Show HN: AI Chatbot for WordPress https://bit.ly/3Pbho3k

Show HN: AI Chatbot for WordPress https://bit.ly/3Pbhp7o September 7, 2023 at 01:28PM

Show HN: Security Compliance in Context (I am starting a side project) https://bit.ly/460kHRN

Show HN: Security Compliance in Context (I am starting a side project) https://bit.ly/44CxXet September 7, 2023 at 01:07PM

Show HN: All visitors pointers on a webpage (How-to) https://bit.ly/44KrHkC

Show HN: All visitors pointers on a webpage (How-to) https://bit.ly/44NITpn September 7, 2023 at 12:26PM

Show HN: Curated custom search engine portal https://bit.ly/44L1S45

Show HN: Curated custom search engine portal Hey friends, a little background - A friend of mine who’s also a newsletter writer once told me that when he’s interested in certain topic he would like to know the opinion of some of his favorite writers on the same topic. He would use Google to search something like discomfort site:sive.rs, and then expand his search across various authors he admires. It turns out that Google’s programmable search engine does exactly that. By simply adding a few websites (much like subscribing to newsletters or RSS feeds), we can craft a search engine tailored for our specific needs. I think that's a great idea - more signal and less noise. So I built a portal to facilitate sharing personal curations https://bit.ly/3L9umgL . Thanks for reading, lemme know what you think :D https://bit.ly/3L9umgL September 7, 2023 at 06:49AM

Show HN: Formstr: An open source and decentralized alternative to Google Forms https://bit.ly/3Rfgupd

Show HN: Formstr: An open source and decentralized alternative to Google Forms https://bit.ly/3sFqhux September 7, 2023 at 04:24AM

Show HN: uDSV.js – A faster CSV parser https://bit.ly/3PsrLAf

Show HN: uDSV.js – A faster CSV parser Hey folks! I know CSV parsers (especially in JS) aren't terribly exciting and someone writes a "better" one every week. I'm in the middle of my parental leave, and this was a project that came out of me looking for the fastest/smallest CSV parser. It all started so innocently, and then turned into a benchmark-validation-athon; the library itself took ~2 weeks to write, but the performance comparisons took another ~4 weeks (on and off). The benchmarks were a huge effort, but I think they are the most thorough to date, both in breadth and in depth, so hopefully you find them useful: https://bit.ly/3rcacfk Let me know if you have specific concerns / questions / improvements :) cheers! Leon https://bit.ly/48gnXKW September 4, 2023 at 05:04PM

Show HN: A better way to read blogs https://bit.ly/3RawBEH

Show HN: A better way to read blogs https://bit.ly/3R4gFUm September 6, 2023 at 12:12PM

Show HN: Automated Pull Request Reviews https://bit.ly/45FcAdM

Show HN: Automated Pull Request Reviews https://bit.ly/3sDeTPH September 6, 2023 at 06:42AM

Show HN: I built an extension that never lets you overpay for a book again https://bit.ly/45YZ849

Show HN: I built an extension that never lets you overpay for a book again https://bit.ly/461jxFA September 6, 2023 at 02:40AM

Show HN: ColorMood https://bit.ly/3P6Wwu7

Show HN: ColorMood Does your mood affect which color you like - a tool that attempts to find your favourite color right now https://bit.ly/3P6WwKD September 6, 2023 at 05:48AM

Show HN: Trellis – open-source Python framework to build DAG-based LLM workflows https://bit.ly/44Hheq8

Show HN: Trellis – open-source Python framework to build DAG-based LLM workflows Hey HN! Trellis is an open-source framework for programmatically orchestrating LLM workflows as Directed Acyclic Graphs (DAGs) in Python. My friend and I started working on this a few weeks ago after we tried building applications using mainstream LLM frameworks, and faced all the common complaints (too abstracted, hard to customize, bad docs/support). After talking to a few other people building with LLMs, we also noticed that these frameworks were not inherently built to support DAG-based LLM workflows. We designed Trellis to be as minimal and flat as possible, so developers can have lower level control over their DAGs. Trellis is composed of only three abstractions: Node, DAG, and LLM. Node: the atomic unit of Trellis. Nodes are chained together to form a DAG. Node is an abstract class with only one method required to implement. DAG: a directed acyclic graph of Nodes. It is the primary abstraction for orchestrating LLM workflows. When you add edges between Nodes, you can specify a transformation function to reuse Nodes and connect any two Nodes. Trellis verifies the data flowing between Nodes in a DAG to ensure the flow of data is validated. LLM: a wrapper around a large language model with simple catches for common OpenAI errors. Currently, the only provider that Trellis supports is OpenAI. Check out our docs if this sounds interesting: https://bit.ly/3EsVio1... We'd love it if you tried hacking with it and give us any feedback you have! :) https://bit.ly/3sHq4XJ September 6, 2023 at 03:04AM

Show HN: Fully client-side GPT2 prediction visualizer https://bit.ly/45TSJHe

Show HN: Fully client-side GPT2 prediction visualizer https://bit.ly/45HJM4g September 5, 2023 at 11:42PM

Show HN: Simple passwordless authentication for your website https://bit.ly/45APSU3

Show HN: Simple passwordless authentication for your website Solo founder here - built a passwordless authentication service after getting frustrated with the very high pricing and lack of customizability, easy passwordless authentication on existing solutions. Check it out here and let me know what you think! https://bit.ly/45FSkbS September 5, 2023 at 08:47AM

Show HN: Chalk.ist – Create beautiful images of your source code https://bit.ly/3R9hL1c

Show HN: Chalk.ist – Create beautiful images of your source code https://bit.ly/3MevSwk September 5, 2023 at 07:54AM

Show HN: Subsidian – Visualize a Substack archive in Obsidian graph view https://bit.ly/45TuclN

Show HN: Subsidian – Visualize a Substack archive in Obsidian graph view https://bit.ly/3Z66TCU September 4, 2023 at 03:08PM

Show HN: Keep – GitHub Actions for your monitoring tools https://bit.ly/48e39DV

Show HN: Keep – GitHub Actions for your monitoring tools Hi Hacker News! Shahar and Tal from Keep here. A few months ago, we introduced here at HN ( https://bit.ly/3EcwLE1 ) Keep as an “open source alerting CLI” and got some interesting feedback - mainly around UI, automation, and supporting more tools. We were VERY early back then, and we understood that although the current DX around creating alerts is not great, it's not that critical and developers don’t need another tool just for that. But we did find something else. While talking to developers and devops, we found that a lot of companies use many tools that generate alerts - from Cloudwatch, Prometheus, Grafana, and Datadog to tools such as Zabbix or Nagios. We definitely agree consolidation in the observability space is a real thing, but while talking to those companies we feel that there are still real use cases for having more than one tool (and for example, according to Grafana’s 2023 observability survey, 52% of the companies uses more than 6 observability tools https://bit.ly/47Ysbqh ). So we that in mind, we rebuilt Keep with a simple mindset: (1) Integrate with every tool that triggers alerts - it can be either pushing alerts to Keep via webhooks or routing policies or Keep to pull alerts via the tools API. (2) Create a simple abstraction layer to run workflows on top of these alerts. (3) Maintain a great developer experience - open source, API-first, workflows as code and generally having a developer mindset while building Keep. During the time we rebuilt Keep, Datadog released their workflow automation tool ( https://bit.ly/3L75FBC ) which led us to the understanding that's exactly what we solve - but for everyone who uses tools other than Datadog. A short demo of Keep with a simple use case: https://www.youtube.com/watch?v=FPMRCZM8ZYg You can try it yourself by signing into https://bit.ly/3Nipgk6 Like always - we invite you to try Keep and we are eager to hear any feedback. https://bit.ly/3IrvGuF September 4, 2023 at 04:15PM

Show HN: Recognize license plates using fine-tuned yolov8, OCR and IP camera https://bit.ly/46fACfp

Show HN: Recognize license plates using fine-tuned yolov8, OCR and IP camera Hey, just a work related project I made, which could be open sourced :D If you're looking for an example on how to use/fine-tune yolov8, I feel like taking a look at this repo and reading the README could help you get up to speed (also linked some nice refs)! This is actually a full rewrite of a proprietary project I made (and documented on my site) like a year ago, will do some finishing touches (write blog post about it, mark the old version deprecated, record a tutorial on how to set it up on an Ubuntu server, etc, etc) in the following month, but felt like sharing it now, cuz I consider it done The only proprietary part is the client, which receives the images and does stuff with db (has to interact with internal APIs, so there's no reason to make it oss anyways). Also, the client contains only the business logic, all of the fun ai/web server stuff is fully open under AGPL-3.0 (and an example client without the business logic is available ... in rust btw xdd). https://bit.ly/3qYiJme September 4, 2023 at 08:56PM

Show HN: TTop – System monitoring tool with historical data, triggers and TUI https://bit.ly/3qZ9Ngl

Show HN: TTop – System monitoring tool with historical data, triggers and TUI It is not top/htop replacement because of historical snapshots which can help you to find problems back in time https://bit.ly/44G50y9 September 4, 2023 at 04:50PM

Show HN: finetune LLMs via the Finetuning Hub https://bit.ly/3PnIQMB

Show HN: finetune LLMs via the Finetuning Hub Hi HN community, I have been working on benchmarking publicly available LLMs these past couple of weeks. More precisely, I am interested on the finetuning piece since a lot of businesses are starting to entertain the idea of self-hosting LLMs trained on their proprietary data rather than relying on third party APIs. To this point, I am tracking the following 4 pillars of evaluation that businesses are typically look into: - Performance - Time to train an LLM - Cost to train an LLM - Inference (throughput / latency / cost per token) For each LLM, my aim is to benchmark them for popular tasks, i.e., classification and summarization. Moreover, I would like to compare them against each other. So far, I have benchmarked Flan-T5-Large, Falcon-7B and RedPajama and have found them to be very efficient in low-data situations, i.e., when there are very few annotated samples. Llama2-7B/13B and Writer’s Palmyra are in the pipeline. But there’s so many LLMs out there! In case this work interests you, would be great to join forces. GitHub repo attached — feedback is always welcome :) Happy hacking! https://bit.ly/3ElU8uG September 4, 2023 at 04:16PM

Show HN: Rapidgzip – Truly Parallel Gzip Decompression with 10 GB/s https://bit.ly/3PoJJVi

Show HN: Rapidgzip – Truly Parallel Gzip Decompression with 10 GB/s I have posted a much earlier version of this over a year ago [0]. Since then a lot has changed. Obviously, the name has changed. This happened for the paper publication [1]. I have also optimized the speed, integrated ISA-L for special cases, limited the compression-ratio-dependent maximum memory consumption, and finally added parallelized CRC32 computation, which adds ~5% overhead no matter the number of cores used. At this point, I am leaning towards calling it production-ready although there are still many ideas for improvements. Redoing the benchmarks of the older Show HN, would look like this: time pigz -d -c 4GiB-base64.gz | wc -c # real ~13.4 s -> ~320 MB/s time rapidgzip -d -c 4GiB-base64.gz | wc -c # real ~1.26 s -> ~3.4 GB/s However, at this point, the piping itself becomes a problem. Rapidgzip is actually slightly faster than cat when comparing the piped bandwidth! E.g., compare these additional benchmarks: time cat 4GiB-base64.gz | wc -c # real ~1.06 s -> ~3.1 GB/s time fcat 4GiB-base64.gz | wc -c # real ~0.41 s -> ~8.0 GB/s time rapidgzip -o /dev/null -d 4GiB-base64.gz # real ~0.68 s -> ~6.5 GB/s fcat is an alternative cat implementation that uses vmsplice to speed up piping. According to the ReadMe it currently is broken, but it works fine on my system and piping it to md5sum yields consistent results [2]. So, at this point, I/O and actually also allocations have become a limiting factor and if you want full speed, you would have to interface with the rapidgzip library interface directly (in C++ or via the Python bindings) and process the decompressed data in memory. The project ReadMe contains further benchmarks with Silesia and FASTQ data and scaling up to 128 cores, for which rapidgzip achieves 12 GB/s for Silesia and 24 GB/s when an index has been created with --export-index and is used with --import-index. It can also be tested with ratarmount 0.14.0, which now uses rapidgzip as a backend by default for .gz and .tar.gz files [3]. [0] https://bit.ly/3P3MRT2 [1] https://bit.ly/3Evkvhz [2] https://bit.ly/3EItoot [3] https://bit.ly/3hZj1Ba https://bit.ly/3qX04Hi September 4, 2023 at 09:29AM