The best Hacker News stories from Show from the past day
Latest posts:
Show HN: Finetune LLaMA-7B on commodity GPUs using your own text
I've been playing around with <a href="https://github.com/zphang/minimal-llama/">https://github.com/zphang/minimal-llama/</a> and <a href="https://github.com/tloen/alpaca-lora/blob/main/finetune.py">https://github.com/tloen/alpaca-lora/blob/main/finetune.py</a>, and wanted to create a simple UI where you can just paste text, tweak the parameters, and finetune the model quickly using a modern GPU.<p>To prepare the data, simply separate your text with two blank lines.<p>There's an inference tab, so you can test how the tuned model behaves.<p>This is my first foray into the world of LLM finetuning, Python, Torch, Transformers, LoRA, PEFT, and Gradio.<p>Enjoy!
Show HN: ChatLLaMA – A ChatGPT style chatbot for Facebook's LLaMA
ChatLLaMA is an experimental chatbot interface for interacting with variants of Facebook's LLaMA. Currently, we support the 7 billion parameter variant that was fine-tuned on the Alpaca dataset. This early versions isn't as conversational as we'd like, but over the next week or so, we're planning on adding support for the 30 billion parameter variant, another variant fine-tuned on LAION's OpenAssistant dataset and more as we explore what this model is capable of.<p>If you want deploy your own instance is the model powering the chatbot and build something similar we've open sourced the Truss here: <a href="https://github.com/basetenlabs/alpaca-7b-truss">https://github.com/basetenlabs/alpaca-7b-truss</a><p>We'd love to hear any feedback you have. You can reach me on Twitter @aaronrelph or Abu (the engineer behind this) @aqaderb.<p>Disclaimer: We both work at Baseten. This was a weekend project. Not trying to shill anything; just want to build and share cool stuff.
Show HN: ChatLLaMA – A ChatGPT style chatbot for Facebook's LLaMA
ChatLLaMA is an experimental chatbot interface for interacting with variants of Facebook's LLaMA. Currently, we support the 7 billion parameter variant that was fine-tuned on the Alpaca dataset. This early versions isn't as conversational as we'd like, but over the next week or so, we're planning on adding support for the 30 billion parameter variant, another variant fine-tuned on LAION's OpenAssistant dataset and more as we explore what this model is capable of.<p>If you want deploy your own instance is the model powering the chatbot and build something similar we've open sourced the Truss here: <a href="https://github.com/basetenlabs/alpaca-7b-truss">https://github.com/basetenlabs/alpaca-7b-truss</a><p>We'd love to hear any feedback you have. You can reach me on Twitter @aaronrelph or Abu (the engineer behind this) @aqaderb.<p>Disclaimer: We both work at Baseten. This was a weekend project. Not trying to shill anything; just want to build and share cool stuff.
Show HN: ChatLLaMA – A ChatGPT style chatbot for Facebook's LLaMA
ChatLLaMA is an experimental chatbot interface for interacting with variants of Facebook's LLaMA. Currently, we support the 7 billion parameter variant that was fine-tuned on the Alpaca dataset. This early versions isn't as conversational as we'd like, but over the next week or so, we're planning on adding support for the 30 billion parameter variant, another variant fine-tuned on LAION's OpenAssistant dataset and more as we explore what this model is capable of.<p>If you want deploy your own instance is the model powering the chatbot and build something similar we've open sourced the Truss here: <a href="https://github.com/basetenlabs/alpaca-7b-truss">https://github.com/basetenlabs/alpaca-7b-truss</a><p>We'd love to hear any feedback you have. You can reach me on Twitter @aaronrelph or Abu (the engineer behind this) @aqaderb.<p>Disclaimer: We both work at Baseten. This was a weekend project. Not trying to shill anything; just want to build and share cool stuff.
Show HN: Pair: Open Tool for Coding with GPTs, Built by Coding with GPTs
Github Copilot is a great tool for leveraging GPTs while coding, but I find that it is too “open loop” for more complex tasks that require Q&A, feedback to guide it in a particular direction, iteration on code execution errors, etc. There is a large class of tasks that are better accomplished in an iterative, stateful chat-like interface.<p>I have been experimenting with a local command line chat interface to GPT-4 and my mind was blown once again a few days ago when I copied documentation for a pretty involved API into the model context and managed to chat-guide GPT-4 to implement the API in under 30 minutes, complete with a ridiculous amount of unit test coverage.<p>This involved a lot of manual copy and pasting back and forth and other friction points that could be removed by a streamlined REPL interface optimized for code interactions. It occurred to me that it would be fun to build such a tool, and as the ultimate act of dogfooding, try to build it with GPT!<p>So PAIR is the starting point here. You can see a recent commit message has a log of my interactions with the model that produced that commit.<p>Next step is to add better mechanisms to manage the model input context (e.g. make it easy for the model to see the latest version of a source file when needed) followed by mechanisms for allowing the model to suggest changes via diffs that are quickly reviewed and accepted by the human in the loop before being applied to the file and tested.<p>I would love to hear from others who have experimented with GPT pair programming in a chat-style interface and any feedback you might have on your experience with it.
Show HN: Watermelon – GPT-powered code contextualizer
Hey there HN! We're Esteban and Esteban and we are looking to get feedback for the new version of our GPT-powered, open-source code contextualizer.<p>We're starting with a VS Code extension that indexes information from git (GitHub, GitLab, or Bitbucket integrations available), Slack and Jira to explain the context around a file or block of code. Finally, we summarize such aggregated context using the power of GPT.<p>As devs we know that it's very annoying to look at a new codebase and start understanding all the nuances, particularly when the person who wrote the code already left the company. With this problem in mind, we decided to build this solution. You'll be able to get into "the ghost" of the person who left the company.<p>Soon, we will also be building a GitHub Action that does the same thing as the VS Code extension but at the time of creating a PR: Index the most relevant information related to this new PR, and add it as a comment. This way we will provide context at one more moment, and also, we will be making the IDE extension better.<p>Here's our open source repo if you also want to check it out: <a href="https://github.com/watermelontools/watermelon-extension">https://github.com/watermelontools/watermelon-extension</a><p>Please give us your feedback! Thanks.
Show HN: Watermelon – GPT-powered code contextualizer
Hey there HN! We're Esteban and Esteban and we are looking to get feedback for the new version of our GPT-powered, open-source code contextualizer.<p>We're starting with a VS Code extension that indexes information from git (GitHub, GitLab, or Bitbucket integrations available), Slack and Jira to explain the context around a file or block of code. Finally, we summarize such aggregated context using the power of GPT.<p>As devs we know that it's very annoying to look at a new codebase and start understanding all the nuances, particularly when the person who wrote the code already left the company. With this problem in mind, we decided to build this solution. You'll be able to get into "the ghost" of the person who left the company.<p>Soon, we will also be building a GitHub Action that does the same thing as the VS Code extension but at the time of creating a PR: Index the most relevant information related to this new PR, and add it as a comment. This way we will provide context at one more moment, and also, we will be making the IDE extension better.<p>Here's our open source repo if you also want to check it out: <a href="https://github.com/watermelontools/watermelon-extension">https://github.com/watermelontools/watermelon-extension</a><p>Please give us your feedback! Thanks.
Show HN: Public transportation signage based on bloom filters (rough mockup)
Hello,
I was running around Germany, hectically navigating public transportation, and getting lost all the time. I noticed that every station had i platforms, each used lists of n buses (trains, whatever) arriving, each has their list of m destinations. That means I would be scanning i x n x m items just to see if I was at the correct stop. As I was nervous, for every bus that arrived, I would rescan the list of stops to double check. I began thinking how I could make a better system.<p>Linked is a very shoddy mockup of how bloom filters could be used to allow passengers O(1) lookup time for which platform+bus is the correct one. I believe it's likely for public transportation to grow increasingly more complex in the future, as population grows, and under the current list-based system, this will make the signage ever more complex. I think some bloom filter mechanism could reduce that complexity.<p>So, here is my fantasy, my day dream. What do you think?
Show HN: Public transportation signage based on bloom filters (rough mockup)
Hello,
I was running around Germany, hectically navigating public transportation, and getting lost all the time. I noticed that every station had i platforms, each used lists of n buses (trains, whatever) arriving, each has their list of m destinations. That means I would be scanning i x n x m items just to see if I was at the correct stop. As I was nervous, for every bus that arrived, I would rescan the list of stops to double check. I began thinking how I could make a better system.<p>Linked is a very shoddy mockup of how bloom filters could be used to allow passengers O(1) lookup time for which platform+bus is the correct one. I believe it's likely for public transportation to grow increasingly more complex in the future, as population grows, and under the current list-based system, this will make the signage ever more complex. I think some bloom filter mechanism could reduce that complexity.<p>So, here is my fantasy, my day dream. What do you think?
Show HN: Public transportation signage based on bloom filters (rough mockup)
Hello,
I was running around Germany, hectically navigating public transportation, and getting lost all the time. I noticed that every station had i platforms, each used lists of n buses (trains, whatever) arriving, each has their list of m destinations. That means I would be scanning i x n x m items just to see if I was at the correct stop. As I was nervous, for every bus that arrived, I would rescan the list of stops to double check. I began thinking how I could make a better system.<p>Linked is a very shoddy mockup of how bloom filters could be used to allow passengers O(1) lookup time for which platform+bus is the correct one. I believe it's likely for public transportation to grow increasingly more complex in the future, as population grows, and under the current list-based system, this will make the signage ever more complex. I think some bloom filter mechanism could reduce that complexity.<p>So, here is my fantasy, my day dream. What do you think?
Show HN: Professional headshots for remote team with AI
Show HN: Professional headshots for remote team with AI
Show HN: Get a Professional Headshot in Minutes with AI
After playing with AI Avatars (like many of us I guess around here), I started to wonder if we could instead bring real value to people by producing affordable professional head-shots using a combination of Dreambooth and ControlNet.<p>Obviously it's only the beginning and there are still many imperfections, but the foundational tech behind this (Dreambooth and ControlNet) are only respectively 6 months and 1.5 month old, and already delivers pretty amazing results.<p>I came up with this little service "Virtual Face" and I'm looking for feedback if some of you are willing to try it (you can use the HUNTER50 coupon to get 50% off, can't make it free to try yet since the running costs are still non-negligible).<p>Cheers,
Pierre
Show HN: Get a Professional Headshot in Minutes with AI
After playing with AI Avatars (like many of us I guess around here), I started to wonder if we could instead bring real value to people by producing affordable professional head-shots using a combination of Dreambooth and ControlNet.<p>Obviously it's only the beginning and there are still many imperfections, but the foundational tech behind this (Dreambooth and ControlNet) are only respectively 6 months and 1.5 month old, and already delivers pretty amazing results.<p>I came up with this little service "Virtual Face" and I'm looking for feedback if some of you are willing to try it (you can use the HUNTER50 coupon to get 50% off, can't make it free to try yet since the running costs are still non-negligible).<p>Cheers,
Pierre
Show HN: Leetcode but for front end engineers. Bad idea?
Show HN: Find words “halfway” between two others
Show HN: Find words “halfway” between two others
Show HN: Great Books Homeschool beta program
I built this customizable literature-based K-12 homeschool curriculum, based on my experience as a homeschool parent. It's designed especially for intellectually curious kids who love to read.<p>One of the main benefits of homeschooling is the ability to design customized programs of study that let kids learn at their level of challenge in each subject. But since designing custom curricula from scratch requires a huge time commitment and familiarity with children's literature and academic materials, most homeschooling parents don't take advantage of this potential and instead opt for prepackaged curricula.<p>Great Books Homeschool eliminates a lot of the work involved in designing a complete and rigorous curriculum for homeschooled students. The website generates a default program of study for each student, then helps parents customize it. Transcripts and other records are generated automatically.<p>Pricing is normally subscription based, but we're offering complimentary access for twelve months to the first 50 users who sign up for our beta testing program. In return, beta testers are requested to complete a monthly questionnaire about their experience with the curriculum.<p>If you would like to participate in the beta testing program, please first create a free trial account at <a href="https://www.greatbookshomeschool.com" rel="nofollow">https://www.greatbookshomeschool.com</a>. Once signed in, go to <a href="https://www.greatbookshomeschool.com/parent/beta-application?via=hn" rel="nofollow">https://www.greatbookshomeschool.com/parent/beta-application...</a> and complete the application form.<p>Questions and comments are welcome!
Show HN: Great Books Homeschool beta program
I built this customizable literature-based K-12 homeschool curriculum, based on my experience as a homeschool parent. It's designed especially for intellectually curious kids who love to read.<p>One of the main benefits of homeschooling is the ability to design customized programs of study that let kids learn at their level of challenge in each subject. But since designing custom curricula from scratch requires a huge time commitment and familiarity with children's literature and academic materials, most homeschooling parents don't take advantage of this potential and instead opt for prepackaged curricula.<p>Great Books Homeschool eliminates a lot of the work involved in designing a complete and rigorous curriculum for homeschooled students. The website generates a default program of study for each student, then helps parents customize it. Transcripts and other records are generated automatically.<p>Pricing is normally subscription based, but we're offering complimentary access for twelve months to the first 50 users who sign up for our beta testing program. In return, beta testers are requested to complete a monthly questionnaire about their experience with the curriculum.<p>If you would like to participate in the beta testing program, please first create a free trial account at <a href="https://www.greatbookshomeschool.com" rel="nofollow">https://www.greatbookshomeschool.com</a>. Once signed in, go to <a href="https://www.greatbookshomeschool.com/parent/beta-application?via=hn" rel="nofollow">https://www.greatbookshomeschool.com/parent/beta-application...</a> and complete the application form.<p>Questions and comments are welcome!
Show HN: Recursive LLM Prompts
I've been playing with the idea of an LLM prompt that causes the model to generate and return a new prompt. <a href="https://github.com/andyk/recursive_llm">https://github.com/andyk/recursive_llm</a><p>The idea I'm starting with is to implement recursion using English as the programming language and GPT as the runtime.<p>It’s kind of like traditional recursion in code, but instead of having a function that calls itself with a different set of arguments, there is a prompt that returns itself with specific parts updated to reflect the new arguments.<p>Here is a prompt for infinitely generating Fibonacci numbers:<p>> You are a recursive function. Instead of being written in a programming language, you are written in English. You have variables FIB_INDEX = 2, MINUS_TWO = 0, MINUS_ONE = 1, CURR_VALUE = 1. Output this paragraph but with updated variables to compute the next step of the Fibbonaci sequence.<p>Interestingly, I found that to get a base case to work I had to add quite a bit more text (i.e. the prompt I arrived at is more than twice as long <a href="https://raw.githubusercontent.com/andyk/recursive_llm/main/prompt_fibonnaci_include_math.txt" rel="nofollow">https://raw.githubusercontent.com/andyk/recursive_llm/main/p...</a>)