The best Hacker News stories from Show from the past day
Latest posts:
Show HN: Send private messages over a public channel
The other day I was on github assisting a compiler developer replicate an accidentally-quadratic observed behavior in generated code and was asked if I was willing to privately share my full source code with his company. While that was ultimately unnecessary as it was easy enough to find a simpler example, it did get me wondering about how exactly was I going to do that:<p>1. github does not (to my knowledge) allow private messages<p>2. all communication had thus far occurred on the public record (or at least Microsoft's version thereof)<p>3. my peer did not publish any contact information in his public profile, and out of respect for his privacy I did not look elsewhere<p>In yet another case of life imitates art imitates life, while I was building this, another developer on this very forum was complaining about not being able to set up a public "dropbox" on the internet in the context of receiving bluesky invite codes which, according to them, were being immediately redeemed by bots. I can't speak to that, but their structure does seem to make them easy to regexp against.<p>My objectives were to build something that:<p>1. provides reasonable security against doxxing oneself <i>to other participants</i><p>2. runs just about anywhere, even without an internet connection<p>3. requires no setup beforehand; since this turned out to be impossible, it was later relaxed to "no installation, but might benefit from it"<p>4. is easy to audit: no external dependencies other than the standard library<p>What I ended up with was this ~35KiB blob of js + html + css which (I hope) can be broadly summarized as "PGP for people who don't get invited to key signing parties". It is reverse-proxied via Cloudflare (I have a great ISP, but my aging homelab server probably can't handle any significant load) at:<p><pre><code> <https://plaintext.world/>.
</code></pre>
If, for whatever reason, Cloudflare is a part of your threat model, the <i>shasum -a 256</i> for the uncompressed file is:<p><pre><code> 153e8022213bc565b5b914a263162920a6039251dd6da5a77b3a37f35de9b1a3 /var/www/html/plaintext.world/index.html
</code></pre>
You'll find most of the relevant technical information on the page itself, though you may have to <i>view-source:</i> in the rather likely event that I've omitted something from the manual.<p>Since this is a technical, VC-backed forum, I'll add a couple more private FAQ entries that may be of interest to some of you, in exchange for the potential of free publicity. Due to HN posts being limited to 4k, you'll find them as a reply by yours truly.
Show HN: I've Published 28B Molecule Embeddings on AWS Open Data
I’ve finally finished a project that involved gathering 7 billion small molecules, each represented in SMILES notation and having fewer than 50 “heavy” non-hydrogen atoms. Those molecules were “fingerprinted”, producing 28 billion structural embeddings, using MACCS, PubChem, ECFP4, and FCFP4 techniques. These embeddings were indexed using Unum’s open-source tool USearch, to accelerate molecule search. This extensive dataset is now made available globally for free, thanks to AWS Open Data. You can find the complete data sheet and scripts for data visualization on GitHub.
Show HN: Mylens.ai – Create timeline for any topics with AI
MyLens.AI is a free AI that generates a beautiful timeline of key events for any topic you're interested in, all with one click. It is great to combine multiple topics and view their overlapping events in one cohesive timeline. One Timeline, Many Histories.<p>Website: <a href="https://mylens.ai" rel="nofollow noreferrer">https://mylens.ai</a><p>Your feedback is much appreciated.
Show HN: Mylens.ai – Create timeline for any topics with AI
MyLens.AI is a free AI that generates a beautiful timeline of key events for any topic you're interested in, all with one click. It is great to combine multiple topics and view their overlapping events in one cohesive timeline. One Timeline, Many Histories.<p>Website: <a href="https://mylens.ai" rel="nofollow noreferrer">https://mylens.ai</a><p>Your feedback is much appreciated.
Show HN: I built local copilot alternative using Codellama
Hey, HN!<p>I was trying to find a good Copilot alternative since my experience with copilot degraded a lot and recent OpenAI drama leaves me no hope for improvement and i decided to take a look at self-hosting options. At the same time Copilot net is a small one - just 12b parameters and 2k of context. You can run much bigger networks locally in 2023.<p>I have figured out that there are no good local options: most of the VS Code plugins simply connect neural networks to the code while copilot has extensive post and pre processing steps. Some of them still work with original copilot plugin (ie sends data to MS), some require complicated setup to make it work.<p>I implemented very basic ones to check if this would work and in my experience it works really good and i wanted to share it on HN. Feedback is appreciated!<p>Sources are here: <a href="https://github.com/ex3ndr/llama-coder">https://github.com/ex3ndr/llama-coder</a>
Show HN: I built local copilot alternative using Codellama
Hey, HN!<p>I was trying to find a good Copilot alternative since my experience with copilot degraded a lot and recent OpenAI drama leaves me no hope for improvement and i decided to take a look at self-hosting options. At the same time Copilot net is a small one - just 12b parameters and 2k of context. You can run much bigger networks locally in 2023.<p>I have figured out that there are no good local options: most of the VS Code plugins simply connect neural networks to the code while copilot has extensive post and pre processing steps. Some of them still work with original copilot plugin (ie sends data to MS), some require complicated setup to make it work.<p>I implemented very basic ones to check if this would work and in my experience it works really good and i wanted to share it on HN. Feedback is appreciated!<p>Sources are here: <a href="https://github.com/ex3ndr/llama-coder">https://github.com/ex3ndr/llama-coder</a>
I made a catalog of useful GPTs
I made a catalog of useful GPTs
Show HN: Stella Nera – Maddness Hardware Accelerator
Show HN: Stella Nera – Maddness Hardware Accelerator
Show HN: Stella Nera – Maddness Hardware Accelerator
Show HN: Neum AI – Open-source large-scale RAG framework
Over the last couple months we have been supporting developers in building large-scale RAG pipelines to process millions of pieces of data.<p>We documented our approach in an HN post (<a href="https://news.ycombinator.com/item?id=37824547">https://news.ycombinator.com/item?id=37824547</a>) a couple weeks ago. Today, we are open sourcing the framework we have developed.<p>The framework focuses on RAG data pipelines and provides scale, reliability, and data synchronization capabilities out of the box.<p>For those newer to RAG, it is a technique to provide context to Large Language Models. It consists of grabbing pieces of information (i.e. pieces of news articles, papers, descriptions, etc.) and incorporating them into prompts to help contextualize the responses. The technique goes one level deeper in finding the right pieces of information to incorporate. The search for relevant information is done through the use of vector embeddings and vector databases.<p>Those pieces of news articles, papers, etc. are transformed into a vector embedding that represents the semantic meaning of the information. These vector representations are organized into indexes where we can quickly search for the pieces of information that most closely resembles (from a semantic perspective) a given question or query. For example, if I take news articles from this year, vectorize them, and add them to an index, I can quickly search for pieces of information about the US elections.<p>To help achieve this, the Neum AI framework features:<p>Starting with built-in data connectors for common data sources, embedding services and vector stores, the framework provides modularity to build data pipelines to your specification.<p>The connectors support pre-processing capabilities to define loading, chunking and selecting strategies to optimize content to be embedded. This also includes extracting metadata that is going to be associated to a given vector.<p>The generated pipelines support large scale jobs through a high throughput distributed architecture. The connectors allow you to parallelize tasks like downloading documents, processing them, generating embedding and ingesting data into the vector DB.<p>For data sources that might be continuously changing, the framework supports data scheduling and synchronization. This includes delta syncs where only new data is pulled.<p>Once data is transformed into a vector database, the framework supports querying of the data including hybrid search using the available metadata added during pre-processing. As part of the querying process, the framework provides capabilities to capture feedback on retrieved data as well as run evaluations against different pipeline configurations.<p>Try it out and if interested in chatting more about this shoot us an email founders@tryneum.com
Show HN: Neum AI – Open-source large-scale RAG framework
Over the last couple months we have been supporting developers in building large-scale RAG pipelines to process millions of pieces of data.<p>We documented our approach in an HN post (<a href="https://news.ycombinator.com/item?id=37824547">https://news.ycombinator.com/item?id=37824547</a>) a couple weeks ago. Today, we are open sourcing the framework we have developed.<p>The framework focuses on RAG data pipelines and provides scale, reliability, and data synchronization capabilities out of the box.<p>For those newer to RAG, it is a technique to provide context to Large Language Models. It consists of grabbing pieces of information (i.e. pieces of news articles, papers, descriptions, etc.) and incorporating them into prompts to help contextualize the responses. The technique goes one level deeper in finding the right pieces of information to incorporate. The search for relevant information is done through the use of vector embeddings and vector databases.<p>Those pieces of news articles, papers, etc. are transformed into a vector embedding that represents the semantic meaning of the information. These vector representations are organized into indexes where we can quickly search for the pieces of information that most closely resembles (from a semantic perspective) a given question or query. For example, if I take news articles from this year, vectorize them, and add them to an index, I can quickly search for pieces of information about the US elections.<p>To help achieve this, the Neum AI framework features:<p>Starting with built-in data connectors for common data sources, embedding services and vector stores, the framework provides modularity to build data pipelines to your specification.<p>The connectors support pre-processing capabilities to define loading, chunking and selecting strategies to optimize content to be embedded. This also includes extracting metadata that is going to be associated to a given vector.<p>The generated pipelines support large scale jobs through a high throughput distributed architecture. The connectors allow you to parallelize tasks like downloading documents, processing them, generating embedding and ingesting data into the vector DB.<p>For data sources that might be continuously changing, the framework supports data scheduling and synchronization. This includes delta syncs where only new data is pulled.<p>Once data is transformed into a vector database, the framework supports querying of the data including hybrid search using the available metadata added during pre-processing. As part of the querying process, the framework provides capabilities to capture feedback on retrieved data as well as run evaluations against different pipeline configurations.<p>Try it out and if interested in chatting more about this shoot us an email founders@tryneum.com
Show HN: Neum AI – Open-source large-scale RAG framework
Over the last couple months we have been supporting developers in building large-scale RAG pipelines to process millions of pieces of data.<p>We documented our approach in an HN post (<a href="https://news.ycombinator.com/item?id=37824547">https://news.ycombinator.com/item?id=37824547</a>) a couple weeks ago. Today, we are open sourcing the framework we have developed.<p>The framework focuses on RAG data pipelines and provides scale, reliability, and data synchronization capabilities out of the box.<p>For those newer to RAG, it is a technique to provide context to Large Language Models. It consists of grabbing pieces of information (i.e. pieces of news articles, papers, descriptions, etc.) and incorporating them into prompts to help contextualize the responses. The technique goes one level deeper in finding the right pieces of information to incorporate. The search for relevant information is done through the use of vector embeddings and vector databases.<p>Those pieces of news articles, papers, etc. are transformed into a vector embedding that represents the semantic meaning of the information. These vector representations are organized into indexes where we can quickly search for the pieces of information that most closely resembles (from a semantic perspective) a given question or query. For example, if I take news articles from this year, vectorize them, and add them to an index, I can quickly search for pieces of information about the US elections.<p>To help achieve this, the Neum AI framework features:<p>Starting with built-in data connectors for common data sources, embedding services and vector stores, the framework provides modularity to build data pipelines to your specification.<p>The connectors support pre-processing capabilities to define loading, chunking and selecting strategies to optimize content to be embedded. This also includes extracting metadata that is going to be associated to a given vector.<p>The generated pipelines support large scale jobs through a high throughput distributed architecture. The connectors allow you to parallelize tasks like downloading documents, processing them, generating embedding and ingesting data into the vector DB.<p>For data sources that might be continuously changing, the framework supports data scheduling and synchronization. This includes delta syncs where only new data is pulled.<p>Once data is transformed into a vector database, the framework supports querying of the data including hybrid search using the available metadata added during pre-processing. As part of the querying process, the framework provides capabilities to capture feedback on retrieved data as well as run evaluations against different pipeline configurations.<p>Try it out and if interested in chatting more about this shoot us an email founders@tryneum.com
Show HN: New visual language for teaching kids to code
Pickcode is a new language and editor for getting kids started with coding. The code editing experience is totally structured, where you select choices from menus rather than typing.
I made Pickcode after experiences teaching kids both block coding (Scratch, App Inventor) and Python. To me, block coding is too far removed from regular coding for kids to make the connection. Pickcode provides a much clearer transition path for students to Python/JS/Java. Our target market is middle/early high school kids, and that’s who we’ve tested the product with during development.<p>On the site, you can do tutorials to make chatbots, animated drawings, and 2D games. We have a full Intro to Pickcode course, as well as an Intro to Python course where you make regular console programs with a regular text editor. There are 30 or so free lessons accessible with an account, and the rest are paywalled for $5/month.<p>For professional programmers, the editor is probably pretty frustrating to use (no vim keybindings!), but I hope it’s at least interesting to play with from a UI perspective. If you have kids aged 10-14, I’d love any feedback you have from trying it out with them. I love talking to users, reach out at charlie@pickcode.io!
Show HN: New visual language for teaching kids to code
Pickcode is a new language and editor for getting kids started with coding. The code editing experience is totally structured, where you select choices from menus rather than typing.
I made Pickcode after experiences teaching kids both block coding (Scratch, App Inventor) and Python. To me, block coding is too far removed from regular coding for kids to make the connection. Pickcode provides a much clearer transition path for students to Python/JS/Java. Our target market is middle/early high school kids, and that’s who we’ve tested the product with during development.<p>On the site, you can do tutorials to make chatbots, animated drawings, and 2D games. We have a full Intro to Pickcode course, as well as an Intro to Python course where you make regular console programs with a regular text editor. There are 30 or so free lessons accessible with an account, and the rest are paywalled for $5/month.<p>For professional programmers, the editor is probably pretty frustrating to use (no vim keybindings!), but I hope it’s at least interesting to play with from a UI perspective. If you have kids aged 10-14, I’d love any feedback you have from trying it out with them. I love talking to users, reach out at charlie@pickcode.io!
Show HN: New visual language for teaching kids to code
Pickcode is a new language and editor for getting kids started with coding. The code editing experience is totally structured, where you select choices from menus rather than typing.
I made Pickcode after experiences teaching kids both block coding (Scratch, App Inventor) and Python. To me, block coding is too far removed from regular coding for kids to make the connection. Pickcode provides a much clearer transition path for students to Python/JS/Java. Our target market is middle/early high school kids, and that’s who we’ve tested the product with during development.<p>On the site, you can do tutorials to make chatbots, animated drawings, and 2D games. We have a full Intro to Pickcode course, as well as an Intro to Python course where you make regular console programs with a regular text editor. There are 30 or so free lessons accessible with an account, and the rest are paywalled for $5/month.<p>For professional programmers, the editor is probably pretty frustrating to use (no vim keybindings!), but I hope it’s at least interesting to play with from a UI perspective. If you have kids aged 10-14, I’d love any feedback you have from trying it out with them. I love talking to users, reach out at charlie@pickcode.io!
Show HN: New visual language for teaching kids to code
Pickcode is a new language and editor for getting kids started with coding. The code editing experience is totally structured, where you select choices from menus rather than typing.
I made Pickcode after experiences teaching kids both block coding (Scratch, App Inventor) and Python. To me, block coding is too far removed from regular coding for kids to make the connection. Pickcode provides a much clearer transition path for students to Python/JS/Java. Our target market is middle/early high school kids, and that’s who we’ve tested the product with during development.<p>On the site, you can do tutorials to make chatbots, animated drawings, and 2D games. We have a full Intro to Pickcode course, as well as an Intro to Python course where you make regular console programs with a regular text editor. There are 30 or so free lessons accessible with an account, and the rest are paywalled for $5/month.<p>For professional programmers, the editor is probably pretty frustrating to use (no vim keybindings!), but I hope it’s at least interesting to play with from a UI perspective. If you have kids aged 10-14, I’d love any feedback you have from trying it out with them. I love talking to users, reach out at charlie@pickcode.io!
Show HN: Hotlist, a Tauri powered daily todo list for your Mac menu bar
Show HN: Ending the Absurdity of SSO Tax – Introducing ssotax.org
Ever heard of the SSO Tax? In short, it's a tactic where software vendors bully security-conscious companies to upgrade to costly enterprise plans. They do this by gating SSO (Single Sign-On) features behind their priciest options, causing companies to pay up to 70 times their standard rates.<p>As a former CTO at a VC-backed and security-conscious company, I've faced the tough choice of skipping costly enterprise upgrades, even when SSO was crucial.<p>Take a look at Notion: to access SSO, they casually double their standard pricing.<p>Imagine buying a Tesla and being charged extra to unlock full braking power. That's what SSO Tax is - vendors exploiting a built-in feature, essential for security, to extract excessive fees.<p>So, why initiate a new project?<p>Rob Chahin's work on sso.tax initially highlighted this issue. However, the site's updates dwindled, and data became outdated. Despite offering assistance, I received no response, leading to the creation of <a href="https://ssotax.org" rel="nofollow noreferrer">https://ssotax.org</a>. While there has been short spike of activity post-fork, it already stopped again. That’s what we’ve seen often in the last few years. Instead, I want to give the topic the attention it deserves.<p>In addition of integrating all pending PRs and enriching the data, I’ve introduced a new feature: "Friends of SSO". We should not only call out unfair practices but also praise vendors who are committed to security!<p>Furthermore, I’d love to raise awareness about vendor practices by utilizing Twitter and Linkedin to publicly praise or critique them. The goal is to get attention for the topic, ideally sparking conversation with the vendors involved.<p>What are your thoughts on getting rid of the SSO Tax? Excited to hear your ideas!