<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Joe Simo</title>
    <description>The latest articles on DEV Community by Joe Simo (@joesimo).</description>
    <link>https://dev.to/joesimo</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4205555%2F0a5e9360-de08-40f5-9691-a561cd9abe39.png</url>
      <title>DEV Community: Joe Simo</title>
      <link>https://dev.to/joesimo</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/joesimo"/>
    <language>en</language>
    <item>
      <title>Last Commit: an ofrenda for the open source that went quiet</title>
      <dc:creator>Joe Simo</dc:creator>
      <pubDate>Mon, 26 Oct 2026 19:54:51 +0000</pubDate>
      <link>https://dev.to/joesimo/last-commit-an-ofrenda-for-the-open-source-that-went-quiet-1gle</link>
      <guid>https://dev.to/joesimo/last-commit-an-ofrenda-for-the-open-source-that-went-quiet-1gle</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week4-2026-10-26"&gt;Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;Every developer has a few ghosts in their &lt;code&gt;package.json&lt;/code&gt;. A library you once leaned on every day, still installed somewhere, whose repository has not moved in years. Nobody announces when a project dies. The commits just stop, and one day you notice that the last thing anyone ever said to it was "Change hyperlink".&lt;/p&gt;

&lt;p&gt;For Día de los Muertos, families build an ofrenda: an altar with a photo, the things the person loved, candles, marigolds, so the dead are remembered instead of forgotten. &lt;strong&gt;Last Commit&lt;/strong&gt; builds one for a repository.&lt;/p&gt;

&lt;p&gt;Paste any public GitHub repository and you get a remembrance card:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;The years it was alive&lt;/strong&gt;, and whether it is archived or has simply gone quiet.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Last words&lt;/strong&gt;: the real final commit message written by a person (merges and version bumps are skipped).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tended by&lt;/strong&gt;: the people who kept it alive.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;An epitaph and a remembrance&lt;/strong&gt;, written by an open-weight model running inside your browser tab.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Then you can download the card as a PNG, share a &lt;code&gt;?repo=owner/name&lt;/code&gt; link, and, the part I care about most, fork the project or leave its maintainers a star. Some of these ghosts are still holding up half the internet.&lt;/p&gt;

&lt;p&gt;Here is a real card from the live site. Both the epitaph and the "What it gave us" lines were written by Qwen2.5 0.5B, on the CPU, in headless Chrome:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fshioqzyoldsod0go48aj.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fshioqzyoldsod0go48aj.png" alt="Last Commit card for atom/atom" width="" height=""&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Live:&lt;/strong&gt; &lt;a href="https://last-commit-ofrenda.vercel.app" rel="noopener noreferrer"&gt;https://last-commit-ofrenda.vercel.app&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Narrated demo (47 s):&lt;/strong&gt; &lt;a href="https://last-commit-ofrenda.vercel.app/demo/last-commit-demo.mp4" rel="noopener noreferrer"&gt;https://last-commit-ofrenda.vercel.app/demo/last-commit-demo.mp4&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Click one of the examples (atom, brackets, angular.js, BrowserQuest, co) or paste a repository, then &lt;strong&gt;Let it write&lt;/strong&gt;. No account, no key, no server.&lt;/p&gt;

&lt;p&gt;The card fills in instantly from GitHub facts, before any AI is involved. Then the model writes, word by word, and replaces the fact based lines only if its reply passes the checks:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F07lquz42fxs1bb27rvc4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F07lquz42fxs1bb27rvc4.png" alt="The card while the model is writing" width="800" height="1523"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;These are real runs on the live site on October 26, 2026, using the 0.5B model on the CPU (WebAssembly) path in headless Chrome. The first load downloaded 793 MB of weights and took 54.7 s.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Repository&lt;/th&gt;
&lt;th&gt;Epitaph&lt;/th&gt;
&lt;th&gt;Result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;atom/atom&lt;/td&gt;
&lt;td&gt;It was a hackable text editor, not just a hack, but a true hack, by any standards.&lt;/td&gt;
&lt;td&gt;Model, 23 tokens in 29.7 s&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;adobe/brackets&lt;/td&gt;
&lt;td&gt;It gave 7,429 forks a place to begin, then rested in March 2021.&lt;/td&gt;
&lt;td&gt;Model reply set aside (first person), so the fact based line stayed&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;request/request&lt;/td&gt;
&lt;td&gt;It was here, waiting for its day.&lt;/td&gt;
&lt;td&gt;Model, 10 tokens in 18.0 s&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;And the remembrances:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;atom/atom&lt;/strong&gt; (model, 45 tokens in 49.7 s): "Atom was not just a tool; it was a living, breathing ecosystem of features and libraries. It's like a giant sponge that absorbed all the best parts from other tools, making it even more powerful than ever before."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;adobe/brackets&lt;/strong&gt; (model, 47 tokens in 48.1 s): "Brackets was a powerful tool for developers, its simplicity and elegance have been a testament to its longevity. It stands as a reminder of the dedication and hard work of those who created it, and the enduring legacy of its creator."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;request/request&lt;/strong&gt; (model reply set aside for first person, so this is the fact based text): "Simplified HTTP request client. Across 2011 to 2020, it was tended by many hands, mikeal first among them. 3,135 people carried a copy home, so it lives on somewhere. The last change landed on February 11, 2020."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;atom and brackets come from snapshots bundled with the app. request/request was typed in and fetched live from the GitHub API. Its card, "Gone quiet" rather than archived, last words "README: typo in deprecation message (#3268)":&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwxm1wmms1gi4e0e5x53h.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwxm1wmms1gi4e0e5x53h.png" alt="Last Commit card for request/request" width="800" height="1100"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;It works on a phone too, and when a repository does not exist, it says so kindly:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frxgcfpiit6d6kvg9mpr7.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frxgcfpiit6d6kvg9mpr7.png" alt="Last Commit on a phone" width="780" height="1688"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpajswdci11jqqix8pedc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpajswdci11jqqix8pedc.png" alt="Error state for a repository that does not exist" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Joe-Simo" rel="noopener noreferrer"&gt;
        Joe-Simo
      &lt;/a&gt; / &lt;a href="https://github.com/Joe-Simo/last-commit" rel="noopener noreferrer"&gt;
        last-commit
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      An ofrenda for open source that went quiet. An open-weight model in your browser writes the epitaph.
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;Last Commit&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;An ofrenda for open source that went quiet.&lt;/p&gt;
&lt;p&gt;Paste a public GitHub repository and Last Commit lays out a remembrance card for it: the years it was alive, what it gave people, its last words (the real final commit message) and the people who tended it. Then an open-weight model, running entirely in your browser tab, writes its epitaph.&lt;/p&gt;
&lt;p&gt;Live: &lt;a href="https://last-commit-ofrenda.vercel.app" rel="nofollow noopener noreferrer"&gt;https://last-commit-ofrenda.vercel.app&lt;/a&gt;&lt;/p&gt;
&lt;p&gt;&lt;a rel="noopener noreferrer" href="https://github.com/Joe-Simo/last-commit/public/screenshots/card-atom.png"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fraw.githubusercontent.com%2FJoe-Simo%2Flast-commit%2FHEAD%2Fpublic%2Fscreenshots%2Fcard-atom.png" alt="A Last Commit card for atom/atom"&gt;&lt;/a&gt;&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Demo&lt;/h2&gt;
&lt;/div&gt;
&lt;p&gt;A 47 second narrated walkthrough recorded on the live site: &lt;a href="https://last-commit-ofrenda.vercel.app/demo/last-commit-demo.mp4" rel="nofollow noopener noreferrer"&gt;last-commit-demo.mp4&lt;/a&gt;. The model download and writing part is sped up 12x and labeled as such; in real time it took 2 minutes 15 seconds on the CPU path in headless Chrome. The voiceover was generated with ElevenLabs text to speech (&lt;a href="https://last-commit-ofrenda.vercel.app/demo/narration-elevenlabs.mp3" rel="nofollow noopener noreferrer"&gt;narration-elevenlabs.mp3&lt;/a&gt;).&lt;/p&gt;
&lt;p&gt;Made for the DEV Hacktoberfest Open-Source AI Challenge, Week 4: Ghost in the Machine.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;How it works&lt;/h2&gt;
&lt;/div&gt;
&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Facts first.&lt;/strong&gt; Your browser calls GitHub's public REST API with no token: the repository, its…&lt;/li&gt;
&lt;/ol&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Joe-Simo/last-commit" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;p&gt;MIT licensed. Next.js 16, React 19, TypeScript, Tailwind CSS 4, Transformers.js, Vitest (48 tests covering parsing, status, fallback text and reply checks).&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Facts first, AI second.&lt;/strong&gt; The browser calls GitHub's public REST API with no token: the repository, its last ten commits, its top contributors, its README and its latest release. That is five requests per lookup. The whole card is built from those facts alone, so it is complete and correct before a model ever loads. The model can only make it nicer, never make it wrong.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The model.&lt;/strong&gt; &lt;a href="https://huggingface.co/Qwen" rel="noopener noreferrer"&gt;Qwen2.5 Instruct&lt;/a&gt; in the ONNX builds from the Hugging Face onnx-community, under Apache 2.0, run with &lt;a href="https://github.com/huggingface/transformers.js" rel="noopener noreferrer"&gt;Transformers.js&lt;/a&gt; in a Web Worker so the page never freezes. On WebGPU (with 16 bit float support) the app uses 4 bit &lt;code&gt;q4f16&lt;/code&gt; weights and defaults to the 1.5B model, with a one click swap to 0.5B. Without WebGPU it uses WebAssembly on the CPU with &lt;code&gt;q4&lt;/code&gt; weights and the 0.5B model. I tried the 8 bit CPU weights and they produced garbled text. I also tried the 1.5B model on the CPU path: it downloaded about 1.8 GB and then ran out of memory inside WebAssembly, so the app now only offers 1.5B on WebGPU and explains why.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;A tiny prompt.&lt;/strong&gt; The model only sees four lines: name, what it was about, the years it was active, and whether it is archived or how long it has been quiet. Two short examples (left-pad and jshint) show the tone. Decoding is greedy and capped at 32 tokens for the epitaph and 96 for the remembrance.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Every reply is checked.&lt;/strong&gt; Before a reply reaches the card, it is set aside if it invents a number that is not in the facts, slips into first person, includes links or emoji, echoes the prompt, starts chatting, or runs too long. When that happens, the fact based line stays and the panel says why, as with brackets and request above. Small models do this a lot, and I would rather show a plain true line than a confident invented one.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Keepsakes.&lt;/strong&gt; The card exports to PNG with &lt;code&gt;html-to-image&lt;/code&gt;, fonts included. Lookups are cached in localStorage for six hours, the five examples are bundled as snapshots so they work even when the 60 requests an hour GitHub limit runs out, and the rate limit error tells you when it resets.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The demo.&lt;/strong&gt; I recorded the live site with Playwright, sped up the download and writing part 12x (it is labeled on screen; in real time it took 2 minutes 15 seconds), and added a voiceover generated with ElevenLabs text to speech.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;This app only exists because everything in it is open. The facts come from public repositories and a public API. The model is open-weight, so it can run inside a browser tab on my own machine with no server, no account and no bill, and nothing anyone looks up leaves their browser except the GitHub requests. And the projects on the cards are open source, which is why they are not really gone: atom still has 17,160 forks on GitHub. A closed product that shuts down is lost. An open one can be picked up by anyone with a fork button.&lt;/p&gt;

&lt;p&gt;That is the point of the ofrenda. Remembering is not only nostalgia, it is a nudge: before you add a dependency, look at its last words. And if a quiet project still matters to you, fork it, or thank the people who tended it while they still read their notifications.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limitations
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;The 0.5B model writes flat or clumsy lines. You can see it above: "a true hack, by any standards" and "a giant sponge" are not poetry. The checks catch invented facts and first person, not dull prose. The 1.5B model on WebGPU writes better, but my runs above are all 0.5B on the CPU.&lt;/li&gt;
&lt;li&gt;The first model download is large (793 MB on the CPU path) and CPU generation is slow: 18 to 50 seconds per part in my runs.&lt;/li&gt;
&lt;li&gt;Without a token, GitHub allows 60 API requests an hour per IP, and one lookup uses five.&lt;/li&gt;
&lt;li&gt;"Gone quiet" is a heuristic: archived, or no commits on the default branch for a year. A quiet repository is not always a dead one.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Prize Categories
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Best Use of ElevenLabs&lt;/strong&gt;: the demo narration was generated with ElevenLabs text to speech from a script I wrote (&lt;a href="https://last-commit-ofrenda.vercel.app/demo/narration-elevenlabs.mp3" rel="noopener noreferrer"&gt;mp3&lt;/a&gt;). The app itself does not use ElevenLabs.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;AI disclosure: I built Last Commit with help from an AI coding assistant. The epitaphs and remembrances in the app are written in the browser by Qwen2.5 Instruct (0.5B or 1.5B, open-weight, Apache 2.0); every model line quoted in this post is copied from a real run on the live site. The demo voiceover is ElevenLabs text to speech.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>opensource</category>
      <category>ai</category>
    </item>
    <item>
      <title>Lazy Changelog, where Gemma writes your release notes inside the browser</title>
      <dc:creator>Joe Simo</dc:creator>
      <pubDate>Sat, 24 Oct 2026 22:45:54 +0000</pubDate>
      <link>https://dev.to/joesimo/lazy-changelog-where-gemma-writes-your-release-notes-inside-the-browser-5hh8</link>
      <guid>https://dev.to/joesimo/lazy-changelog-where-gemma-writes-your-release-notes-inside-the-browser-5hh8</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week3-2026-10-19"&gt;Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;Release notes are the chore that always gets skipped. Everything users need to know is already in the commits, but it is written for the person who made the change: &lt;code&gt;fix(ui): dark mode toggle not persisting after refresh&lt;/code&gt;. Turning that into "Dark mode toggle persists after refresh" for thirty commits is boring, so it does not happen.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Lazy Changelog&lt;/strong&gt; does it for you. You run one command, paste, and copy:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;git log &lt;span class="nt"&gt;--oneline&lt;/span&gt; &lt;span class="si"&gt;$(&lt;/span&gt;git describe &lt;span class="nt"&gt;--tags&lt;/span&gt; &lt;span class="nt"&gt;--abbrev&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;0&lt;span class="si"&gt;)&lt;/span&gt;..HEAD
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;An instant draft, before any AI.&lt;/strong&gt; The moment you paste, a small parser groups commits into New, Fixed, Faster, Docs and Behind the scenes. It strips &lt;code&gt;feat:&lt;/code&gt; style prefixes, gitmoji, &lt;code&gt;[tags]&lt;/code&gt; and hashes, keeps PR numbers like &lt;code&gt;(#212)&lt;/code&gt;, puts breaking changes at the top, and quietly drops merges, version bumps, &lt;code&gt;wip&lt;/code&gt; and &lt;code&gt;fixup!&lt;/code&gt; commits and duplicates. You can copy this draft right away.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Then Gemma rewrites the lines users care about.&lt;/strong&gt; Gemma 3 1B runs inside the browser tab and rewrites each New, Fixed and Faster line in plain words, one at a time, so you watch them change.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;After the first time, there is nothing to click.&lt;/strong&gt; The weights stay cached. Next time you open the page, the model loads by itself and rewrites whatever you paste.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Who gets to be lazier: anyone who ships a library, an app or a side project and owes their users a "what changed" note.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F26uw0spcmt3av30hroqc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F26uw0spcmt3av30hroqc.png" alt="Release notes rewritten by Gemma in the browser" width="800" height="575"&gt;&lt;/a&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Live:&lt;/strong&gt; &lt;a href="https://lazy-changelog.vercel.app" rel="noopener noreferrer"&gt;https://lazy-changelog.vercel.app&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Click &lt;strong&gt;Try an example&lt;/strong&gt; to load a sample log, then &lt;strong&gt;Load Gemma 3 1B&lt;/strong&gt;. The first load downloads the model once (about 760 MB with WebGPU, about 1 GB on the CPU path). No account, no key.&lt;/p&gt;

&lt;p&gt;Here is a real run of the example log on the live site, on the CPU path:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Commit&lt;/th&gt;
&lt;th&gt;Release note&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;feat(reports): export reports as CSV (#212)&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Export your reports as CSV (#212)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;fix(ui): dark mode toggle not persisting after refresh&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Dark mode toggle persists after refresh&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;fix: resolve race condition in websocket reconnect&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;The websocket reconnect process is now stable&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;perf(gallery): lazy load images&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Lazy load images &lt;em&gt;(kept as written)&lt;/em&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Not every line gets better, and the app says so. Turn on &lt;strong&gt;Show commits&lt;/strong&gt; to see the commit behind each line, and which ones Gemma left alone:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fz71l4077ap2l51ifa7e0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fz71l4077ap2l51ifa7e0.png" alt="Each release note with the commit it came from" width="800" height="670"&gt;&lt;/a&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Joe-Simo" rel="noopener noreferrer"&gt;
        Joe-Simo
      &lt;/a&gt; / &lt;a href="https://github.com/Joe-Simo/lazy-changelog" rel="noopener noreferrer"&gt;
        lazy-changelog
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      Paste your git log, get release notes people can read. Gemma 3 1B rewrites them in your browser.
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;Lazy Changelog&lt;/h1&gt;
&lt;/div&gt;

&lt;p&gt;Paste your git log. Get release notes people can read. Nothing leaves your browser.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Live:&lt;/strong&gt; &lt;a href="https://lazy-changelog.vercel.app" rel="nofollow noopener noreferrer"&gt;https://lazy-changelog.vercel.app&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a rel="noopener noreferrer" href="https://github.com/Joe-Simo/lazy-changelog/public/screenshots/rewritten.png"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fraw.githubusercontent.com%2FJoe-Simo%2Flazy-changelog%2FHEAD%2Fpublic%2Fscreenshots%2Frewritten.png" alt="Release notes rewritten in the browser"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Why&lt;/h2&gt;
&lt;/div&gt;

&lt;p&gt;Writing release notes is the chore everyone skips. The information is already in your commits; it is just written for you, not for your users. Lazy Changelog does the boring part for you:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;p&gt;Run one command in your repo:&lt;/p&gt;
&lt;div class="highlight highlight-source-shell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;git log --oneline &lt;span class="pl-s"&gt;&lt;span class="pl-pds"&gt;$(&lt;/span&gt;git describe --tags --abbrev=0&lt;span class="pl-pds"&gt;)&lt;/span&gt;&lt;/span&gt;..HEAD&lt;/pre&gt;

&lt;/div&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;Paste the output.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;Copy the Markdown.&lt;/p&gt;
&lt;/li&gt;
&lt;/ol&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;How it works&lt;/h2&gt;
&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Instant draft, no model needed.&lt;/strong&gt; A small parser (&lt;code&gt;src/lib/parse.ts&lt;/code&gt;) reads the log the moment you paste it. It groups commits into New, Fixed, Faster, Docs and Behind the scenes, strips conventional commit prefixes, gitmoji, &lt;code&gt;[tags]&lt;/code&gt; and hashes, keeps PR numbers, puts breaking changes first, and drops noise like merges, version bumps, &lt;code&gt;wip&lt;/code&gt; and &lt;code&gt;fixup!&lt;/code&gt; commits and duplicates. This draft is labeled as a draft.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Then Gemma rewrites the lines users care&lt;/strong&gt;…&lt;/p&gt;&lt;/div&gt;


&lt;/div&gt;
&lt;br&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Joe-Simo/lazy-changelog" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;br&gt;
&lt;/div&gt;
&lt;br&gt;


&lt;p&gt;MIT licensed. Next.js, React, TypeScript, Tailwind CSS, Transformers.js. Tests cover the parser and the reply checks (&lt;code&gt;npm test&lt;/code&gt;).&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;The model.&lt;/strong&gt; &lt;a href="https://huggingface.co/onnx-community/gemma-3-1b-it-ONNX" rel="noopener noreferrer"&gt;Gemma 3 1B instruct&lt;/a&gt;, in the ONNX build from the Hugging Face onnx-community, run with &lt;a href="https://github.com/huggingface/transformers.js" rel="noopener noreferrer"&gt;Transformers.js&lt;/a&gt; in a Web Worker so the page never freezes. With WebGPU it uses 4-bit weights (&lt;code&gt;q4f16&lt;/code&gt;). Without WebGPU it falls back to WebAssembly on the CPU with 8-bit weights (&lt;code&gt;uint8&lt;/code&gt;), because the 4-bit file needs an operator the WebAssembly runtime does not have. I tried the smaller Gemma 3 270M first; it was fast, but in my tests its rewrites were wrong too often, so I went with 1B.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Small jobs, not one big prompt.&lt;/strong&gt; A 1B model is good at one short task and bad at a long one. So the parser does everything that can be done with rules (grouping, cleanup, noise removal), and Gemma only gets one commit subject at a time, with a short instruction for its kind and three examples. Fixes get "this bug is now fixed, say what works now", which matters: without it, a small model happily turns "crash when uploading files" into a feature. Decoding is greedy and capped at 32 new tokens.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Never show junk as if it were good.&lt;/strong&gt; Every reply goes through a checker before it reaches the screen. It is thrown away if it is empty, too short or too long, repeats the prompt, starts chatting ("Sure", "I'm ready"), contains stray characters, shares no real word with the commit, or simply echoes the commit back. When that happens, the original line stays and is marked "kept as written". On the CPU path this happens more often, and I would rather show an honest original than a confident wrong sentence.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Lazy by default.&lt;/strong&gt; After the first successful load, the app remembers it and loads the model on its own. Pasting a new log rewrites it automatically after a short pause. Loading, download progress, empty states and errors all have their own clear message, and if the model cannot load at all, the instant draft is still there to copy.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Commit logs are private more often than not: unreleased features, customer names in branch names, security fixes that are not public yet. With a closed API, using this tool would mean sending that history to someone else's server and putting an API key somewhere.&lt;/p&gt;

&lt;p&gt;Because Gemma's weights are open and Transformers.js and ONNX Runtime Web are open source, the whole thing runs in the tab. There is no backend at all, so there is nothing to pay for per request, no key to leak, and no server that ever sees your commits. That is also what lets it be free to host and free to use. An open model running on your own machine is the only way this tool can be both private and lazy.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prize Categories
&lt;/h2&gt;

&lt;p&gt;Best Use of Gemma&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I built this project and wrote this post with heavy help from an AI agent. Every result shown above came from real runs of the live app.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>gemma</category>
      <category>opensource</category>
    </item>
  </channel>
</rss>
