Bushidō

Is anyone making a "Sumbit this paper to XYZ journal for me" bot? @yesnoerror @AdamDraper @erinmagennis

BREAKING. Officially announced as the newest @Solana conference: Accelerate taking place in NYC beginning May 19 👀

What happened with @yesnoerror and $YNE over the past week: - Meetings had or scheduled with 3 extremely prominent people who have a lot of insight into our mission and vision. These are the most influential people I have ever talked to in my life, and I've talked to a lot of people. More on this later. What is important to note here, is our vision and mission are simple, true, and large, and it is attracting a caliber of people I do not think have been involved in this space before. - Progress on v2 of the AI agent framework: Started running our first stress tests. Integrated with multiple research paper sources including arxiv, pubmed, openalex, biorxiv, medrxiv. Added support for both html, pdf, and LaTeX converter. Implemented vision AI for graphs, diagrams, and images of any kind. Interface to communicate with the AI agent and ask it to audit specific types of papers (covid, microplastics, ai, crypto, etc). This is a team effort, I built v1 myself, but me + the team is making v2. - 6,500+ research papers reviewed - 109 research papers with errors found - 14.2k+ token holders - @benparr has been handling 99% of inbound partnership calls doing 10-20 calls a day. - Continuing to build the advisory board. Selected two in total now, evaluating more. We are being *extremely* selective. - Partnerships launched with @bioprotocol @longcovidlabs, @brian_armstrong's @ResearchHubF, and @opencampus_xyz (funded by @BinanceLabs) - Scheduled multiple podcast appearances including @AdamDraper / @BoostVC. - Started to explore opportunities to not only look for errors in research papers, but to also analyze research papers and enable the AI agent to create useful research of it's own (the first experiment here will be done alongside @longcovidlabs to research COVID related trends, insights, and potential drugs that can be explored to treat Long COVID). This coming week: - Continuing work on v2, looking to get this live asap. Once this is live we will be able to process papers at a 100x efficiency. - Building v1 of a system that finds research papers that are discussed in news, social media, and have high citations. The goal is to prioritize papers that would be most notable and impactful if errors were found. - Scheduling more podcasts and spaces appearances. Thank you to everyone who has reached out or suggested a podcast to go on, these have all been organized into a spreadsheet and are being processed. - Booking travel to go to San Francisco to visit @boostvc, @coinbase, and others in early February. - Talking to design teams to do a design overhaul of the @yesnoerror website (@benparr is on a call with one right now). - Continuing to build out the advisory board. - Likely launching application process for people to join a @yesnoerror research and execution team. Look out for this. Special thanks/shoutouts this week to @AdamDraper @HeslinKim @BraytonKey @adaugelli @gregosuri @alex_gurevich @Rewkang @yoheinakajima @SigGravitas @DeFi_Cheetah I am most focused on and excited about the v2 of the AI agent being released, this will be a 0-1 type of upgrade, completely overshadowing the v1 that is live today. From the moment I wake up to the moment I go to sleep, $YNE is what I am focused on. This is not a random project, this is not a meme, this is not small. We are all doing something for the public good and it will only get better and more impactful with every day. This is not a normal project. This is an outlier. We will continue to attract the brightest minds. It doesn't matter that this was started as an experiment, in a way I don't see any other way it could have started. What matters is our commitment and conviction is 1000%, full speed ahead, no looking back. It is happening because we are making it happen. We must go even faster. Audit science. Accelerate humanity. Seek truth. Let's go.

Machine learning of microstructure–property relationships in materials with robust features from foundational vision transformers ⚠️ Mathematical Errors Found (CRITICAL) A major unit conversion error appears in the statement that 1 kgf/mm² equals (10³/g) GPa, which is off by about four orders of magnitude from the correct value (≈0.00981 GPa). 💥 Impact: This miscalculation would severely skew any hardness values converted using that relationship, potentially invalidating property comparisons or model training. ✅ No Discrepancies No additional numerical or data inconsistencies were observed aside from the noted hardness unit conversion mistake. 👶 For a child: They use special computer programs that look at pictures of metals and predict how strong they might be. 🎓 For college students: They explore using pre-trained vision transformers to extract features from images of metal microstructures for predicting their properties such as stiffness and hardness. 🧪 For PhD students: They develop a framework leveraging task-agnostic embeddings from large-scale pre-trained vision transformers, combined with dimensionality reduction and regression, to model microstructure–property linkages in simulated and experimental datasets. arxiv.org/html/2501.1863…