robust_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..f0eb2265d5723de185ec68a44fe6c812edc9535b
--- /dev/null
+++ b/data/sampled_jsons/causal_representation_learning_variational_autoencoder_noise_robust_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Causality - Wikipedia", "date": "", "ddg_snippet": "In general, a process can have multiple causes, [1] which are also said to be causal factors for it, and all lie in its past. An effect can in turn be a cause of, or causal factor for, many other effects, which all lie in its future.", "subpage_snippet": "", "source": "en.m.wikipedia.org", "link": "https://en.m.wikipedia.org/wiki/Causality", "content": "In general, a process can have multiple causes, [1] which are also said to be causal factors for it, and all lie in its past. An effect can in turn be a cause of, or causal factor for, many other effects, which all lie in its future."}
+{"idx": 1, "title": "CAUSAL Definition & Meaning - Merriam-Webster", "date": "", "ddg_snippet": "The meaning of CAUSAL is expressing or indicating cause : causative. How to use causal in a sentence.", "subpage_snippet": "", "source": "www.merriam-webster.com", "link": "https://www.merriam-webster.com/dictionary/causal", "content": "The meaning of CAUSAL is expressing or indicating cause : causative. How to use causal in a sentence."}
+{"idx": 2, "title": "CAUSAL | English meaning - Cambridge Dictionary", "date": "", "ddg_snippet": "CAUSAL definition: 1. a relationship, link, etc. between two things in which one causes the other: 2. a relationship…. Learn more.", "subpage_snippet": "", "source": "dictionary.cambridge.org", "link": "https://dictionary.cambridge.org/dictionary/english/causal", "content": "CAUSAL definition: 1. a relationship, link, etc. between two things in which one causes the other: 2. a relationship…. Learn more."}
+{"idx": 3, "title": "CAUSAL Definition & Meaning | Dictionary .com", "date": "", "ddg_snippet": "Causal definition: of, constituting, or implying a cause.. See examples of CAUSAL used in a sentence.", "subpage_snippet": "", "source": "www.dictionary.com", "link": "https://www.dictionary.com/browse/causal", "content": "Causal definition: of, constituting, or implying a cause.. See examples of CAUSAL used in a sentence."}
+{"idx": 4, "title": "CAUSAL definition and meaning | Collins English Dictionary", "date": "", "ddg_snippet": "If there is a causal relationship between two things, one thing is responsible for causing the other thing.", "subpage_snippet": "", "source": "www.collinsdictionary.com", "link": "https://www.collinsdictionary.com/dictionary/english/causal", "content": "If there is a causal relationship between two things, one thing is responsible for causing the other thing."}
+{"idx": 5, "title": "Causal - definition of causal by The Free Dictionary", "date": "", "ddg_snippet": "1. Of, involving, or constituting a cause: a causal relationship between scarcity of goods and higher prices. 2. Indicative of or expressing a cause.", "subpage_snippet": "", "source": "www.thefreedictionary.com", "link": "https://www.thefreedictionary.com/causal", "content": "1. Of, involving, or constituting a cause: a causal relationship between scarcity of goods and higher prices. 2. Indicative of or expressing a cause."}
+{"idx": 6, "title": "causal adjective - Definition, pictures, pronunciation and usage...", "date": "", "ddg_snippet": "Definition of causal adjective in Oxford Advanced Learner's Dictionary. Meaning, pronunciation, picture, example sentences, grammar, usage notes, synonyms and more.", "subpage_snippet": "", "source": "www.oxfordlearnersdictionaries.com", "link": "https://www.oxfordlearnersdictionaries.com/definition/english/causal", "content": "Definition of causal adjective in Oxford Advanced Learner's Dictionary. Meaning, pronunciation, picture, example sentences, grammar, usage notes, synonyms and more."}
+{"idx": 7, "title": "causal , adj. & n. meanings, etymology and more | Oxford English...", "date": "", "ddg_snippet": "causal , adj. & n. meanings, etymology, pronunciation and more in the Oxford English Dictionary", "subpage_snippet": "", "source": "www.oed.com", "link": "https://www.oed.com/dictionary/causal_adj", "content": "causal , adj. & n. meanings, etymology, pronunciation and more in the Oxford English Dictionary"}
+{"idx": 8, "title": "causal - Wiktionary, the free dictionary", "date": "", "ddg_snippet": "Aug 28, 2025 · causal (comparative more causal , superlative most causal ) There is no causal relationship between eating carrots and seeing in the dark.", "subpage_snippet": "", "source": "en.m.wiktionary.org", "link": "https://en.m.wiktionary.org/wiki/causal", "content": "Aug 28, 2025 · causal (comparative more causal , superlative most causal ) There is no causal relationship between eating carrots and seeing in the dark."}
+{"idx": 9, "title": "Causal - Definition, Meaning & Synonyms | Vocabulary.com", "date": "", "ddg_snippet": "Causal is a variation of the word cause , which should be a clue to its meaning. A cause is what makes something happen: the notebook flew across the room because you threw it, so your throwing it was causal. If a bolt of lightning set a statue on fire, the lightning was causal for the fire.", "subpage_snippet": "", "source": "www.vocabulary.com", "link": "https://www.vocabulary.com/dictionary/causal", "content": "Causal is a variation of the word cause , which should be a clue to its meaning. A cause is what makes something happen: the notebook flew across the room because you threw it, so your throwing it was causal. If a bolt of lightning set a statue on fire, the lightning was causal for the fire."}
diff --git a/data/sampled_jsons/cross-blended_images_deepfake_detection_paper_title.jsonl b/data/sampled_jsons/cross-blended_images_deepfake_detection_paper_title.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..3407831c9d87096c27b1bbb5ea2f50c3c43217e0
--- /dev/null
+++ b/data/sampled_jsons/cross-blended_images_deepfake_detection_paper_title.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2204.08376] Detecting Deepfakes with Self-Blended Images", "date": "", "ddg_snippet": "View a PDF of the paper titled Detecting Deepfakes with Self- Blended Images , by Kaede Shiohara and Toshihiko Yamasaki", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2204.08376", "content": "View a PDF of the paper titled Detecting Deepfakes with Self- Blended Images , by Kaede Shiohara and Toshihiko Yamasaki"}
+{"idx": 1, "title": "SocialDF: Benchmark Dataset and Detection Model for Mitigating", "date": "", "ddg_snippet": "2024 ) , which integrate both audio and visual modalities, are considered the most robust for deepfake detection , as they enable cross -modal ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.05538v1", "content": "2024 ) , which integrate both audio and visual modalities, are considered the most robust for deepfake detection , as they enable cross -modal ..."}
+{"idx": 2, "title": "GitHub - flyingby/Awesome-Deepfake-Generation-and-Detection: A", "date": "", "ddg_snippet": "... manipulation in Deepfake , encompassing Face Swapping , Face Reenactment , Talking Face Generation , Face Attribute Editing and Forgery Detection .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/flyingby/Awesome-Deepfake-Generation-and-Detection", "content": "... manipulation in Deepfake , encompassing Face Swapping , Face Reenactment , Talking Face Generation , Face Attribute Editing and Forgery Detection ."}
+{"idx": 3, "title": "GitHub - mapooon/SelfBlendedImages: [CVPR 2022 Oral] Detecting", "date": "", "ddg_snippet": "inproceedings { shiohara2022detecting , title = { Detecting Deepfakes with Self- Blended Images } , author = { Shiohara, Kaede and Yamasaki, Toshihiko ...", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/mapooon/SelfBlendedImages", "content": "inproceedings { shiohara2022detecting , title = { Detecting Deepfakes with Self- Blended Images } , author = { Shiohara, Kaede and Yamasaki, Toshihiko ..."}
+{"idx": 4, "title": "CVPR 2024 Papers", "date": "", "ddg_snippet": "Image -to- Image Matching via Foundation Models: A New Perspective for Open-Vocabulary Semantic Segmentation ... based Visual Relationship Detection", "subpage_snippet": "", "source": "cvpr.thecvf.com", "link": "https://cvpr.thecvf.com/virtual/2024/papers.html", "content": "Image -to- Image Matching via Foundation Models: A New Perspective for Open-Vocabulary Semantic Segmentation ... based Visual Relationship Detection"}
+{"idx": 5, "title": "CVPR 2023 Papers", "date": "", "ddg_snippet": "All-in-One Image Restoration for Unknown Degradations ... Towards Robust Tampered Text Detection in Document Image : New Dataset and New Solution", "subpage_snippet": "", "source": "cvpr.thecvf.com", "link": "https://cvpr.thecvf.com/virtual/2023/papers.html", "content": "All-in-One Image Restoration for Unknown Degradations ... Towards Robust Tampered Text Detection in Document Image : New Dataset and New Solution"}
+{"idx": 6, "title": "Researchers Identify a Resilient Trait of Deepfakes That Could", "date": "", "ddg_snippet": "The new paper is titled Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption , and comes from researchers at Tsinghua University, the ...", "subpage_snippet": "", "source": "www.unite.ai", "link": "https://www.unite.ai/researchers-identify-a-resilient-trait-of-deepfakes-that-could-aid-long-term-detection/", "content": "The new paper is titled Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption , and comes from researchers at Tsinghua University, the ..."}
+{"idx": 7, "title": "Enhancing Authenticity Verification with Transfer Learning and", "date": "", "ddg_snippet": "Detecting and mitigating these threats is imperative for fortifying defenses and safeguarding information integrity.This paper tackles the ...", "subpage_snippet": "", "source": "repository.uel.ac.uk", "link": "https://repository.uel.ac.uk/item/8z540", "content": "Detecting and mitigating these threats is imperative for fortifying defenses and safeguarding information integrity.This paper tackles the ..."}
+{"idx": 8, "title": "GFADE: generalized feature adaptation and discrimination", "date": "", "ddg_snippet": "... demonstrates state-of-the-art performance across various deepfake detection scenarios, including cross -dataset and cross -manipulation settings.", "subpage_snippet": "", "source": "peerj.com", "link": "https://peerj.com/articles/cs-2879/", "content": "... demonstrates state-of-the-art performance across various deepfake detection scenarios, including cross -dataset and cross -manipulation settings."}
+{"idx": 9, "title": "27TH EUROPEAN CONFERENCE ON ARTIFICIAL INTELLIGENCE", "date": "", "ddg_snippet": "Detect Closer Surfaces that can be ... SaccadeMOT: Enhancing Object Detection and Tracking in Gigapixel Images via Scale-Aware Density Estimation", "subpage_snippet": "", "source": "www.ecai2024.eu", "link": "https://www.ecai2024.eu/programme/accepted-papers", "content": "Detect Closer Surfaces that can be ... SaccadeMOT: Enhancing Object Detection and Tracking in Gigapixel Images via Scale-Aware Density Estimation"}
diff --git a/data/sampled_jsons/data-dependent_privacy_machine_unlearning_differential_privacy.jsonl b/data/sampled_jsons/data-dependent_privacy_machine_unlearning_differential_privacy.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..1c6acfa281dc62baf62d7c3a270dd806f4130b5d
--- /dev/null
+++ b/data/sampled_jsons/data-dependent_privacy_machine_unlearning_differential_privacy.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Differential Privacy in Federated Learning: Mitigating", "date": "", "ddg_snippet": "Furthermore, recent literature also investigates machine unlearning techniques as a way to address privacy and compliance challenges in federated ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.13987v1", "content": "Furthermore, recent literature also investigates machine unlearning techniques as a way to address privacy and compliance challenges in federated ..."}
+{"idx": 1, "title": "Game-Theoretic Machine Unlearning: Mitigating Extra Privacy", "date": "", "ddg_snippet": "First, to satisfy privacy protection requirements, data removal conducted by machine unlearning usually reduces model performance [ 9 ] .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.03914v1", "content": "First, to satisfy privacy protection requirements, data removal conducted by machine unlearning usually reduces model performance [ 9 ] ."}
+{"idx": 2, "title": "Machine Unlearning of Traffic State Estimation and Prediction", "date": "", "ddg_snippet": "After receiving a data deletion request, machine unlearning can ”erase” the influence of the traffic data to be forgotten (e.g., a subset of ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2507.17984v1", "content": "After receiving a data deletion request, machine unlearning can ”erase” the influence of the traffic data to be forgotten (e.g., a subset of ..."}
+{"idx": 3, "title": "Towards Reliable Forgetting: A Survey on Machine Unlearning", "date": "", "ddg_snippet": "While algorithmic methods for machine unlearning offer mechanisms to erase specific training data from models, concerns about trust and transparency ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2506.15115v1", "content": "While algorithmic methods for machine unlearning offer mechanisms to erase specific training data from models, concerns about trust and transparency ..."}
+{"idx": 4, "title": "Unveiling Privacy Risks in Machine Unlearning: Reconstruction", "date": "", "ddg_snippet": "Highlighting privacy risks in data deletion or machine unlearning , the findings emphasize the need for techniques like differential privacy .", "subpage_snippet": "", "source": "www.marktechpost.com", "link": "https://www.marktechpost.com/2024/12/27/unveiling-privacy-risks-in-machine-unlearning-reconstruction-attacks-on-deleted-data/", "content": "Highlighting privacy risks in data deletion or machine unlearning , the findings emphasize the need for techniques like differential privacy ."}
+{"idx": 5, "title": "Infosys Knowledge Institute | Privacy in the Digital World", "date": "", "ddg_snippet": "... or enacting data privacy regulations that make organizations accountable and respect individual choices about their data and protect their privacy ...", "subpage_snippet": "", "source": "www.infosys.com", "link": "https://www.infosys.com/iki/perspectives/privacy-digital.html", "content": "... or enacting data privacy regulations that make organizations accountable and respect individual choices about their data and protect their privacy ..."}
+{"idx": 6, "title": "How to Use Heuristics for Differential Privacy - Article -", "date": "", "ddg_snippet": "... dependent privacy guarantees to worst-case privacy guarantee that hold even when the heuristic standing in for the oracle might fail in adversarial ...", "subpage_snippet": "", "source": "www.hbs.edu", "link": "https://www.hbs.edu/faculty/Pages/item.aspx?num=60698", "content": "... dependent privacy guarantees to worst-case privacy guarantee that hold even when the heuristic standing in for the oracle might fail in adversarial ..."}
+{"idx": 7, "title": "Machine unlearning gets a practical privacy upgrade - Help Net", "date": "", "ddg_snippet": "... idea is to first train on a version of the dataset that has been transformed using a formal privacy model, either k-anonymity or differential privacy ...", "subpage_snippet": "", "source": "www.helpnetsecurity.com", "link": "https://www.helpnetsecurity.com/2025/07/17/machine-unlearning-privacy-upgrade/", "content": "... idea is to first train on a version of the dataset that has been transformed using a formal privacy model, either k-anonymity or differential privacy ..."}
+{"idx": 8, "title": "TPDP 2024 – Theory and Practice of Differential Privacy", "date": "", "ddg_snippet": "Next, I will cover data curation and potential intersections with differential privacy . ... Differential privacy and sublinear algorithms are both ...", "subpage_snippet": "", "source": "tpdp.journalprivacyconfidentiality.org", "link": "https://tpdp.journalprivacyconfidentiality.org/2024/", "content": "Next, I will cover data curation and potential intersections with differential privacy . ... Differential privacy and sublinear algorithms are both ..."}
+{"idx": 9, "title": "Publications – Privacy Technology Research Group", "date": "", "ddg_snippet": "Personalized Federated Learning With Differential Privacy and Convergence Guarantee Wei, Kang and Li, Jun and Ma, Chuan and Ding, Ming and Chen, Wen ...", "subpage_snippet": "", "source": "research.csiro.au", "link": "https://research.csiro.au/isp/research/publications/", "content": "Personalized Federated Learning With Differential Privacy and Convergence Guarantee Wei, Kang and Li, Jun and Ma, Chuan and Ding, Ming and Chen, Wen ..."}
diff --git a/data/sampled_jsons/do-calculus_Pearl_causal_inference_interventional_probability_since_2010.jsonl b/data/sampled_jsons/do-calculus_Pearl_causal_inference_interventional_probability_since_2010.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..61587c8f8ff7fa2f40ac6f26a0fbf3d139b47f47
--- /dev/null
+++ b/data/sampled_jsons/do-calculus_Pearl_causal_inference_interventional_probability_since_2010.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Do-calculus - Wikipedia", "date": "", "ddg_snippet": "Do-calculus is a set of mathematical rules devised by Judea Pearl in 1995 to determine whether causal effects can be identified from observational data under specific assumptions encoded in a causal graph. It provides a systematic method for transforming expressions involving the do -operator (representing interventions) into expressions involving only observable probabilities, enabling the ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Do-calculus", "content": "Do-calculus is a set of mathematical rules devised by Judea Pearl in 1995 to determine whether causal effects can be identified from observational data under specific assumptions encoded in a causal graph. It provides a systematic method for transforming expressions involving the do -operator (representing interventions) into expressions involving only observable probabilities, enabling the ..."}
+{"idx": 1, "title": "Causality - Wikipedia", "date": "", "ddg_snippet": "2.2 Probabilistic causation . 2.3 Causal calculus . Since causality is a subtle metaphysical notion, considerable intellectual effort, along with exhibition of evidence, is needed to establish knowledge of it in particular empirical circumstances.", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Causality", "content": "2.2 Probabilistic causation . 2.3 Causal calculus . Since causality is a subtle metaphysical notion, considerable intellectual effort, along with exhibition of evidence, is needed to establish knowledge of it in particular empirical circumstances."}
+{"idx": 2, "title": "The Do-Calculus Revisited Judea Pearl Keynote Lecture, August ...", "date": "", "ddg_snippet": "The do-calculus was developed in 1995 to facilitate the identification of causal effects in non-parametric mod-els. The completeness proofs of [Huang and Valtorta, 2006] and [Shpitser and Pearl , 2006] and the graphi-cal criteria of [Tian and Shpitser, 2010 ] have laid this identification problem to rest.", "subpage_snippet": "", "source": "ftp.cs.ucla.edu", "link": "https://ftp.cs.ucla.edu/pub/stat_ser/r402.pdf", "content": "The do-calculus was developed in 1995 to facilitate the identification of causal effects in non-parametric mod-els. The completeness proofs of [Huang and Valtorta, 2006] and [Shpitser and Pearl , 2006] and the graphi-cal criteria of [Tian and Shpitser, 2010 ] have laid this identification problem to rest."}
+{"idx": 3, "title": "Pearl’s Calculus of Intervention Is Complete", "date": "", "ddg_snippet": "Theorem (Rules of Do-Calculus ) [ Pearl , 2000] Let G be the directed acyclic graph associated with a causal model, and let P( ̇) stand for the probability distribution induced by that model.", "subpage_snippet": "", "source": "cse.sc.edu", "link": "https://cse.sc.edu/~mgv/papers/HuangValtortaUAI06.pdf", "content": "Theorem (Rules of Do-Calculus ) [ Pearl , 2000] Let G be the directed acyclic graph associated with a causal model, and let P( ̇) stand for the probability distribution induced by that model."}
+{"idx": 4, "title": "Causal Models > Supplement 2. The do-calculus (Stanford ...", "date": "", "ddg_snippet": "Supplement to Causal Models Supplement 2. The do- calculus The do -calculus is an axiomatic system for replacing probability formulas containing the do operator with ordinary conditional probabilities. It consists of three axiom schemas that provide graphical criteria for when certain substitutions may be made.", "subpage_snippet": "", "source": "plato.stanford.edu", "link": "https://plato.stanford.edu/entries/causal-models/do-calculus.html", "content": "Supplement to Causal Models Supplement 2. The do- calculus The do -calculus is an axiomatic system for replacing probability formulas containing the do operator with ordinary conditional probabilities. It consists of three axiom schemas that provide graphical criteria for when certain substitutions may be made."}
+{"idx": 5, "title": "The 3 rules of do-calculus – Joshua ... - Joshua Entrop", "date": "", "ddg_snippet": "Feb 10, 2024 · Pearl ’s do-calculus offers a comprehensive set of rules for identifying causal effects from a causal directed acyclic graph (DAG). Using those rules, one can identify causal effects even in situations in which the commonly applied back-door criteria does not hold. In this blog post, I demonstrate how you can use Pearl ’s do-calculus to identify causal effects based on various example DAGs.", "subpage_snippet": "", "source": "www.joshua-entrop.com", "link": "https://www.joshua-entrop.com/post/the_3_rules_of_do_calculus.html", "content": "Feb 10, 2024 · Pearl ’s do-calculus offers a comprehensive set of rules for identifying causal effects from a causal directed acyclic graph (DAG). Using those rules, one can identify causal effects even in situations in which the commonly applied back-door criteria does not hold. In this blog post, I demonstrate how you can use Pearl ’s do-calculus to identify causal effects based on various example DAGs."}
+{"idx": 6, "title": "ML beyond Curve Fitting: An Intro to Causal Inference and do ...", "date": "", "ddg_snippet": "May 24, 2018 · ML beyond Curve Fitting: An Intro to Causal Inference and do-Calculus Since writing this post back in 2018, I have extended this to a 4-part series on causal inference : ️️ Part 1: Intro to causal inference and do-calculus Part 2: Illustrating Interventions with a Toy Example Part 3: Counterfactuals Part 4: Causal Diagrams, Markov Factorization, Structural Equation Models You might have ...", "subpage_snippet": "", "source": "www.inference.vc", "link": "https://www.inference.vc/untitled/", "content": "May 24, 2018 · ML beyond Curve Fitting: An Intro to Causal Inference and do-Calculus Since writing this post back in 2018, I have extended this to a 4-part series on causal inference : ️️ Part 1: Intro to causal inference and do-calculus Part 2: Illustrating Interventions with a Toy Example Part 3: Counterfactuals Part 4: Causal Diagrams, Markov Factorization, Structural Equation Models You might have ..."}
+{"idx": 7, "title": "A Potential Outcomes Calculus for Identifying Conditional ...", "date": "", "ddg_snippet": "The do-calculus is a well-known deductive system for deriving connections between interventional and observed distributions, and has been proven complete for a number of important identifiability problems in causal inference [1, 8, 18]. ...", "subpage_snippet": "", "source": "pmc.ncbi.nlm.nih.gov", "link": "https://pmc.ncbi.nlm.nih.gov/articles/PMC6935349/", "content": "The do-calculus is a well-known deductive system for deriving connections between interventional and observed distributions, and has been proven complete for a number of important identifiability problems in causal inference [1, 8, 18]. ..."}
+{"idx": 8, "title": "Position: Probabilistic Modelling is Sufficient for Causal Inference", "date": "", "ddg_snippet": "In this paper, we demonstrated that you can do causal infer -ence through probabilistic modelling by defining a model over all settings of interest.In Probabilistic and causal inference : the works of judea pearl , pp. 507–556. 2022. Billingsley, P. Probability and Measure.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=V1FP9WDKa7", "content": "In this paper, we demonstrated that you can do causal infer -ence through probabilistic modelling by defining a model over all settings of interest.In Probabilistic and causal inference : the works of judea pearl , pp. 507–556. 2022. Billingsley, P. Probability and Measure."}
+{"idx": 9, "title": "Causal Inference and Data Fusion in Econometrics", "date": "", "ddg_snippet": "Causal Inference Engine: Three inference rules of do - calculus . Solution exists?Simultaneously controlling for confounding and selection biases introduces a new chal-lenge to the do - calculus . Not only is it necessary to transform interventional distributions into do-free expressions, but...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/1912.09104", "content": "Causal Inference Engine: Three inference rules of do - calculus . Solution exists?Simultaneously controlling for confounding and selection biases introduces a new chal-lenge to the do - calculus . Not only is it necessary to transform interventional distributions into do-free expressions, but..."}
diff --git a/data/sampled_jsons/given_two_users_with_the_same_level_of_sympathy_participate_in_similar_subreddits_interacts_with_an__year_2024.jsonl b/data/sampled_jsons/given_two_users_with_the_same_level_of_sympathy_participate_in_similar_subreddits_interacts_with_an__year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..6ac362507700c38b8649b40fb38e7ded7d52aa51
--- /dev/null
+++ b/data/sampled_jsons/given_two_users_with_the_same_level_of_sympathy_participate_in_similar_subreddits_interacts_with_an__year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Factitious disorder imposed on self - Wikipedia", "date": "", "ddg_snippet": "... on self is related to factitious disorder imposed on another , which refers to the abuse of another person in order to seek attention or sympathy for ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Factitious_disorder_imposed_on_self", "content": "... on self is related to factitious disorder imposed on another , which refers to the abuse of another person in order to seek attention or sympathy for ..."}
+{"idx": 1, "title": "Let’s have a serious discussion about the consequences of", "date": "", "ddg_snippet": "... these links to engage with the source subreddit in any manner that would violate their rules or the Reddit content policy, and do not encourage any ...", "subpage_snippet": "", "source": "www.reddit.com", "link": "https://www.reddit.com/r/REBubble/comments/w6v44f/lets_have_a_serious_discussion_about_the/", "content": "... these links to engage with the source subreddit in any manner that would violate their rules or the Reddit content policy, and do not encourage any ..."}
+{"idx": 2, "title": "With friends like these, who needs enemies? :", "date": "", "ddg_snippet": "... user summon people with the intent ... Similarly, when I m smuggling 20 kilos of cocaine into the states, I don t stop to blow up a federal building.", "subpage_snippet": "", "source": "www.reddit.com", "link": "https://www.reddit.com/r/NonPoliticalTwitter/comments/148o5r7/with_friends_like_these_who_needs_enemies/", "content": "... user summon people with the intent ... Similarly, when I m smuggling 20 kilos of cocaine into the states, I don t stop to blow up a federal building."}
+{"idx": 3, "title": "CRISIS ON INFINITE SUBREDDITS | MetaFilter", "date": "", "ddg_snippet": "In reaction, the 150K subscribers of FPH and their sympathizers in other fringe subreddits went on a rampage, creating countless clones (all banned), ...", "subpage_snippet": "", "source": "www.metafilter.com", "link": "https://www.metafilter.com/150396/CRISIS-ON-INFINITE-SUBREDDITS", "content": "In reaction, the 150K subscribers of FPH and their sympathizers in other fringe subreddits went on a rampage, creating countless clones (all banned), ..."}
+{"idx": 4, "title": "Causal Modeling of Climate Activism on Reddit", "date": "", "ddg_snippet": "14 Oct 2024 — That is, given two users with the same level of sympathy who participate in similar subreddits , if one interacts with an activist , the odds of ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.10562v1", "content": "14 Oct 2024 — That is, given two users with the same level of sympathy who participate in similar subreddits , if one interacts with an activist , the odds of ..."}
+{"idx": 5, "title": "TIFU by not knowing what a gallbladder attack was and thinking", "date": "", "ddg_snippet": "In our experience people who frequent these types of subreddits do not participate in good faith and are not a good fit for the type of atmosphere we ...", "subpage_snippet": "", "source": "www.reddit.com", "link": "https://www.reddit.com/r/tifu/comments/xryfjc/tifu_by_not_knowing_what_a_gallbladder_attack_was/", "content": "In our experience people who frequent these types of subreddits do not participate in good faith and are not a good fit for the type of atmosphere we ..."}
+{"idx": 6, "title": "Important New Rules/Guidelines Regarding Twitter Content Posted", "date": "", "ddg_snippet": "Warnings are enforced by the moderators and depend on the context and a user s history; a similar offense may thus be met with very different ...", "subpage_snippet": "", "source": "www.reddit.com", "link": "https://www.reddit.com/r/syriancivilwar/comments/55nobo/important_new_rulesguidelines_regarding_twitter/", "content": "Warnings are enforced by the moderators and depend on the context and a user s history; a similar offense may thus be met with very different ..."}
+{"idx": 7, "title": "if you know someone who thoughts of suicide, please reach out,", "date": "", "ddg_snippet": "... the right to verify or not, to remove and ban ... We reserve the right to bar entry to participants of subreddits known to harass our userbase.", "subpage_snippet": "", "source": "www.reddit.com", "link": "https://www.reddit.com/r/WhitePeopleTwitter/comments/zerrma/if_you_know_someone_who_thoughts_of_suicide/", "content": "... the right to verify or not, to remove and ban ... We reserve the right to bar entry to participants of subreddits known to harass our userbase."}
+{"idx": 8, "title": "Redditors, in defense of Reddit, destroy Reddit | MetaFilter", "date": "", "ddg_snippet": "... don't folks just switch to the official app, or the mobile web interface? Because they 're notoriously crappy , ridiculously bloated , and filled with ...", "subpage_snippet": "", "source": "www.metafilter.com", "link": "https://www.metafilter.com/199589/Redditors-in-defense-of-Reddit-destroy-Reddit", "content": "... don't folks just switch to the official app, or the mobile web interface? Because they 're notoriously crappy , ridiculously bloated , and filled with ..."}
+{"idx": 9, "title": "My Little Pony: Friendship Is Magic/Season 2/Headscratchers |", "date": "", "ddg_snippet": "This is where the Headscratchers for Season 2 go, and you can find the other seasons from this index . ... with this, they 'll either affect a more ...", "subpage_snippet": "", "source": "tropedia.fandom.com", "link": "https://tropedia.fandom.com/wiki/My_Little_Pony:_Friendship_Is_Magic/Season_2/Headscratchers", "content": "This is where the Headscratchers for Season 2 go, and you can find the other seasons from this index . ... with this, they 'll either affect a more ..."}
diff --git a/data/sampled_jsons/httpsarxiv.orgabs2003.08934.jsonl b/data/sampled_jsons/httpsarxiv.orgabs2003.08934.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..29add97d9e90ed512367c450051d275132768a3d
--- /dev/null
+++ b/data/sampled_jsons/httpsarxiv.orgabs2003.08934.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Ava Labs: Digitize All The World's Assets", "date": "", "ddg_snippet": "Ava Labs makes it simple to deploy high-performance solutions for Web3, led by innovations on Avalanche.", "subpage_snippet": "", "source": "www.avalabs.org", "link": "https://www.avalabs.org/", "content": "Ava Labs makes it simple to deploy high-performance solutions for Web3, led by innovations on Avalanche."}
+{"idx": 1, "title": "Organize Sanayi Bölgeleri Arama Motoru", "date": "", "ddg_snippet": "Org -San, Organize Sanayi Bölgeleri arama yapmayı sağlayan arama motorudur.", "subpage_snippet": "", "source": "www.org-san.org", "link": "https://www.org-san.org/", "content": "Org -San, Organize Sanayi Bölgeleri arama yapmayı sağlayan arama motorudur."}
+{"idx": 2, "title": "Presentation and Sorting (The Org Manual)", "date": "", "ddg_snippet": "Before displaying items in an agenda view, Org mode visually prepares the items and sorts them. Each item occupies a single line. The line starts with a prefix that contains the category (see...", "subpage_snippet": "", "source": "orgmode.org", "link": "https://orgmode.org/manual/Presentation-and-Sorting.html", "content": "Before displaying items in an agenda view, Org mode visually prepares the items and sorts them. Each item occupies a single line. The line starts with a prefix that contains the category (see..."}
+{"idx": 3, "title": "Critical Alerts | 511. org", "date": "", "ddg_snippet": "511 is a free phone and web service that provides Bay Area transportation information. Call 511 or visit 511. org to get information about Traffic, Transit, Carpool, Vanpool, or Bicycling.", "subpage_snippet": "", "source": "511.org", "link": "https://511.org/alerts/critical", "content": "511 is a free phone and web service that provides Bay Area transportation information. Call 511 or visit 511. org to get information about Traffic, Transit, Carpool, Vanpool, or Bicycling."}
+{"idx": 4, "title": "[ 2003 . 08934 ] NeRF: Representing Scenes as Neural Radiance Fields...", "date": "", "ddg_snippet": "We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2003.08934", "content": "We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image."}
+{"idx": 5, "title": "arxiv .uz — O’zbek tilida eng katta referatlar, slaydlar, kurs ishlari...", "date": "", "ddg_snippet": "Bu sayt O’zbekistondagi eng ko’p va eng sifatli ma’lumotlarga ega bo’lgan .referatlar, taqdimotlar, bitiruv malakaviy ishlar, kurs ishlarini yuklab olishingiz mumkin.", "subpage_snippet": "", "source": "arxiv.uz", "link": "https://arxiv.uz/uz/", "content": "Bu sayt O’zbekistondagi eng ko’p va eng sifatli ma’lumotlarga ega bo’lgan .referatlar, taqdimotlar, bitiruv malakaviy ishlar, kurs ishlarini yuklab olishingiz mumkin."}
+{"idx": 6, "title": "ABS light suddenly appeared? | EK9. org JDM EK9 Honda Civic Type...", "date": "", "ddg_snippet": "Today my car was fine drove it to my mates to underseal behind the bumpers, raise the coilovers, and strip some interior out to remove the bolted in head unit...", "subpage_snippet": "", "source": "www.ek9.org", "link": "https://www.ek9.org/index.php?threads/abs-light-suddenly-appeared.11490/", "content": "Today my car was fine drove it to my mates to underseal behind the bumpers, raise the coilovers, and strip some interior out to remove the bolted in head unit..."}
+{"idx": 7, "title": "Text to 3D Avatar Animation: A New Era in Virtual... - MarkTechPost", "date": "", "ddg_snippet": "https:// arxiv . org / abs / 2003 . 08934 .", "subpage_snippet": "", "source": "www.marktechpost.com", "link": "https://www.marktechpost.com/2024/05/06/text-to-3d-avatar-animation-a-new-era-in-virtual-character-creation/", "content": "https:// arxiv . org / abs / 2003 . 08934 ."}
+{"idx": 8, "title": "GitHub - JamunaSMurthy/2D-Video-to-3D-Synthesis: Transforming...", "date": "", "ddg_snippet": "1 MiDaS: Vision Transformers for Depth Estimation – GitHub (https://github.com/isl- org /MiDaS).", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/JamunaSMurthy/2D-Video-to-3D-Synthesis", "content": "1 MiDaS: Vision Transformers for Depth Estimation – GitHub (https://github.com/isl- org /MiDaS)."}
+{"idx": 9, "title": "nerfacc.com/_sources/methodology/sampling.rst.txt", "date": "", "ddg_snippet": "NeRF-Synthetic`: https:// arxiv . org / abs / 2003 . 08934 .", "subpage_snippet": "", "source": "www.nerfacc.com", "link": "https://www.nerfacc.com/_sources/methodology/sampling.rst.txt", "content": "NeRF-Synthetic`: https:// arxiv . org / abs / 2003 . 08934 ."}
diff --git a/data/sampled_jsons/hybrid_group_prototypes_incomplete_multi-view_clustering_intra-cluster_diversity_year_2024.jsonl b/data/sampled_jsons/hybrid_group_prototypes_incomplete_multi-view_clustering_intra-cluster_diversity_year_2024.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..4cd2014b1467e03e7a70b2c8a780f0da5f217266
--- /dev/null
+++ b/data/sampled_jsons/hybrid_group_prototypes_incomplete_multi-view_clustering_intra-cluster_diversity_year_2024.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Cluster analysis - Wikipedia", "date": "", "ddg_snippet": "The result of a cluster analysis shown as the coloring of the squares into three clusters . Cluster analysis, or clustering , is a data analysis technique aimed at partitioning a set of objects into groups such that objects within the same group exhibi...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Cluster_analysis", "content": "The result of a cluster analysis shown as the coloring of the squares into three clusters . Cluster analysis, or clustering , is a data analysis technique aimed at partitioning a set of objects into groups such that objects within the same group exhibi..."}
+{"idx": 1, "title": "Simple yet Effective Incomplete Multi-view Clustering:...", "date": "", "ddg_snippet": "Jan 22, 2025 · Simple yet Effective Incomplete Multi-view Clustering : Similarity-level Imputation and Intra -view Hybrid - group Prototype Construction Shengju Yu, Zhibin Dong, Siwei Wang, Pei Zhang, Yi Zhang, Xinwang Liu, Naiyang Guan, Tiejun Li, Yiu-ming Cheung Published: 22 Jan 2025, Last Modified: 28 Feb 2025 ICLR 2025 Spotlight Everyone Revisions BibTeX CC ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=KijslFbfOL", "content": "Jan 22, 2025 · Simple yet Effective Incomplete Multi-view Clustering : Similarity-level Imputation and Intra -view Hybrid - group Prototype Construction Shengju Yu, Zhibin Dong, Siwei Wang, Pei Zhang, Yi Zhang, Xinwang Liu, Naiyang Guan, Tiejun Li, Yiu-ming Cheung Published: 22 Jan 2025, Last Modified: 28 Feb 2025 ICLR 2025 Spotlight Everyone Revisions BibTeX CC ..."}
+{"idx": 2, "title": "[2301.11045] Incomplete Multi-view Clustering via Prototype ... Incomplete Multi-view Clustering via Prototype-based Imputation Prototype Matching Learning for Incomplete Multi-View Clustering GitHub - whbdmu/MIMB: Incomplete Multi-view Clustering Simple yet Effective Incomplete Multi-view Clustering ... Prototype Matching Learning for Incomplete Multi-View Clustering GitHub - whbdmu/MIMB: Incomplete Multi-view Clustering Incomplete Multi-view Clustering via Prototype-based Imputation Incomplete Multi-view Clustering via Prototype-based Imputation Prototype Matching Learning for Incomplete Multi-View Clustering Imputation-free and Alignment-free: Incomplete Multi-view Clustering Imputation-free and Alignment-free: Incomplete Multi-view ...", "date": "", "ddg_snippet": "Jan 26, 2023 · In this paper, we study how to achieve two characteristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance commonality refers to that within- cluster instances should share a common pattern, and ii) view versatility refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a dual attention ... Abstract In this paper, we study how to achieve two charac-teristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance common-ality refers to that within- cluster instances should share a common pattern, and ii) view versatil-ity refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a ... Jan 20, 2025 · As information acquisition diversifies, data is acquired and stored in increasing modalities. However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype -based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross ... Introduction In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization. Gather the information from other views at the similarity level to assist imputing the incomplete parts of similarity on each view. Associate a group of hybrid prototype quantities for each individual view so that it can flexibly exploit features according to the characteristics of each view. What causes incomplete multi-view clustering (imvc) problems? However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype-based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross-view prototypes are aligned. How can a manifold based incomplete multi-view clustering achieve bi-consistency guidance? In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization . What is incomplete multi-view clustering? To this end, we propose an incomplete multi-view clustering method based on a novel dual-stream model consisting of a dual attention layer and a dual con-trastive learning loss. To be specific, the dual attention layer aims to enhance the instance commonality by representing samples and prototypes with each other. Can a dual-stream model achieve incomplete multi-view clustering? In this section, we propose a dual-stream model dubbed ProImp to achieve incomplete multi-view clustering. As illus-trated in Fig. 2, ProImp is composed of a dual attention layer to model the relationship between samples and prototypes, as well as a dual contrastive learning loss to learn attention and view-specific prototypes. What is prototype matching learning for incomplete multi-view clustering? To address these issues, we propose Prototype Matching Learning for Incomplete Multi-view Clustering ( PMIMC ). Specifically, PMIMC leverages relational consistency learning to mitigate the heterogeneity of multi-view data. Subsequently, we design a robust prototype contrastive learning loss for the generated prototypes to reduce the effects of PUP. What is a consensus prototype based contrastive clustering? For consensus semantics learning, a consensus prototypes, linearly weighted by all available data, is introduced to promote consensus assignments between paired observations in a shared semantic space , called as prototype-based contrastive clustering. In incomplete multi-view clustering (IMVC), missing data introduces noise, causing view prototypes to shift and exhibit inconsistent semantics across views. A feasible solution is to explore cross-view consistency in paired complete observations for imputation and alignment. However, existing paradigm is limited to instance- or cluster -level.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2301.11045", "content": "Jan 26, 2023 · In this paper, we study how to achieve two characteristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance commonality refers to that within- cluster instances should share a common pattern, and ii) view versatility refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a dual attention ... Abstract In this paper, we study how to achieve two charac-teristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance common-ality refers to that within- cluster instances should share a common pattern, and ii) view versatil-ity refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a ... Jan 20, 2025 · As information acquisition diversifies, data is acquired and stored in increasing modalities. However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype -based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross ... Introduction In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization. Gather the information from other views at the similarity level to assist imputing the incomplete parts of similarity on each view. Associate a group of hybrid prototype quantities for each individual view so that it can flexibly exploit features according to the characteristics of each view. What causes incomplete multi-view clustering (imvc) problems? However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype-based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross-view prototypes are aligned. How can a manifold based incomplete multi-view clustering achieve bi-consistency guidance? In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization . What is incomplete multi-view clustering? To this end, we propose an incomplete multi-view clustering method based on a novel dual-stream model consisting of a dual attention layer and a dual con-trastive learning loss. To be specific, the dual attention layer aims to enhance the instance commonality by representing samples and prototypes with each other. Can a dual-stream model achieve incomplete multi-view clustering? In this section, we propose a dual-stream model dubbed ProImp to achieve incomplete multi-view clustering. As illus-trated in Fig. 2, ProImp is composed of a dual attention layer to model the relationship between samples and prototypes, as well as a dual contrastive learning loss to learn attention and view-specific prototypes. What is prototype matching learning for incomplete multi-view clustering? To address these issues, we propose Prototype Matching Learning for Incomplete Multi-view Clustering ( PMIMC ). Specifically, PMIMC leverages relational consistency learning to mitigate the heterogeneity of multi-view data. Subsequently, we design a robust prototype contrastive learning loss for the generated prototypes to reduce the effects of PUP. What is a consensus prototype based contrastive clustering? For consensus semantics learning, a consensus prototypes, linearly weighted by all available data, is introduced to promote consensus assignments between paired observations in a shared semantic space , called as prototype-based contrastive clustering. In incomplete multi-view clustering (IMVC), missing data introduces noise, causing view prototypes to shift and exhibit inconsistent semantics across views. A feasible solution is to explore cross-view consistency in paired complete observations for imputation and alignment. However, existing paradigm is limited to instance- or cluster -level."}
+{"idx": 3, "title": "Incomplete Multi-view Clustering via Prototype-based Imputation", "date": "", "ddg_snippet": "Abstract In this paper, we study how to achieve two charac-teristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance common-ality refers to that within- cluster instances should share a common pattern, and ii) view versatil-ity refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a ...", "subpage_snippet": "", "source": "www.ijcai.org", "link": "https://www.ijcai.org/proceedings/2023/0435.pdf", "content": "Abstract In this paper, we study how to achieve two charac-teristics highly-expected by incomplete multi-view clustering (IMvC). Namely, i) instance common-ality refers to that within- cluster instances should share a common pattern, and ii) view versatil-ity refers to that cross-view samples should own view-specific patterns. To this end, we design a novel dual-stream model which employs a ..."}
+{"idx": 4, "title": "Prototype Matching Learning for Incomplete Multi-View Clustering", "date": "", "ddg_snippet": "Jan 20, 2025 · As information acquisition diversifies, data is acquired and stored in increasing modalities. However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype -based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross ...", "subpage_snippet": "", "source": "ieeexplore.ieee.org", "link": "https://ieeexplore.ieee.org/document/10847794", "content": "Jan 20, 2025 · As information acquisition diversifies, data is acquired and stored in increasing modalities. However, sensor failures or equipment issues can lead to partial data loss in certain views, resulting in incomplete multi-view clustering (IMVC) problems. Although some prototype -based IMVC methods have achieved satisfactory performance, almost all of these methods implicitly assume that the cross ..."}
+{"idx": 5, "title": "GitHub - whbdmu/MIMB: Incomplete Multi-view Clustering", "date": "", "ddg_snippet": "Introduction In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/whbdmu/MIMB", "content": "Introduction In this paper, we have proposed a novel method termed Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance (MIMB), which flexibly recover the incomplete data among various views and intends to achieve the bi-consistency guidance via reverse regularization."}
+{"idx": 6, "title": "Simple yet Effective Incomplete Multi-view Clustering ...", "date": "", "ddg_snippet": "Gather the information from other views at the similarity level to assist imputing the incomplete parts of similarity on each view. Associate a group of hybrid prototype quantities for each individual view so that it can flexibly exploit features according to the characteristics of each view.", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/media/iclr-2025/Slides/30038.pdf", "content": "Gather the information from other views at the similarity level to assist imputing the incomplete parts of similarity on each view. Associate a group of hybrid prototype quantities for each individual view so that it can flexibly exploit features according to the characteristics of each view."}
+{"idx": 7, "title": "Imputation-free and Alignment-free: Incomplete Multi-view ...", "date": "", "ddg_snippet": "In incomplete multi-view clustering (IMVC), missing data introduces noise, causing view prototypes to shift and exhibit inconsistent semantics across views. A feasible solution is to explore cross-view consistency in paired complete observations for imputation and alignment. However, existing paradigm is limited to instance- or cluster -level.", "subpage_snippet": "", "source": "cvpr.thecvf.com", "link": "https://cvpr.thecvf.com/virtual/2025/poster/33279", "content": "In incomplete multi-view clustering (IMVC), missing data introduces noise, causing view prototypes to shift and exhibit inconsistent semantics across views. A feasible solution is to explore cross-view consistency in paired complete observations for imputation and alignment. However, existing paradigm is limited to instance- or cluster -level."}
+{"idx": 8, "title": "ICLR Poster Simple yet Effective Incomplete Multi - view Clustering ...", "date": "", "ddg_snippet": "Diversity & Inclusion.Abstract: Most of incomplete multi - view clustering (IMVC) methods typically choose to ignore the missing samples and only utilize observed unpaired samples to construct bipartite similarity.", "subpage_snippet": "", "source": "iclr.cc", "link": "https://iclr.cc/virtual/2025/poster/30038", "content": "Diversity & Inclusion.Abstract: Most of incomplete multi - view clustering (IMVC) methods typically choose to ignore the missing samples and only utilize observed unpaired samples to construct bipartite similarity."}
+{"idx": 9, "title": "M Ulti - view", "date": "", "ddg_snippet": "Intra -View Hybrid - Group Prototypes . Incomplete multi - view clustering network via nonlinear manifold embedding and probability-induced loss.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=KijslFbfOL", "content": "Intra -View Hybrid - Group Prototypes . Incomplete multi - view clustering network via nonlinear manifold embedding and probability-induced loss."}
diff --git a/data/sampled_jsons/neural_operator_multi-time-step_PDE_solver.jsonl b/data/sampled_jsons/neural_operator_multi-time-step_PDE_solver.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ff9c13f4ac177f1b2c5ecf8f09f92f1527499f1a
--- /dev/null
+++ b/data/sampled_jsons/neural_operator_multi-time-step_PDE_solver.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2507.17763] Multi-Head Neural Operator for Modelling ...", "date": "", "ddg_snippet": "by MS Eshaghi · 2025 · Cited by 1 — This design allows MHNO to predict all time steps after a single forward pass, while effectively capturing long-term dependencies and avoiding ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2507.17763", "content": "by MS Eshaghi · 2025 · Cited by 1 — This design allows MHNO to predict all time steps after a single forward pass, while effectively capturing long-term dependencies and avoiding ..."}
+{"idx": 1, "title": "Multi-scale time-stepping of Partial Differential Equations ...", "date": "", "ddg_snippet": "by AP Hemmasian · 2024 · Cited by 16 — In order to assess the capability of our model, we train it on two challenging tasks used to evaluate powerful and novel PDE solvers like neural operators .", "subpage_snippet": "", "source": "www.sciencedirect.com", "link": "https://www.sciencedirect.com/science/article/pii/S0045782524002391", "content": "by AP Hemmasian · 2024 · Cited by 16 — In order to assess the capability of our model, we train it on two challenging tasks used to evaluate powerful and novel PDE solvers like neural operators ."}
+{"idx": 2, "title": "PDE-constrained Learning with Multi-time-stepping for ...", "date": "", "ddg_snippet": "by Q Wang — This paper introduces MultiPDENet , a neural network architecture designed to accelerate fluid dynamic simulations by combining classical numerical methods, such ...", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=stcN89QGfL", "content": "by Q Wang — This paper introduces MultiPDENet , a neural network architecture designed to accelerate fluid dynamic simulations by combining classical numerical methods, such ..."}
+{"idx": 3, "title": "MultiPDENet: PDE-embedded Learning with Multi-time ...", "date": "", "ddg_snippet": "27 Jan 2025 — To accurately predict at the micro-scale step, we developed a neural solver , referred to as the Physics Block, as illustrated in Figure 1(b).", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2501.15987v1", "content": "27 Jan 2025 — To accurately predict at the micro-scale step, we developed a neural solver , referred to as the Physics Block, as illustrated in Figure 1(b)."}
+{"idx": 4, "title": "PDE-Refiner: Achieving Accurate Long Rollouts with Neural ...", "date": "", "ddg_snippet": "A novel model class that enables more accurate modeling of all frequency components via a multi - step refinement process.", "subpage_snippet": "", "source": "phlippe.github.io", "link": "https://phlippe.github.io/PDERefiner/", "content": "A novel model class that enables more accurate modeling of all frequency components via a multi - step refinement process."}
+{"idx": 5, "title": "MultiPDENet: PDE-embedded Learning with Multi-time ...", "date": "", "ddg_snippet": "15 Jul 2025 — To this end, we propose a PDEembedded network with multiscale time stepping ( MultiPDENet ), which fuses the scheme of numerical methods and ...", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/46029", "content": "15 Jul 2025 — To this end, we propose a PDEembedded network with multiscale time stepping ( MultiPDENet ), which fuses the scheme of numerical methods and ..."}
+{"idx": 6, "title": "HMgNO: Hybrid multigrid neural operator with low-order ...", "date": "", "ddg_snippet": "by Y Hu · 2025 — The HMgNO couples a low-order numerical solver with a multigrid neural operator , and the neural operator is used to correct the low-order numerical solutions.", "subpage_snippet": "", "source": "www.sciencedirect.com", "link": "https://www.sciencedirect.com/science/article/pii/S0893608025005295", "content": "by Y Hu · 2025 — The HMgNO couples a low-order numerical solver with a multigrid neural operator , and the neural operator is used to correct the low-order numerical solutions."}
+{"idx": 7, "title": "Multi-resolution partial differential equations preserved ...", "date": "", "ddg_snippet": "by XY Liu · 2024 · Cited by 65 — The neural solver is formulated as a next- step DNN model by learning the dynamic transitions from the current step t to the next time step t + ...", "subpage_snippet": "", "source": "www.nature.com", "link": "https://www.nature.com/articles/s42005-024-01521-z", "content": "by XY Liu · 2024 · Cited by 65 — The neural solver is formulated as a next- step DNN model by learning the dynamic transitions from the current step t to the next time step t + ..."}
+{"idx": 8, "title": "Newton Informed Neural Operator for Solving Nonlinear ...", "date": "", "ddg_snippet": "by W Hao · 2024 · Cited by 5 — Overall, the Newton Informed. Neural Operator efficiently solves nonlinear PDEs with multiple solutions by learning the Newton nonlinear solver. It addresses ...", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/dae8afc6b990aa0b3b5efaa096fbd7fa-Paper-Conference.pdf", "content": "by W Hao · 2024 · Cited by 5 — Overall, the Newton Informed. Neural Operator efficiently solves nonlinear PDEs with multiple solutions by learning the Newton nonlinear solver. It addresses ..."}
+{"idx": 9, "title": "Neural Operator - Zongyi Li", "date": "", "ddg_snippet": "In this experiment, we use neural operators to learn the operator mapping from the vorticity of the first time 10 time steps to that up to a later time step .", "subpage_snippet": "", "source": "zongyi-li.github.io", "link": "https://zongyi-li.github.io/neural-operator/", "content": "In this experiment, we use neural operators to learn the operator mapping from the vorticity of the first time 10 time steps to that up to a later time step ."}
diff --git a/data/sampled_jsons/neural_scaling_law_error_rate_exponent_pre-training_Kaplan_Hoffmann.jsonl b/data/sampled_jsons/neural_scaling_law_error_rate_exponent_pre-training_Kaplan_Hoffmann.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..fbe7181bb50e39019a197832389b801becbaecc4
--- /dev/null
+++ b/data/sampled_jsons/neural_scaling_law_error_rate_exponent_pre-training_Kaplan_Hoffmann.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Neural scaling law - Wikipedia", "date": "", "ddg_snippet": "Neural scaling law Performance of AI models on various benchmarks from 1998 to 2024 In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, [1][2] and training ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Neural_scaling_law", "content": "Neural scaling law Performance of AI models on various benchmarks from 1998 to 2024 In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, [1][2] and training ..."}
+{"idx": 1, "title": "Explaining neural scaling laws - PNAS", "date": "", "ddg_snippet": "Jun 24, 2024 · The population loss of trained deep neural networks often follows precise power- law scaling relations with either the size of the training dataset or the number of parameters in the network. We pro...", "subpage_snippet": "", "source": "www.pnas.org", "link": "https://www.pnas.org/doi/10.1073/pnas.2311878121", "content": "Jun 24, 2024 · The population loss of trained deep neural networks often follows precise power- law scaling relations with either the size of the training dataset or the number of parameters in the network. We pro..."}
+{"idx": 2, "title": "[2001.08361] Scaling Laws for Neural Language Models - arXiv.org Neural scaling law - Wikipedia Scaling Laws for Neural Language Models - papers.baulab.info Scaling Laws for Neural Language Models - Semantic Scholar Scaling Laws and Compute-Optimal Training Beyond Fixed ... Neural scaling law - Wikipedia Explaining neural scaling laws - PNAS Scaling Laws of Neural Language Models - GitHub Scaling Laws for Neural Language Models - papers.baulab.info Scaling Laws for Neural Language Models - papers.baulab.info Scaling Laws for Neural Language Models - papers.baulab.info Scaling Laws of Neural Language Models - GitHub", "date": "", "ddg_snippet": "Jan 23, 2020 · We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern the ... Neural scaling law Performance of AI models on various benchmarks from 1998 to 2024 In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, [1][2] and training ... Abstract We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern ... Larger models are significantly more sample-efficient, such that optimally compute-efficient training involves training very large models on a relatively modest amount of data and stopping significantly before convergence. We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of ... Scaling Law Experiments for Neural Language Models. Kaplan et al. (2020) were the first to establish scaling laws for language models by training a suite of models for a fixed token count. What is a neural scaling law? A neural scaling law is a theoretical or empirical statistical law between these parameters . There are also other parameters with other scaling laws. In most cases, the model's size is simply the number of parameters. However, one complication arises with the use of sparse models, such as mixture-of-expert models. What are scaling laws in deep neural networks? There have been a number of recent works demonstrating empirical scaling laws (1 – 5) in deep neural networks, including scaling laws with model size, dataset size, compute, and other observables such as mutual information and pruning. Some precursors (6, 7) can be found in earlier literature. How does the Test loss of a language model scale as a power law? Kaplan et al discovered that the test loss of a language model scales as a power law with the number of (non-embedding) model parameters N , the number of tokens D in the training dataset, and the amount of compute C used to train a model: L ( X ) = ( X c X ) α X where X = N , D , or C . Do power-law exponents have a rough interpretation? Power-law scalings with model and dataset size in density estimation [Was06] and in random forest models [Bia12] may be connected with our results. These models suggest that power-law exponents may have a very rough interpretation as the inverse of the number of relevant features in the data. Does the optimally trained Test loss match the scaling law? Here we will study the performance of a model of size N trained on a dataset with D tokens while varying N and D simultaneously. We will empirically demonstrate that the optimally trained test loss accords with the scaling law of Equation (1.5). Is there a power-law scaling between data size and performance? Some early [BB01, Goo01] work found power-law scalings between performance and dataset size . More recent work [HNA+17, HAD19] also investigated scaling between model size and data size; their work is perhaps the closest to ours in the literature8. Scaling Laws of Neural Language Models This repository contains an implementation of scaling laws as first found by Kaplan et al in Scaling Laws of Neural Language Models. We find how the test loss of a language model scales with parameter count, dataset size, total compute, and the number of training steps.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2001.08361", "content": "Jan 23, 2020 · We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern the ... Neural scaling law Performance of AI models on various benchmarks from 1998 to 2024 In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, [1][2] and training ... Abstract We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern ... Larger models are significantly more sample-efficient, such that optimally compute-efficient training involves training very large models on a relatively modest amount of data and stopping significantly before convergence. We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of ... Scaling Law Experiments for Neural Language Models. Kaplan et al. (2020) were the first to establish scaling laws for language models by training a suite of models for a fixed token count. What is a neural scaling law? A neural scaling law is a theoretical or empirical statistical law between these parameters . There are also other parameters with other scaling laws. In most cases, the model's size is simply the number of parameters. However, one complication arises with the use of sparse models, such as mixture-of-expert models. What are scaling laws in deep neural networks? There have been a number of recent works demonstrating empirical scaling laws (1 – 5) in deep neural networks, including scaling laws with model size, dataset size, compute, and other observables such as mutual information and pruning. Some precursors (6, 7) can be found in earlier literature. How does the Test loss of a language model scale as a power law? Kaplan et al discovered that the test loss of a language model scales as a power law with the number of (non-embedding) model parameters N , the number of tokens D in the training dataset, and the amount of compute C used to train a model: L ( X ) = ( X c X ) α X where X = N , D , or C . Do power-law exponents have a rough interpretation? Power-law scalings with model and dataset size in density estimation [Was06] and in random forest models [Bia12] may be connected with our results. These models suggest that power-law exponents may have a very rough interpretation as the inverse of the number of relevant features in the data. Does the optimally trained Test loss match the scaling law? Here we will study the performance of a model of size N trained on a dataset with D tokens while varying N and D simultaneously. We will empirically demonstrate that the optimally trained test loss accords with the scaling law of Equation (1.5). Is there a power-law scaling between data size and performance? Some early [BB01, Goo01] work found power-law scalings between performance and dataset size . More recent work [HNA+17, HAD19] also investigated scaling between model size and data size; their work is perhaps the closest to ours in the literature8. Scaling Laws of Neural Language Models This repository contains an implementation of scaling laws as first found by Kaplan et al in Scaling Laws of Neural Language Models. We find how the test loss of a language model scales with parameter count, dataset size, total compute, and the number of training steps."}
+{"idx": 3, "title": "Scaling Laws for Neural Language Models - papers.baulab.info", "date": "", "ddg_snippet": "Abstract We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern ...", "subpage_snippet": "", "source": "papers.baulab.info", "link": "https://papers.baulab.info/papers/Kaplan-2020.pdf", "content": "Abstract We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of compute used for training , with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern ..."}
+{"idx": 4, "title": "Scaling Laws for Neural Language Models - Semantic Scholar", "date": "", "ddg_snippet": "Larger models are significantly more sample-efficient, such that optimally compute-efficient training involves training very large models on a relatively modest amount of data and stopping significantly before convergence. We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of ...", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Scaling-Laws-for-Neural-Language-Models-Kaplan-McCandlish/e6c561d02500b2596a230b341a8eb8b921ca5bf2", "content": "Larger models are significantly more sample-efficient, such that optimally compute-efficient training involves training very large models on a relatively modest amount of data and stopping significantly before convergence. We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power- law with model size, dataset size, and the amount of ..."}
+{"idx": 5, "title": "Scaling Laws and Compute-Optimal Training Beyond Fixed ...", "date": "", "ddg_snippet": "Scaling Law Experiments for Neural Language Models. Kaplan et al. (2020) were the first to establish scaling laws for language models by training a suite of models for a fixed token count.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/pdf?id=ompl7supoX", "content": "Scaling Law Experiments for Neural Language Models. Kaplan et al. (2020) were the first to establish scaling laws for language models by training a suite of models for a fixed token count."}
+{"idx": 6, "title": "Scaling Laws of Neural Language Models - GitHub", "date": "", "ddg_snippet": "Scaling Laws of Neural Language Models This repository contains an implementation of scaling laws as first found by Kaplan et al in Scaling Laws of Neural Language Models. We find how the test loss of a language model scales with parameter count, dataset size, total compute, and the number of training steps.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/shehper/scaling_laws", "content": "Scaling Laws of Neural Language Models This repository contains an implementation of scaling laws as first found by Kaplan et al in Scaling Laws of Neural Language Models. We find how the test loss of a language model scales with parameter count, dataset size, total compute, and the number of training steps."}
+{"idx": 7, "title": "Reconciling Kaplan and Chinchilla Scaling Laws", "date": "", "ddg_snippet": "Both informed large- scale efforts on how to trade off model parameters ( N 𝑁 N italic_N ) and training tokens ( D 𝐷 D italic_D ) for a given ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2406.12907v3", "content": "Both informed large- scale efforts on how to trade off model parameters ( N 𝑁 N italic_N ) and training tokens ( D 𝐷 D italic_D ) for a given ..."}
+{"idx": 8, "title": "How Does Critical Batch Size Scale in Pre-training?", "date": "", "ddg_snippet": "Efficient optimization is critical in pre - training large models (LMs) at scale (McCandlish et al., 2018 ; Shoeybi et al., 2019 ; Kaplan et al ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.21676v4", "content": "Efficient optimization is critical in pre - training large models (LMs) at scale (McCandlish et al., 2018 ; Shoeybi et al., 2019 ; Kaplan et al ..."}
+{"idx": 9, "title": "Scaling Laws of Synthetic Data for Language Models", "date": "", "ddg_snippet": "This prompts a natural question: Are there scaling laws for synthetic data? While scaling laws for pre - training data are well-documented [ 20 , 18 ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.19551v2", "content": "This prompts a natural question: Are there scaling laws for synthetic data? While scaling laws for pre - training data are well-documented [ 20 , 18 ..."}
diff --git a/data/sampled_jsons/peeling_technique_concentration_inequalities_bandit_variance_estimation.jsonl b/data/sampled_jsons/peeling_technique_concentration_inequalities_bandit_variance_estimation.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b6819318857c57ffe1998d5908708bf2ea8efcc6
--- /dev/null
+++ b/data/sampled_jsons/peeling_technique_concentration_inequalities_bandit_variance_estimation.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "PDF Empirical Process: Peeling Technique - Stanford University", "date": "", "ddg_snippet": "The two main technique used in the paper is the peeling lemma and the Talagrand Concentration Inequality. In this section, we have a slightly simpler version of the proof instead of the original one in the paper.", "subpage_snippet": "", "source": "web.stanford.edu", "link": "https://web.stanford.edu/~yplu/note/localization.pdf", "content": "The two main technique used in the paper is the peeling lemma and the Talagrand Concentration Inequality. In this section, we have a slightly simpler version of the proof instead of the original one in the paper."}
+{"idx": 1, "title": "PDF Lecture 3: Concentration Inequalities and Mean Estimation", "date": "", "ddg_snippet": "3 Concentration Inequalities for Mean Estimation First, we'll derive two concentration inequalities { Hoe ding's inequality and Bernstein's inequality { and use them to evaluate the sample complexity of the sample mean when estimating the true mean of a collection of i.i.d. random variables.", "subpage_snippet": "", "source": "cs.brown.edu", "link": "https://cs.brown.edu/courses/csci1951-w/lec/lec+3+notes.pdf", "content": "3 Concentration Inequalities for Mean Estimation First, we'll derive two concentration inequalities { Hoe ding's inequality and Bernstein's inequality { and use them to evaluate the sample complexity of the sample mean when estimating the true mean of a collection of i.i.d. random variables."}
+{"idx": 2, "title": "\"Peeling Technique\" in Probability - Mathematics Stack Exchange", "date": "", "ddg_snippet": "Never heard of the \" Peeling Argument\", but the first inequality just seems to be the standard inequality $\\mathbb P (A\\cup B)\\leq \\mathbb P (A)+\\mathbb P (B)$ for (probability) measures. And the second inequality should come from a comparison of the summands and their respective events (monotonicity of (probability) measure).", "subpage_snippet": "", "source": "math.stackexchange.com", "link": "https://math.stackexchange.com/questions/4947385/peeling-technique-in-probability", "content": "Never heard of the \" Peeling Argument\", but the first inequality just seems to be the standard inequality $\\mathbb P (A\\cup B)\\leq \\mathbb P (A)+\\mathbb P (B)$ for (probability) measures. And the second inequality should come from a comparison of the summands and their respective events (monotonicity of (probability) measure)."}
+{"idx": 3, "title": "PDF Concentration Inequalities and Multi-Armed Bandits", "date": "", "ddg_snippet": "Can we directly apply Hoeffding's inequality here with n as the number of coin tosses? If you want to derive a concentration bound for this problem, look up Azuma's inequality.", "subpage_snippet": "", "source": "nanjiang.web.engr.illinois.edu", "link": "https://nanjiang.web.engr.illinois.edu/files/cs598/note_bandit.pdf", "content": "Can we directly apply Hoeffding's inequality here with n as the number of coin tosses? If you want to derive a concentration bound for this problem, look up Azuma's inequality."}
+{"idx": 4, "title": "Variance-Adaptive Algorithm for Probabilistic ... - NSF Public Access", "date": "", "ddg_snippet": "For each arm's over- estimation , we use the peeling technique to handle the observation probability and carefully design a series of events to derive the final regrets. We believe our proof techniques are tight and novel, which may be of independent interest to improve other works that share the similar feedback model or over- estimation terms.", "subpage_snippet": "", "source": "par.nsf.gov", "link": "https://par.nsf.gov/servlets/purl/10463406", "content": "For each arm's over- estimation , we use the peeling technique to handle the observation probability and carefully design a series of events to derive the final regrets. We believe our proof techniques are tight and novel, which may be of independent interest to improve other works that share the similar feedback model or over- estimation terms."}
+{"idx": 5, "title": "PDF Lecture 7: Martingales and Concentration - University of Cambridge", "date": "", "ddg_snippet": "Xi indicates if the i-th edge belongs to a random graph G, and f(X1; : : : ; Xm) represent the number of connected components of G We can simply prove concentration of X around it means by the so-called Method of Bounded Differences", "subpage_snippet": "", "source": "www.cl.cam.ac.uk", "link": "https://www.cl.cam.ac.uk/teaching/1819/Probablty/materials/Lecture7.pdf", "content": "Xi indicates if the i-th edge belongs to a random graph G, and f(X1; : : : ; Xm) represent the number of connected components of G We can simply prove concentration of X around it means by the so-called Method of Bounded Differences"}
+{"idx": 6, "title": "PDF Bandit value estimation as an excuse to get some new concentration ...", "date": "", "ddg_snippet": "Bandit value estimation as an excuse to get some new concentration inequalities Csaba Szepesv ́ari November 30, 2021 DeepMind and University of Alberta BIRS Workshop on Math.Stat & Learning", "subpage_snippet": "", "source": "stats.birs.ca", "link": "https://stats.birs.ca/workshops/2021/21w5070/files/Csaba+Szepesvari/contextual_mini_talk+BIRS.pdf", "content": "Bandit value estimation as an excuse to get some new concentration inequalities Csaba Szepesv ́ari November 30, 2021 DeepMind and University of Alberta BIRS Workshop on Math.Stat & Learning"}
+{"idx": 7, "title": "PDF Basics of Concentration Inequalities - Stanford University", "date": "", "ddg_snippet": "Tensorization identities I variance inequality familiar: if Xi are independent, Var", "subpage_snippet": "", "source": "web.stanford.edu", "link": "https://web.stanford.edu/class/stats300b/Slides/06-concentration.pdf", "content": "Tensorization identities I variance inequality familiar: if Xi are independent, Var"}
+{"idx": 8, "title": "FraPPE: Fast and Efficient Preference-based Pure Exploration", "date": "", "ddg_snippet": "Together, these techniques solve the maxmin optimisation problem in 𝒪 ( K L 2 ) \\mathcal{O}(KL^{2}) time for a bandit instance with K K ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2508.16487v1", "content": "Together, these techniques solve the maxmin optimisation problem in 𝒪 ( K L 2 ) \\mathcal{O}(KL^{2}) time for a bandit instance with K K ..."}
+{"idx": 9, "title": "WC 2021", "date": "", "ddg_snippet": "We show that at a fixed interior point in the design space, the estimation error of the max-min block estimator converges in distribution to a non ...", "subpage_snippet": "", "source": "duetone.org", "link": "https://duetone.org/wc21/day/3", "content": "We show that at a fixed interior point in the design space, the estimation error of the max-min block estimator converges in distribution to a non ..."}
diff --git a/data/sampled_jsons/permutation_pattern_3412_definition.jsonl b/data/sampled_jsons/permutation_pattern_3412_definition.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..e1b9cb91e07aa84fc0ae3bcd4760c04d211e9572
--- /dev/null
+++ b/data/sampled_jsons/permutation_pattern_3412_definition.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Skew-merged permutation - Wikipedia", "date": "", "ddg_snippet": "The two smallest permutations that cannot be partitioned into an increasing and a decreasing sequence are 3412 and 2143. Stankova (1994) was the first to establish that a skew-merged permutation can also be equivalently defined as a permutation that avoids the two patterns 3412 and 2143. A permutation is skew-merged if and only if its associated permutation graph is a split graph, a graph that ...", "subpage_snippet": "", "source": "en.wikipedia.org", "link": "https://en.wikipedia.org/wiki/Skew-merged_permutation", "content": "The two smallest permutations that cannot be partitioned into an increasing and a decreasing sequence are 3412 and 2143. Stankova (1994) was the first to establish that a skew-merged permutation can also be equivalently defined as a permutation that avoids the two patterns 3412 and 2143. A permutation is skew-merged if and only if its associated permutation graph is a split graph, a graph that ..."}
+{"idx": 1, "title": "Counting the Nontrivial Equivalence Classes of Under 1234 3412 -Pattern ...", "date": "", "ddg_snippet": "Recall that we define two permutations α, β ∈ Sn to be equivalent under the {1234, 3412}-equivalence if α can be reached from β by performing a series of 1234 → 3412 and 3412 → 1234 pattern -replacements. A 1234 → 3412 pattern -replacement in a permutation π simply takes an increasing 4-letter subsequence πi1, πi2, πi3, πi3 (where i1 < i2 < i3 < i4), and places each of πi3 ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2008.02380", "content": "Recall that we define two permutations α, β ∈ Sn to be equivalent under the {1234, 3412}-equivalence if α can be reached from β by performing a series of 1234 → 3412 and 3412 → 1234 pattern -replacements. A 1234 → 3412 pattern -replacement in a permutation π simply takes an increasing 4-letter subsequence πi1, πi2, πi3, πi3 (where i1 < i2 < i3 < i4), and places each of πi3 ..."}
+{"idx": 2, "title": "Permutation patterns - SymCat", "date": "", "ddg_snippet": "Permutation patterns A permutation π ∈ S n is said to contain the pattern σ ∈ S k, if there is a subsequence of π which is order-isomorphic to σ For example, π = [1, 8, 5, 3, 7, 4, 6, 2] contains the pattern [2, 3, 1] as the subsequence 3, 6, 2 in π have its elements in the same relative order.", "subpage_snippet": "", "source": "www.symmetricfunctions.com", "link": "https://www.symmetricfunctions.com/permutationPatterns.htm", "content": "Permutation patterns A permutation π ∈ S n is said to contain the pattern σ ∈ S k, if there is a subsequence of π which is order-isomorphic to σ For example, π = [1, 8, 5, 3, 7, 4, 6, 2] contains the pattern [2, 3, 1] as the subsequence 3, 6, 2 in π have its elements in the same relative order."}
+{"idx": 3, "title": "PDF 3412, 4231 patterns produce singular points of essential sets", "date": "", "ddg_snippet": "Abstract This article introduces new concepts dual bigrassmannian permutations and dual essential sets with the discussion of singular patterns ( 3412 , 4231). We show that (dual) bigrassmannian permutations may contain only 3412 (4231). Then we define singularity of points of essential sets by singularity of certain corresponding (dual) bigrassmannian permutations (this discussion depends on ...", "subpage_snippet": "", "source": "www.rimath.saitama-u.ac.jp", "link": "https://www.rimath.saitama-u.ac.jp/research/pdf/smj28-2.pdf", "content": "Abstract This article introduces new concepts dual bigrassmannian permutations and dual essential sets with the discussion of singular patterns ( 3412 , 4231). We show that (dual) bigrassmannian permutations may contain only 3412 (4231). Then we define singularity of points of essential sets by singularity of certain corresponding (dual) bigrassmannian permutations (this discussion depends on ..."}
+{"idx": 4, "title": "PDF Counting the Nontrivial Equivalence Classes of Under 1234 3412 -Pattern ...", "date": "", "ddg_snippet": "We study the {1234, 3412 }- pattern -replacement equivalence relation on the set Sn of permutations of length n, which is conceptually similar to the Knuth relation. In par-ticular, we enumerate and characterize the nontrivial equivalence classes, or equivalence classes with size greater than 1, in Sn for n ≥ 7 under the {1234, 3412}-equivalence. This proves a conjecture by Ma, who found three ...", "subpage_snippet": "", "source": "cs.uwaterloo.ca", "link": "https://cs.uwaterloo.ca/journals/JIS/VOL23/Zhang/zhang6.pdf", "content": "We study the {1234, 3412 }- pattern -replacement equivalence relation on the set Sn of permutations of length n, which is conceptually similar to the Knuth relation. In par-ticular, we enumerate and characterize the nontrivial equivalence classes, or equivalence classes with size greater than 1, in Sn for n ≥ 7 under the {1234, 3412}-equivalence. This proves a conjecture by Ma, who found three ..."}
+{"idx": 5, "title": "Permutation Pattern -- from Wolfram MathWorld", "date": "", "ddg_snippet": "Let F(n,sigma) denote the number of permutations on the symmetric group S_n which avoid sigma in S_k as a subpattern, where \"tau contains sigma as a subpattern\" is interpreted to mean that there exist 1<=x_1<=x_2<=...<=x_k<=n such that for 1<=i,j<=k, tau(x_i)<tau(x_j) iff sigma(i)<sigma(j). For example, a permutation contains the pattern (123) iff it has an ascending subsequence of length ...", "subpage_snippet": "", "source": "mathworld.wolfram.com", "link": "https://mathworld.wolfram.com/PermutationPattern.html", "content": "Let F(n,sigma) denote the number of permutations on the symmetric group S_n which avoid sigma in S_k as a subpattern, where \"tau contains sigma as a subpattern\" is interpreted to mean that there exist 1<=x_1<=x_2<=...<=x_k<=n such that for 1<=i,j<=k, tau(x_i)<tau(x_j) iff sigma(i)<sigma(j). For example, a permutation contains the pattern (123) iff it has an ascending subsequence of length ..."}
+{"idx": 6, "title": "PDF Permutation patterns: basic definitions and notati", "date": "", "ddg_snippet": "A permutation , or pattern , p is said to be contained in, or to be a subpermutation of, another permutation s, written p 6 s or p 4 s, if s has a (not necessarily contiguous) subsequence whose terms are order isomorphic to (i.e. have the same relative ordering as) p. From the graphical perspective, s contains p if the plot of p results from erasing zero or more points from the plot of s and ...", "subpage_snippet": "", "source": "2018.permutationpatterns.com", "link": "https://2018.permutationpatterns.com/docs/basic-definitions.pdf", "content": "A permutation , or pattern , p is said to be contained in, or to be a subpermutation of, another permutation s, written p 6 s or p 4 s, if s has a (not necessarily contiguous) subsequence whose terms are order isomorphic to (i.e. have the same relative ordering as) p. From the graphical perspective, s contains p if the plot of p results from erasing zero or more points from the plot of s and ..."}
+{"idx": 7, "title": "ds.algorithms - More questions about permutations - Theoretical ...", "date": "", "ddg_snippet": "I'm interested in permutations classes defined by one or several excluded patterns of length 4. In particular, the following classes seem interesting but not so-well studied algorithmically. the skew-merged permutations consist of the permutations that can be partitioned in an increasing and a decreasing subsequence; their basis is $\\ { 3412 , 2143\\}$. the vexillary permutations have basis ...", "subpage_snippet": "", "source": "cstheory.stackexchange.com", "link": "https://cstheory.stackexchange.com/questions/22306/more-questions-about-permutations", "content": "I'm interested in permutations classes defined by one or several excluded patterns of length 4. In particular, the following classes seem interesting but not so-well studied algorithmically. the skew-merged permutations consist of the permutations that can be partitioned in an increasing and a decreasing subsequence; their basis is $\\ { 3412 , 2143\\}$. the vexillary permutations have basis ..."}
+{"idx": 8, "title": "Permutation patterns: basic definitions and notati - arXiv.org", "date": "", "ddg_snippet": "Classical permutation classes The subpermutation relation is a partial order on the set of all permutations . A classical per-mutation class, sometimes called a pattern class, is a set of permutations closed downwards (a down-set) under this partial order. Thus, if σ is a member of a permutation class C and τ is contained in σ, then it must be the case that τ is also a member of C. From a ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/1506.06673", "content": "Classical permutation classes The subpermutation relation is a partial order on the set of all permutations . A classical per-mutation class, sometimes called a pattern class, is a set of permutations closed downwards (a down-set) under this partial order. Thus, if σ is a member of a permutation class C and τ is contained in σ, then it must be the case that τ is also a member of C. From a ..."}
+{"idx": 9, "title": "PermPal Database", "date": "", "ddg_snippet": "The Permutation Pattern Avoidance Library (PermPAL) PermPAL is a database of algorithmically-derived theorems about permutation classes. The Combinatorial Exploration framework produces rigorously verified combinatorial specifications for families of combinatorial objects.", "subpage_snippet": "", "source": "www.permpal.com", "link": "https://www.permpal.com/", "content": "The Permutation Pattern Avoidance Library (PermPAL) PermPAL is a database of algorithmically-derived theorems about permutation classes. The Combinatorial Exploration framework produces rigorously verified combinatorial specifications for families of combinatorial objects."}
diff --git a/data/sampled_jsons/qtuxDy2qEB_Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models.jsonl b/data/sampled_jsons/qtuxDy2qEB_Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..bae682836bf0fa926424965dce99204096b351bf
--- /dev/null
+++ b/data/sampled_jsons/qtuxDy2qEB_Parallel_Simulation_for_Log-concave_Sampling_and_Score-based_Diffusion_Models.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Parallel Simulation for Log - concave Sampling and Score - based ...", "date": "", "ddg_snippet": "TL;DR: We propose new parallel algorithms for log - concave sampling and diffusion models , and rigorously prove that our algorithms enjoy $\\log d$ iteration complexity.Keywords: parallel sampling , log - concave sampling , diffusion model . Submission Number: 9486.", "subpage_snippet": "", "source": "openreview.net", "link": "https://openreview.net/forum?id=qtuxDy2qEB&referrer=[the+profile+of+Masashi+Sugiyama](/profile?id=~Masashi_Sugiyama1)", "content": "TL;DR: We propose new parallel algorithms for log - concave sampling and diffusion models , and rigorously prove that our algorithms enjoy $\\log d$ iteration complexity.Keywords: parallel sampling , log - concave sampling , diffusion model . Submission Number: 9486."}
+{"idx": 1, "title": "ICML Poster Parallel Simulation for Sampling under Isoperimetry...", "date": "", "ddg_snippet": "d)$, which is even optimal for log - concave sampling with some specific adaptive complexity. Our approach builds on parallel simulation techniques from scientific computing.", "subpage_snippet": "", "source": "icml.cc", "link": "https://icml.cc/virtual/2025/poster/43916", "content": "d)$, which is even optimal for log - concave sampling with some specific adaptive complexity. Our approach builds on parallel simulation techniques from scientific computing."}
+{"idx": 2, "title": "Parallel simulation for sampling under isoperimetry and...", "date": "", "ddg_snippet": "In score - based diffusion models , one considers forward process.-strongly log - concave and. β𝛽\\betaitalic_β.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.07435v1", "content": "In score - based diffusion models , one considers forward process.-strongly log - concave and. β𝛽\\betaitalic_β."}
+{"idx": 3, "title": "Parallel Sampling Algorithms", "date": "", "ddg_snippet": "2.4. Diffusion -based Parallel Sampling and Scientific Simulation . Recent algorithms for sampling from high-dimensional, possibly non- log - concave , or data-driven ( score - based ) distributions use methods rooted in ODE/SDE discretizations", "subpage_snippet": "", "source": "www.emergentmind.com", "link": "https://www.emergentmind.com/topics/parallel-sampling-algorithms", "content": "2.4. Diffusion -based Parallel Sampling and Scientific Simulation . Recent algorithms for sampling from high-dimensional, possibly non- log - concave , or data-driven ( score - based ) distributions use methods rooted in ODE/SDE discretizations"}
+{"idx": 4, "title": "(PDF) Parallel simulation for sampling under isoperimetry and...", "date": "", "ddg_snippet": "score - based diffusion models . Huanjian Zhou1, 2 and Masashi Sugiyama2, 1. 1The University of Tokyo, Tokyo, Japan.methods in scientific computation for dynamics-based sampling and diffusion models .", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/386976226_Parallel_simulation_for_sampling_under_isoperimetry_and_score-based_diffusion_models", "content": "score - based diffusion models . Huanjian Zhou1, 2 and Masashi Sugiyama2, 1. 1The University of Tokyo, Tokyo, Japan.methods in scientific computation for dynamics-based sampling and diffusion models ."}
+{"idx": 5, "title": "Improved Convergence Rate for Diffusion Probabilistic Models", "date": "", "ddg_snippet": "Improving our understanding of diffusion models , including mainly convergence analysis for such models , has attracted a lot of interests.(2024). Our theory accommodates $\\varepsilon$-accurate score estimates, and does not require log - concavity on the target distribution.", "subpage_snippet": "", "source": "proceedings.iclr.cc", "link": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/dcd9ed95f4abeaa1c22c4c2fd4231930-Abstract-Conference.html", "content": "Improving our understanding of diffusion models , including mainly convergence analysis for such models , has attracted a lot of interests.(2024). Our theory accommodates $\\varepsilon$-accurate score estimates, and does not require log - concavity on the target distribution."}
+{"idx": 6, "title": "[Literature Review] Faster Sampling without Isoperimetry via...", "date": "", "ddg_snippet": "[Literature Review] Parallel simulation for sampling under isoperimetry and score - based diffusion models .", "subpage_snippet": "", "source": "www.themoonlight.io", "link": "https://www.themoonlight.io/en/review/faster-sampling-without-isoperimetry-via-diffusion-based-monte-carlo", "content": "[Literature Review] Parallel simulation for sampling under isoperimetry and score - based diffusion models ."}
+{"idx": 7, "title": "Zeroth-Order Sampling Methods for", "date": "", "ddg_snippet": "Zeroth-Order Sampling Methods for Non- Log - Concave Distributions: Alleviating.It first describes a framework, Denoising Diffusion Monte Carlo (DDMC), based on the simulation of a denoising diffusion process with its score function approximated by a generic Monte Carlo estimator.", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper_files/paper/2024/file/8337b176767d948275d0b6d17cf4221e-Paper-Conference.pdf", "content": "Zeroth-Order Sampling Methods for Non- Log - Concave Distributions: Alleviating.It first describes a framework, Denoising Diffusion Monte Carlo (DDMC), based on the simulation of a denoising diffusion process with its score function approximated by a generic Monte Carlo estimator."}
+{"idx": 8, "title": "Score - based Diffusion Models via... -- a Technical Tutorial-Bohrium", "date": "", "ddg_snippet": "Score - based Diffusion Models via Stochastic Differential Equations -- a Technical Tutorial.", "subpage_snippet": "", "source": "www.bohrium.com", "link": "https://www.bohrium.com/paper-details/score-based-diffusion-models-via-stochastic-differential-equations-a-technical-tutorial/964420384331923471-108614", "content": "Score - based Diffusion Models via Stochastic Differential Equations -- a Technical Tutorial."}
+{"idx": 9, "title": "Faster Sampling without Isoperimetry via Diffusion - based Monte Carlo", "date": "", "ddg_snippet": "Recursive Score Estimation and Reverse Diffusion Sampling .", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v247/huang24a/huang24a.pdf", "content": "Recursive Score Estimation and Reverse Diffusion Sampling ."}
diff --git a/data/sampled_jsons/sitearxiv.org_2007.00016.jsonl b/data/sampled_jsons/sitearxiv.org_2007.00016.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..700a72da4f0a19cf88eb5c2d57da8abf440471e9
--- /dev/null
+++ b/data/sampled_jsons/sitearxiv.org_2007.00016.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2007.00016] Solver-in-the-Loop: Learning from Differentiable Physics ...", "date": "", "ddg_snippet": "Abstract page for arXiv paper 2007.00016 : Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solvers", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2007.00016", "content": "Abstract page for arXiv paper 2007.00016 : Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solvers"}
+{"idx": 1, "title": "Abstract page for arXiv paper 0705.0164: Inflationary Cosmology", "date": "", "ddg_snippet": "I give a general review of the history of inflationary cosmology and of its present status.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0705.0164", "content": "I give a general review of the history of inflationary cosmology and of its present status."}
+{"idx": 2, "title": "[2007.00016] Solver-in-the-Loop: Learning from Differentiable Physics ...", "date": "", "ddg_snippet": "Finding accurate solutions to partial differential equations (PDEs) is a crucial task in all scientific and engineering disciplines. It has recently been shown that machine learning methods can improve the solution acc…", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2007.00016", "content": "Finding accurate solutions to partial differential equations (PDEs) is a crucial task in all scientific and engineering disciplines. It has recently been shown that machine learning methods can improve the solution acc…"}
+{"idx": 3, "title": "[0704.2043v1] Enhanced Half-Metallicity in Edge-Oxidized Zigzag...", "date": "", "ddg_snippet": "[Submitted on 16 Apr 2007]. Title:Enhanced Half-Metallicity in Edge-Oxidized Zigzag Graphene Nanoribbons. Authors:Oded Hod, Veronica Barone, Juan E. Peralta, Gustavo E. Scuseria.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0704.2043v1", "content": "[Submitted on 16 Apr 2007]. Title:Enhanced Half-Metallicity in Edge-Oxidized Zigzag Graphene Nanoribbons. Authors:Oded Hod, Veronica Barone, Juan E. Peralta, Gustavo E. Scuseria."}
+{"idx": 4, "title": "[0704.1880] Quantum Quenches in Extended Systems", "date": "", "ddg_snippet": "We study in general the time-evolution of correlation functions in a extended quantum system after the quench of a parameter in the hamiltonian.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0704.1880", "content": "We study in general the time-evolution of correlation functions in a extended quantum system after the quench of a parameter in the hamiltonian."}
+{"idx": 5, "title": "Random-phase reservoir and a quantum resistor: The Lloyd model", "date": "", "ddg_snippet": "This conversion of the random phase into the random Cauchy potential is a notable feature of our work.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0705.2815v3", "content": "This conversion of the random phase into the random Cauchy potential is a notable feature of our work."}
+{"idx": 6, "title": "[0705.0665] Lagrangian subcategories and braided tensor...", "date": "", "ddg_snippet": "In view of results of 0704.0195v2 this gives a complete description of all braided tensor equivalent pairs of twisted quantum doubles of finite groups.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0705.0665", "content": "In view of results of 0704.0195v2 this gives a complete description of all braided tensor equivalent pairs of twisted quantum doubles of finite groups."}
+{"idx": 7, "title": "[0704.2049v1] Magnetosonic solitons in a Fermionic quantum plasma", "date": "", "ddg_snippet": "In the limit of a weak magnetic field and an ultracold plasma, the electron spin contribution to the soliton formation may dominate.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/0704.2049v1", "content": "In the limit of a weak magnetic field and an ultracold plasma, the electron spin contribution to the soliton formation may dominate."}
+{"idx": 8, "title": "Neural Operator: Learning Maps Between Function Spaces With ...", "date": "", "ddg_snippet": "The classical development of neural networks has primarily focused on learning mappings between finite dimensional Euclidean spaces or finite sets. We propose a generalization of neural networks to learn operators, termed neural operators, that map between infinite dimensional function spaces. We formulate the neural operator as a composition of linear integral operators and nonlinear ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2108.08481v6", "content": "The classical development of neural networks has primarily focused on learning mappings between finite dimensional Euclidean spaces or finite sets. We propose a generalization of neural networks to learn operators, termed neural operators, that map between infinite dimensional function spaces. We formulate the neural operator as a composition of linear integral operators and nonlinear ..."}
+{"idx": 9, "title": "P3D: Scalable Neural Surrogates for High-Resolution 3D Physics ...", "date": "", "ddg_snippet": "Abstract We present a scalable framework for learning deterministic and probabilistic neural surrogates for high-resolution 3D physics simulations. We introduce a hybrid CNN-Transformer backbone architecture targeted for 3D physics simulations, which significantly outperforms existing architectures in terms of speed and accuracy. Our proposed network can be pretrained on small patches of the ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2509.10186v1", "content": "Abstract We present a scalable framework for learning deterministic and probabilistic neural surrogates for high-resolution 3D physics simulations. We introduce a hybrid CNN-Transformer backbone architecture targeted for 3D physics simulations, which significantly outperforms existing architectures in terms of speed and accuracy. Our proposed network can be pretrained on small patches of the ..."}
diff --git a/data/sampled_jsons/sitearxiv.org_2109.01903_abstract.jsonl b/data/sampled_jsons/sitearxiv.org_2109.01903_abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..84ad9753affc300850af07fda99be7952350732a
--- /dev/null
+++ b/data/sampled_jsons/sitearxiv.org_2109.01903_abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "[2109.01903] Robust fine-tuning of zero-shot models - arXiv.org", "date": "", "ddg_snippet": "Abstract page for arXiv paper 2109.01903 : Robust fine-tuning of zero-shot models", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2109.01903", "content": "Abstract page for arXiv paper 2109.01903 : Robust fine-tuning of zero-shot models"}
+{"idx": 1, "title": "[2109.01903] Robust fine-tuning of zero-shot models - ar5iv", "date": "", "ddg_snippet": "Abstract Large pre-trained models such as CLIP or ALIGN offer consistent accuracy across a range of data distributions when performing zero-shot inference (i.e., without fine-tuning on a specific dataset). Although existing fine-tuning methods substantially improve accuracy on a given target distribution, they often reduce robustness to distribution shifts. We address this tension by ...", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2109.01903", "content": "Abstract Large pre-trained models such as CLIP or ALIGN offer consistent accuracy across a range of data distributions when performing zero-shot inference (i.e., without fine-tuning on a specific dataset). Although existing fine-tuning methods substantially improve accuracy on a given target distribution, they often reduce robustness to distribution shifts. We address this tension by ..."}
+{"idx": 2, "title": "Robust ne-tuning of zero-shot models - arXiv.org", "date": "", "ddg_snippet": "Abstract out ne-tuning on a speci c dataset). Although existing ne-tuning methods substantially improve accuracy on a given target distribution, they often redu e robustness to distribution shifts. We address this tension by introducing a simple and e ective method for improving robustness while ne-tuning: ensembling the weights of the zer", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2109.01903", "content": "Abstract out ne-tuning on a speci c dataset). Although existing ne-tuning methods substantially improve accuracy on a given target distribution, they often redu e robustness to distribution shifts. We address this tension by introducing a simple and e ective method for improving robustness while ne-tuning: ensembling the weights of the zer"}
+{"idx": 3, "title": "Matryoshka Representation Learning - arXiv.org", "date": "", "ddg_snippet": "Abstract Learned representations are a central component in modern ML systems, serving a multitude of downstream tasks. When training such representations, it is often the case that computational and statistical constraints for each downstream task are unknown. In this context, rigid fixed-capacity representations can be either over or under-accommodating to the task at hand. This leads us to ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2205.13147v4", "content": "Abstract Learned representations are a central component in modern ML systems, serving a multitude of downstream tasks. When training such representations, it is often the case that computational and statistical constraints for each downstream task are unknown. In this context, rigid fixed-capacity representations can be either over or under-accommodating to the task at hand. This leads us to ..."}
+{"idx": 4, "title": "[2307.01903] Probing general relativistic spin-orbit coupling with ...", "date": "", "ddg_snippet": "Abstract page for arXiv paper 2307.01903: Probing general relativistic spin-orbit coupling with gravitational waves from hierarchical triple systems", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2307.01903", "content": "Abstract page for arXiv paper 2307.01903: Probing general relativistic spin-orbit coupling with gravitational waves from hierarchical triple systems"}
+{"idx": 5, "title": "Contrastive Abstraction for Reinforcement Learning", "date": "", "ddg_snippet": "Abstract Learning agents with reinforcement learning is difficult when dealing with long trajectories that involve a large number of states. To address these learning problems effectively, the number of states can be reduced by abstract representations that cluster states. In principle, deep reinforcement learning can find abstract states, but end-to-end learning is unstable. We propose ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2410.00704v1", "content": "Abstract Learning agents with reinforcement learning is difficult when dealing with long trajectories that involve a large number of states. To address these learning problems effectively, the number of states can be reduced by abstract representations that cluster states. In principle, deep reinforcement learning can find abstract states, but end-to-end learning is unstable. We propose ..."}
+{"idx": 6, "title": "[2109.01903v3] Robust fine-tuning of zero-shot models", "date": "", "ddg_snippet": "Large pre-trained models such as CLIP or ALIGN offer consistent accuracy across a range of data distributions when performing zero-shot inference (i.e., without fine-tuning on a specific dataset). Although existing fine-tuning methods substantially improve accuracy on a given target distribution, they often reduce robustness to distribution shifts. We address this tension by introducing a ...", "subpage_snippet": "", "source": "export.arxiv.org", "link": "http://export.arxiv.org/abs/2109.01903v3", "content": "Large pre-trained models such as CLIP or ALIGN offer consistent accuracy across a range of data distributions when performing zero-shot inference (i.e., without fine-tuning on a specific dataset). Although existing fine-tuning methods substantially improve accuracy on a given target distribution, they often reduce robustness to distribution shifts. We address this tension by introducing a ..."}
+{"idx": 7, "title": "Visual Classification via Description from Large Language Models", "date": "", "ddg_snippet": "Visual Classification via Description from Large Language Models Abstract Vision-language models (VLMs) such as CLIP have shown promising performance on a variety of recognition tasks using the standard zero-shot classification procedure - computing similarity between the query image and the embedded words for each category.", "subpage_snippet": "", "source": "ar5iv.labs.arxiv.org", "link": "https://ar5iv.labs.arxiv.org/html/2210.07183", "content": "Visual Classification via Description from Large Language Models Abstract Vision-language models (VLMs) such as CLIP have shown promising performance on a variety of recognition tasks using the standard zero-shot classification procedure - computing similarity between the query image and the embedded words for each category."}
+{"idx": 8, "title": "RaVL: Discovering and Mitigating Spurious Correlations in Fine-Tuned ...", "date": "", "ddg_snippet": "Abstract Fine-tuned vision-language models (VLMs) often capture spurious correlations between image features and textual attributes, resulting in degraded zero-shot performance at test time. Existing approaches for addressing spurious correlations (i) primarily operate at the global image-level rather than intervening directly on fine-grained image features and (ii) are predominantly designed ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2411.04097v1", "content": "Abstract Fine-tuned vision-language models (VLMs) often capture spurious correlations between image features and textual attributes, resulting in degraded zero-shot performance at test time. Existing approaches for addressing spurious correlations (i) primarily operate at the global image-level rather than intervening directly on fine-grained image features and (ii) are predominantly designed ..."}
+{"idx": 9, "title": "Learning Neural Causal Models with Active Interventions", "date": "", "ddg_snippet": "Discovering causal structures from data is a challenging inference problem of fundamental importance in all areas of science. The appealing properties of neural networks have recently led to a surge of interest in differentiable neural network-based methods for learning causal structures from data. So far, differentiable causal discovery has focused on static datasets of observational or fixed ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2109.02429", "content": "Discovering causal structures from data is a challenging inference problem of fundamental importance in all areas of science. The appealing properties of neural networks have recently led to a surge of interest in differentiable neural network-based methods for learning causal structures from data. So far, differentiable causal discovery has focused on static datasets of observational or fixed ..."}
diff --git a/data/sampled_jsons/sitearxiv.org_2111.03980_abstract.jsonl b/data/sampled_jsons/sitearxiv.org_2111.03980_abstract.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..7b353ac627cf5171b369a01718fc74dd9e9b37db
--- /dev/null
+++ b/data/sampled_jsons/sitearxiv.org_2111.03980_abstract.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Dynamic Algorithms Against an Adaptive Adversary", "date": "", "ddg_snippet": "by A Beimel · 2021 · Cited by 53 — Abstract page for arXiv paper 2111.03980 : Dynamic Algorithms Against an Adaptive Adversary: Generic Constructions and Lower Bounds.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2111.03980", "content": "by A Beimel · 2021 · Cited by 53 — Abstract page for arXiv paper 2111.03980 : Dynamic Algorithms Against an Adaptive Adversary: Generic Constructions and Lower Bounds."}
+{"idx": 1, "title": "Dynamic Algorithms Against an Adaptive Adversary", "date": "", "ddg_snippet": "by A Beimel · 2021 · Cited by 53 — Abstract . A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2111.03980", "content": "by A Beimel · 2021 · Cited by 53 — Abstract . A dynamic algorithm against an adaptive adversary is required to be correct when the adversary chooses the next update after ..."}
+{"idx": 2, "title": "Robust Cardinality Sketches for Adaptive Queries", "date": "", "ddg_snippet": "8 Feb 2025 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Ben-Eliezer et al. (2021a)", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2502.05723v1", "content": "8 Feb 2025 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Ben-Eliezer et al. (2021a)"}
+{"idx": 3, "title": "Tight Quadratic Black-Box Attacks on Sketches for l2 Norm ...", "date": "", "ddg_snippet": "by S Ahmadian · 2025 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. [8] Omri Ben-Eliezer, ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2507.16345", "content": "by S Ahmadian · 2025 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. [8] Omri Ben-Eliezer, ..."}
+{"idx": 4, "title": "Adaptive Data Analysis with Correlated Observations", "date": "", "ddg_snippet": "by A Kontorovich · 2022 · Cited by 13 — Abstract . The vast majority of the work on adaptive data analysis focuses on the case where the samples in the dataset are independent.", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2201.08704", "content": "by A Kontorovich · 2022 · Cited by 13 — Abstract . The vast majority of the work on adaptive data analysis focuses on the case where the samples in the dataset are independent."}
+{"idx": 5, "title": "A Framework for Adversarial Streaming via Differential ...", "date": "", "ddg_snippet": "by I Attias · 2021 · Cited by 42 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. 15. Page 16. [BNS16] Amos ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2107.14527", "content": "by I Attias · 2021 · Cited by 42 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. 15. Page 16. [BNS16] Amos ..."}
+{"idx": 6, "title": "Tight Quadratic Bounds for Adaptive Queries on Cardinality ...", "date": "", "ddg_snippet": "by E Cohen · 2024 · Cited by 4 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Omri Ben-Eliezer, Talya ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2411.06370", "content": "by E Cohen · 2024 · Cited by 4 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Omri Ben-Eliezer, Talya ..."}
+{"idx": 7, "title": "The Target-Charging Technique for Privacy Analysis ...", "date": "", "ddg_snippet": "by E Cohen · 2023 · Cited by 2 — Abstract. We propose the Target Charging Technique (TCT), a unified privacy analysis framework for interactive settings where a sensitive ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2302.11044", "content": "by E Cohen · 2023 · Cited by 2 — Abstract. We propose the Target Charging Technique (TCT), a unified privacy analysis framework for interactive settings where a sensitive ..."}
+{"idx": 8, "title": "A Tight Lower Bound on the Robustness of CountSketch to ...", "date": "", "ddg_snippet": "by E Cohen · 2022 · Cited by 12 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. [8] Omri Ben-Eliezer, ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2207.00956", "content": "by E Cohen · 2022 · Cited by 12 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. [8] Omri Ben-Eliezer, ..."}
+{"idx": 9, "title": "Robust Cardinality Sketches for Adaptive Queries", "date": "", "ddg_snippet": "by E Cohen · 2025 · Cited by 1 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Ben-Eliezer, O., Eden, T., ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/pdf/2502.05723", "content": "by E Cohen · 2025 · Cited by 1 — Dynamic algorithms against an adaptive adversary : Generic constructions and lower bounds. CoRR, abs/2111.03980, 2021. Ben-Eliezer, O., Eden, T., ..."}
diff --git a/data/sampled_jsons/sitegithub.com_uiuc-kang-lab_cve-bench_CVE_list_Table_6_CVSS_rating.jsonl b/data/sampled_jsons/sitegithub.com_uiuc-kang-lab_cve-bench_CVE_list_Table_6_CVSS_rating.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..439e0337b71fca37f07c60534e328d2e5d4de2ca
--- /dev/null
+++ b/data/sampled_jsons/sitegithub.com_uiuc-kang-lab_cve-bench_CVE_list_Table_6_CVSS_rating.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "GitHub - uiuc-kang-lab/cve-bench: CVE-Bench: A Benchmark for ...", "date": "", "ddg_snippet": "Apr 24, 2025 · This repository contains data and code used in the CVE-Bench (paper, blog), which is for evaluating AI agents on real world web vulnerabilities and exploits collected from National Vulnerability Database. CVE-Bench includes 40 critical-severity Common Vulnerability and Exposures ( CVE ) with the reference automatic exploits available on requests.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench", "content": "Apr 24, 2025 · This repository contains data and code used in the CVE-Bench (paper, blog), which is for evaluating AI agents on real world web vulnerabilities and exploits collected from National Vulnerability Database. CVE-Bench includes 40 critical-severity Common Vulnerability and Exposures ( CVE ) with the reference automatic exploits available on requests."}
+{"idx": 1, "title": "cve - bench /.env.sample.solution at main · uiuc - kang - lab / cve - bench", "date": "", "ddg_snippet": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /.env.sample.solution at main · uiuc - kang - lab / cve - bench .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/blob/main/.env.sample.solution", "content": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /.env.sample.solution at main · uiuc - kang - lab / cve - bench ."}
+{"idx": 2, "title": "cve - bench /pyproject.toml at main · uiuc - kang - lab / cve - bench · GitHub", "date": "", "ddg_snippet": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /pyproject.toml at main · uiuc - kang - lab / cve - bench .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/blob/main/pyproject.toml", "content": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /pyproject.toml at main · uiuc - kang - lab / cve - bench ."}
+{"idx": 3, "title": "cve - bench /LICENSE at main · uiuc - kang - lab / cve - bench · GitHub", "date": "", "ddg_snippet": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /LICENSE at main · uiuc - kang - lab / cve - bench .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/blob/main/LICENSE", "content": "CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve - bench /LICENSE at main · uiuc - kang - lab / cve - bench ."}
+{"idx": 4, "title": "yuxuan18 (Yuxuan Zhu) · GitHub", "date": "", "ddg_snippet": "uiuc - kang - lab / cve - bench uiuc - kang - lab / cve - bench Public. CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities .Created a pull request in uiuc - kang - lab / cve - bench that received 5 comments.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/yuxuan18", "content": "uiuc - kang - lab / cve - bench uiuc - kang - lab / cve - bench Public. CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities .Created a pull request in uiuc - kang - lab / cve - bench that received 5 comments."}
+{"idx": 5, "title": "aokellermann has 77 repositories available. Follow their code on GitHub.", "date": "", "ddg_snippet": "uiuc - kang - lab / cve - bench uiuc - kang - lab / cve - bench Public. CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities .", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/aokellermann", "content": "uiuc - kang - lab / cve - bench uiuc - kang - lab / cve - bench Public. CVE - Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities ."}
+{"idx": 6, "title": "uiuc-kang-lab · GitHub", "date": "", "ddg_snippet": "uiuc-kang-lab Popular repositories cve-bench Public CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities Python 96 16 InjecAgent Public Python 79 12 agentic-benchmarks Public", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab", "content": "uiuc-kang-lab Popular repositories cve-bench Public CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities Python 96 16 InjecAgent Public Python 79 12 agentic-benchmarks Public"}
+{"idx": 7, "title": "fix: incorrect service version on CVE-2024-5314 and CVE-2024 ...", "date": "", "ddg_snippet": "CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - fix: incorrect service version on CVE -2024-5314 and CVE -2024-5315 by ethanho-oai · Pull Request #16 · uiuc-kang-lab / cve-bench", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/pull/16/commits", "content": "CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - fix: incorrect service version on CVE -2024-5314 and CVE -2024-5315 by ethanho-oai · Pull Request #16 · uiuc-kang-lab / cve-bench"}
+{"idx": 8, "title": "cve-bench/README.md at main · uiuc-kang-lab/cve-bench", "date": "", "ddg_snippet": "Apr 24, 2025 · This repository contains data and code used in the CVE-Bench (paper, blog), which is for evaluating AI agents on real world web vulnerabilities and exploits collected from National Vulnerability Database. CVE-Bench includes 40 critical-severity Common Vulnerability and Exposures ( CVE ) with the reference automatic exploits available on requests.", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/blob/main/README.md", "content": "Apr 24, 2025 · This repository contains data and code used in the CVE-Bench (paper, blog), which is for evaluating AI agents on real world web vulnerabilities and exploits collected from National Vulnerability Database. CVE-Bench includes 40 critical-severity Common Vulnerability and Exposures ( CVE ) with the reference automatic exploits available on requests."}
+{"idx": 9, "title": "cve-bench/run at main · uiuc-kang-lab/cve-bench · GitHub", "date": "", "ddg_snippet": "CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve-bench /run at main · uiuc-kang-lab / cve-bench", "subpage_snippet": "", "source": "github.com", "link": "https://github.com/uiuc-kang-lab/cve-bench/blob/main/run", "content": "CVE-Bench : A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities - cve-bench /run at main · uiuc-kang-lab / cve-bench"}
diff --git a/data/sampled_jsons/siteresearchgate.net_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Mo.jsonl b/data/sampled_jsons/siteresearchgate.net_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Mo.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..6ce61d9c0b5c19b41c58026627db0fb6b80948b4
--- /dev/null
+++ b/data/sampled_jsons/siteresearchgate.net_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Mo.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Generalizable Origin Identification for Text-Guided Image-to ...", "date": "", "ddg_snippet": "Jan 4, 2025 · The curated dataset, OriPID, contains abundant Origins and guided Prompts, which can be used to train and test potential IDentification models across various diffusion models .", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/387767437_Generalizable_Origin_Identification_for_Text-Guided_Image-to-Image_Diffusion_Models", "content": "Jan 4, 2025 · The curated dataset, OriPID, contains abundant Origins and guided Prompts, which can be used to train and test potential IDentification models across various diffusion models ."}
+{"idx": 1, "title": "(PDF) Enhancing Text-to-Image Generation: Integrating CLIP ... Text To Image Generation By Using Stable Diffusion Model With ... Generalizable Synthetic Image Detection via Language-guided ... Text-to-Image Synthesis with Generative Models: Methods ... Perception and evaluation of text-to-image generative AI ... SG-Adapter: Enhancing Text-to-Image Generation with Scene ...", "date": "", "ddg_snippet": "Nov 15, 2024 · In this work, we develop a novel deep architecture and GAN formulation to effectively bridge these advances in text and image modeling , translating visual concepts from characters to pixels. Oct 31, 2023 · We demonstrate that our model produces higher quality samples than other approaches and generates images with novel scene compositions corresponding to previously unseen captions in the dataset. May 23, 2023 · In this work, we propose a simple yet very effective synthetic image detection method via a language-guided contrastive learning and a new formulation of the detection problem. Jan 1, 2024 · This study offers a concise overview of text - to - image generative models by examining the existing body of literature and provide a deeper understanding of this topic. Oct 1, 2024 · This study evaluates four prominent text - to - image generative models -DALL-E, Google Imagen, Stable Diffusion, and GROK AI emphasizing on the text - to - image diffusion models . May 24, 2024 · In this paper, we introduce the Scene Graph Adapter (SG-Adapter), leveraging the structured representation of scene graphs to rectify inaccuracies in the original text embeddings.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/385904859_Enhancing_Text-to-Image_Generation_Integrating_CLIP_and_Diffusion_Models_for_Improved_Visual_Accuracy_and_Semantic_Consistency", "content": "Nov 15, 2024 · In this work, we develop a novel deep architecture and GAN formulation to effectively bridge these advances in text and image modeling , translating visual concepts from characters to pixels. Oct 31, 2023 · We demonstrate that our model produces higher quality samples than other approaches and generates images with novel scene compositions corresponding to previously unseen captions in the dataset. May 23, 2023 · In this work, we propose a simple yet very effective synthetic image detection method via a language-guided contrastive learning and a new formulation of the detection problem. Jan 1, 2024 · This study offers a concise overview of text - to - image generative models by examining the existing body of literature and provide a deeper understanding of this topic. Oct 1, 2024 · This study evaluates four prominent text - to - image generative models -DALL-E, Google Imagen, Stable Diffusion, and GROK AI emphasizing on the text - to - image diffusion models . May 24, 2024 · In this paper, we introduce the Scene Graph Adapter (SG-Adapter), leveraging the structured representation of scene graphs to rectify inaccuracies in the original text embeddings."}
+{"idx": 2, "title": "Text To Image Generation By Using Stable Diffusion Model With ...", "date": "", "ddg_snippet": "Oct 31, 2023 · We demonstrate that our model produces higher quality samples than other approaches and generates images with novel scene compositions corresponding to previously unseen captions in the dataset.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/375113719_Text_To_Image_Generation_By_Using_Stable_Diffusion_Model_With_Variational_Autoencoder_Decoder", "content": "Oct 31, 2023 · We demonstrate that our model produces higher quality samples than other approaches and generates images with novel scene compositions corresponding to previously unseen captions in the dataset."}
+{"idx": 3, "title": "Generalizable Synthetic Image Detection via Language-guided ...", "date": "", "ddg_snippet": "May 23, 2023 · In this work, we propose a simple yet very effective synthetic image detection method via a language-guided contrastive learning and a new formulation of the detection problem.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/370981524_Generalizable_Synthetic_Image_Detection_via_Language-guided_Contrastive_Learning", "content": "May 23, 2023 · In this work, we propose a simple yet very effective synthetic image detection method via a language-guided contrastive learning and a new formulation of the detection problem."}
+{"idx": 4, "title": "Text-to-Image Synthesis with Generative Models: Methods ...", "date": "", "ddg_snippet": "Jan 1, 2024 · This study offers a concise overview of text - to - image generative models by examining the existing body of literature and provide a deeper understanding of this topic.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/378122729_Text-to-Image_Synthesis_with_Generative_Models_Methods_Datasets_Performance_Metrics_Challenges_and_Future_Direction", "content": "Jan 1, 2024 · This study offers a concise overview of text - to - image generative models by examining the existing body of literature and provide a deeper understanding of this topic."}
+{"idx": 5, "title": "Perception and evaluation of text-to-image generative AI ...", "date": "", "ddg_snippet": "Oct 1, 2024 · This study evaluates four prominent text - to - image generative models -DALL-E, Google Imagen, Stable Diffusion, and GROK AI emphasizing on the text - to - image diffusion models .", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/385290574_Perception_and_evaluation_of_text-to-image_generative_AI_models_a_comparative_study_of_DALL-E_Google_Imagen_GROK_and_Stable_Diffusion", "content": "Oct 1, 2024 · This study evaluates four prominent text - to - image generative models -DALL-E, Google Imagen, Stable Diffusion, and GROK AI emphasizing on the text - to - image diffusion models ."}
+{"idx": 6, "title": "SG-Adapter: Enhancing Text-to-Image Generation with Scene ...", "date": "", "ddg_snippet": "May 24, 2024 · In this paper, we introduce the Scene Graph Adapter (SG-Adapter), leveraging the structured representation of scene graphs to rectify inaccuracies in the original text embeddings.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/380895117_SG-Adapter_Enhancing_Text-to-Image_Generation_with_Scene_Graph_Guidance", "content": "May 24, 2024 · In this paper, we introduce the Scene Graph Adapter (SG-Adapter), leveraging the structured representation of scene graphs to rectify inaccuracies in the original text embeddings."}
+{"idx": 7, "title": "(PDF) EMOCONV-Diff: Diffusion-Based Speech Emotion Conversion", "date": "", "ddg_snippet": "... Diffusion -Based Speech Emotion Conversion ... The stop gradient function stops the accumulation of the gradients of the inputs during the training.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/379817290_EMOCONV-Diff_Diffusion-Based_Speech_Emotion_Conversion_for_Non-Parallel_and_in-the-Wild_Data", "content": "... Diffusion -Based Speech Emotion Conversion ... The stop gradient function stops the accumulation of the gradients of the inputs during the training."}
+{"idx": 8, "title": "Efficient Graph-Based Image Segmentation", "date": "", "ddg_snippet": "Training-free open-vocabulary semantic segmentation (OVS) aims to segment images given a set of arbitrary textual categories without costly model ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/340575292_Efficient_Graph-Based_Image_Segmentation", "content": "Training-free open-vocabulary semantic segmentation (OVS) aims to segment images given a set of arbitrary textual categories without costly model ..."}
+{"idx": 9, "title": "Learning to Exploit Temporal Structure for Biomedical", "date": "", "ddg_snippet": "... g ., 10, [12] [13][14] , incorporated information from X-ray acquired at previous examination, their goal was to obtain better alignment between text ...", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/373324805_Learning_to_Exploit_Temporal_Structure_for_Biomedical_Vision-Language_Processing", "content": "... g ., 10, [12] [13][14] , incorporated information from X-ray acquired at previous examination, their goal was to obtain better alignment between text ..."}
diff --git a/data/sampled_jsons/stochastic_combinatorial_bandits_non-continuous_loss_function_year_2022.jsonl b/data/sampled_jsons/stochastic_combinatorial_bandits_non-continuous_loss_function_year_2022.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9652f94b65c0648460fe9124111f285d86069520
--- /dev/null
+++ b/data/sampled_jsons/stochastic_combinatorial_bandits_non-continuous_loss_function_year_2022.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "A Framework for Adapting Offline Algorithms to Solve Combinatorial ...", "date": "", "ddg_snippet": "We investigate the problem of stochastic , combinatorial multi-armed bandits where the learner only has access to bandit feedback and the reward function can be non -linear.", "subpage_snippet": "", "source": "deepai.org", "link": "https://deepai.org/publication/a-framework-for-adapting-offline-algorithms-to-solve-combinatorial-multi-armed-bandit-problems-with-bandit-feedback", "content": "We investigate the problem of stochastic , combinatorial multi-armed bandits where the learner only has access to bandit feedback and the reward function can be non -linear."}
+{"idx": 1, "title": "[2101.01301] Adversarial Combinatorial Bandits with General...", "date": "", "ddg_snippet": "{The adversarial combinatorial bandit with general non -linear reward is an important open problem in bandit literature, and it is still unclear whether there is a significant gap from the case of linear reward, stochastic bandit , or semi- bandit feedback.}", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2101.01301", "content": "{The adversarial combinatorial bandit with general non -linear reward is an important open problem in bandit literature, and it is still unclear whether there is a significant gap from the case of linear reward, stochastic bandit , or semi- bandit feedback.}"}
+{"idx": 2, "title": "(PDF) Combinatorial Multi-Armed Bandit with General Reward...", "date": "", "ddg_snippet": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB) framework that allows a general nonlinear reward function , whose expected value may not depend only on the means of the input random variables but possibly on the entire distributions of these variables.", "subpage_snippet": "", "source": "www.researchgate.net", "link": "https://www.researchgate.net/publication/309388321_Combinatorial_Multi-Armed_Bandit_with_General_Reward_Functions", "content": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB) framework that allows a general nonlinear reward function , whose expected value may not depend only on the means of the input random variables but possibly on the entire distributions of these variables."}
+{"idx": 3, "title": "Adversarial Combinatorial Bandits with General Non -linear Reward...", "date": "", "ddg_snippet": "torial bandit with general non -linear reward is an. important open problem in bandit literature, and. it is still unclear whether there is a signicant gap. from the case of linear reward, stochastic bandit", "subpage_snippet": "", "source": "proceedings.mlr.press", "link": "https://proceedings.mlr.press/v139/han21b/han21b.pdf", "content": "torial bandit with general non -linear reward is an. important open problem in bandit literature, and. it is still unclear whether there is a signicant gap. from the case of linear reward, stochastic bandit"}
+{"idx": 4, "title": "Combinatorial Bandits Revisited", "date": "", "ddg_snippet": "4 Stochastic Combinatorial Bandits under Semi- bandit Feedback.in a straightforward manner to the case of continuous linear bandit problems, where the set of arms. is the unit sphere and one wants to maximize the dot product between the arm and an unknown.", "subpage_snippet": "", "source": "proceedings.neurips.cc", "link": "https://proceedings.neurips.cc/paper/2015/file/0ce2ffd21fc958d9ef0ee9ba5336e357-Paper.pdf", "content": "4 Stochastic Combinatorial Bandits under Semi- bandit Feedback.in a straightforward manner to the case of continuous linear bandit problems, where the set of arms. is the unit sphere and one wants to maximize the dot product between the arm and an unknown."}
+{"idx": 5, "title": "[PDF] Combinatorial Multi-Armed Bandit with... | Semantic Scholar", "date": "", "ddg_snippet": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB) framework that allows a general nonlinear reward function , whose expected value may not depend only on the means of the input random variables but possibly on the entire distributions of these variables.", "subpage_snippet": "", "source": "www.semanticscholar.org", "link": "https://www.semanticscholar.org/paper/Combinatorial-Multi-Armed-Bandit-with-General-Chen-Hu/1b50cee7c4ce8d576cb98e2da42d82b979eac4ff", "content": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB) framework that allows a general nonlinear reward function , whose expected value may not depend only on the means of the input random variables but possibly on the entire distributions of these variables."}
+{"idx": 6, "title": "(PDF) Regret Bounds for Stochastic Combinatorial Multi-Armed...", "date": "", "ddg_snippet": "This is the first work for combinatorial bandit where the reward received can be a non -linear function of the chosen $K$ arms. The direct use of multi-armed bandit requires choosing among $N$-choose-$K$ options making the state space large.", "subpage_snippet": "", "source": "www.academia.edu", "link": "https://www.academia.edu/123981524/Regret_Bounds_for_Stochastic_Combinatorial_Multi_Armed_Bandits_with_Linear_Space_Complexity", "content": "This is the first work for combinatorial bandit where the reward received can be a non -linear function of the chosen $K$ arms. The direct use of multi-armed bandit requires choosing among $N$-choose-$K$ options making the state space large."}
+{"idx": 7, "title": "Nonstochastic Contextual Combinatorial Bandits", "date": "", "ddg_snippet": "On the side of non - stochastic losses , the only relevant works we are aware of are those of Kale et al.As outlined in the introduction, we are considering a non - stochastic bandit problem with combinatorial actions and contexts provided in each timestep.", "subpage_snippet": "", "source": "www.cs.bme.hu", "link": "https://www.cs.bme.hu/~gergo/files/ZHCBN23.pdf", "content": "On the side of non - stochastic losses , the only relevant works we are aware of are those of Kale et al.As outlined in the introduction, we are considering a non - stochastic bandit problem with combinatorial actions and contexts provided in each timestep."}
+{"idx": 8, "title": "Combinatorial Multi-Armed Bandit with General Reward Functions", "date": "", "ddg_snippet": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB). framework that allows a general nonlinear reward function , whose expected value. may not depend only on the means of the input random variables but possibly.", "subpage_snippet": "", "source": "papers.nips.cc", "link": "https://papers.nips.cc/paper/2016/file/aa169b49b583a2b5af89203c2b78c67c-Paper.pdf", "content": "In this paper, we study the stochastic combinatorial multi-armed bandit (CMAB). framework that allows a general nonlinear reward function , whose expected value. may not depend only on the means of the input random variables but possibly."}
+{"idx": 9, "title": "DART: Adaptive Accept Reject Algorithm for Non -Linear...", "date": "", "ddg_snippet": "non -linear reward function and additional feedback, where. the feedback is a linear combination of the rewards of the K.Regret bounds for stochastic combinatorial multi-armed bandits with linear space complexity. arXiv preprint arXiv:1811.11925 .", "subpage_snippet": "", "source": "cdn.aaai.org", "link": "https://cdn.aaai.org/ojs/16812/16812-13-20306-1-2-20210518.pdf", "content": "non -linear reward function and additional feedback, where. the feedback is a linear combination of the rewards of the K.Regret bounds for stochastic combinatorial multi-armed bandits with linear space complexity. arXiv preprint arXiv:1811.11925 ."}
diff --git a/data/sampled_jsons/stochastic_localization_samplers.jsonl b/data/sampled_jsons/stochastic_localization_samplers.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9ba6eecbfa88b83549ccb3123511d553a6a7f855
--- /dev/null
+++ b/data/sampled_jsons/stochastic_localization_samplers.jsonl
@@ -0,0 +1,10 @@
+{"idx": 0, "title": "Coherently mitigating boson samplers with stochastic errors", "date": "", "ddg_snippet": "Given access to multiple stochastic boson samplers randomly picked from a distribution centred around the ideal boson sampler , we show that the ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2505.00102v1", "content": "Given access to multiple stochastic boson samplers randomly picked from a distribution centred around the ideal boson sampler , we show that the ..."}
+{"idx": 1, "title": "Noise-based Local Learning using Stochastic Magnetic Tunnel", "date": "", "ddg_snippet": "The demonstration of noise-based local learning is generally feasible with any neuromorphic material featuring stochastic behavior, such as ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2412.12783v3", "content": "The demonstration of noise-based local learning is generally feasible with any neuromorphic material featuring stochastic behavior, such as ..."}
+{"idx": 2, "title": "#1 - Slow stochastic settings forex", "date": "", "ddg_snippet": "Gulyani, targeted chemoradiation followed by surgical salvage is a slow stochastic settings forex effective approach for regional control of patients ...", "subpage_snippet": "", "source": "topforex-brokers.com", "link": "http://topforex-brokers.com/slow-stochastic-settings-forex.html", "content": "Gulyani, targeted chemoradiation followed by surgical salvage is a slow stochastic settings forex effective approach for regional control of patients ..."}
+{"idx": 3, "title": "[2502.00921] Blink of an eye: a simple theory for feature", "date": "", "ddg_snippet": "In this work we develop a simple, unifying theory to explain this phenomenon using the formalism of stochastic localization samplers .", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/abs/2502.00921", "content": "In this work we develop a simple, unifying theory to explain this phenomenon using the formalism of stochastic localization samplers ."}
+{"idx": 4, "title": "SMT(LIA) Sampling with High Diversity", "date": "", "ddg_snippet": "... mathit{HighDiv} italic_HighDiv sampling framework is composed of two distinct phases: the local search phase and the stochastic CDCL(T) phase ...", "subpage_snippet": "", "source": "arxiv.org", "link": "https://arxiv.org/html/2503.04782v1", "content": "... mathit{HighDiv} italic_HighDiv sampling framework is composed of two distinct phases: the local search phase and the stochastic CDCL(T) phase ..."}
+{"idx": 5, "title": "MeerKAT Pulsar Timing Array: the 4.5-yr data release and the", "date": "", "ddg_snippet": "Dispersion measure (DM) noise is a result of the stochastic variations in the column density of electrons along the line of sight to the pulsar, and ...", "subpage_snippet": "", "source": "academic.oup.com", "link": "https://academic.oup.com/mnras/article/536/2/1467/7912547", "content": "Dispersion measure (DM) noise is a result of the stochastic variations in the column density of electrons along the line of sight to the pulsar, and ..."}
+{"idx": 6, "title": "Chips-n-Salsa: A Java Library of Customizable, Hybridizable,", "date": "", "ddg_snippet": "... makes a small random modification to move to another local candidate solution; and also includes constructive algorithms like stochastic samplers ...", "subpage_snippet": "", "source": "www.cicirello.org", "link": "https://www.cicirello.org/publications/cicirello2020joss.html", "content": "... makes a small random modification to move to another local candidate solution; and also includes constructive algorithms like stochastic samplers ..."}
+{"idx": 7, "title": "Gbman2DN | SuperCollider 3.14.0-dev Help", "date": "", "ddg_snippet": "Classes (extension) | UGens > Generators'>Generators > Generators> Stochastic '> Stochastic", "subpage_snippet": "", "source": "doc.sccode.org", "link": "http://doc.sccode.org/Classes/Gbman2DN.html", "content": "Classes (extension) | UGens > Generators'>Generators > Generators> Stochastic '> Stochastic"}
+{"idx": 8, "title": "CS 8803: Counting and Sampling (Fall 2024) – Zongchen Chen", "date": "", "ddg_snippet": "... local-to-global induction, trickle-down theorem Oct 2, Oct 7, Oct 9: Spectral independence, optimal relaxation time, universality, disagreement ...", "subpage_snippet": "", "source": "sites.gatech.edu", "link": "https://sites.gatech.edu/zongchenchen/cs8803fall24/", "content": "... local-to-global induction, trickle-down theorem Oct 2, Oct 7, Oct 9: Spectral independence, optimal relaxation time, universality, disagreement ..."}
+{"idx": 9, "title": "Henon2DL | SuperCollider 3.14.0-dev Help", "date": "", "ddg_snippet": "Henon2DL.ar(SampleRate.ir/4, SampleRate.ir, MouseX.kr(1,1.4), MouseY.kr(0,0.3)) * 0.2 }.play(s); ( { Henon2DL.ar( 2200, 8800, LFNoise2.kr(1, 0.2, 1.2 ...", "subpage_snippet": "", "source": "doc.sccode.org", "link": "http://doc.sccode.org/Classes/Henon2DL.html", "content": "Henon2DL.ar(SampleRate.ir/4, SampleRate.ir, MouseX.kr(1,1.4), MouseY.kr(0,0.3)) * 0.2 }.play(s); ( { Henon2DL.ar( 2200, 8800, LFNoise2.kr(1, 0.2, 1.2 ..."}