Comment by jeff_ciesielski

Back to stories Open in a window
Comment

Comment by jeff_ciesielski

Howdy folks.

I was inspired by the really neat "Jev in 25 lines of Python" post a few days ago (and System One models in general), and after playing with the idea I tried whipping up something a little more complex, and Lichen is what popped out: https://github.com/Mushroom-Systems/lichen

My question after reading that post (and I know I wasn't the only one; antirez came right out and said it in the comments) was whether Google Research's prompt repetition paper for non-reasoning models (https://arxiv.org/abs/2512.14982) would improve the outcomes. Turns out it did a fair bit. After some experimentation I added two more tricks: the options are listed a second time in a rotated order to wash out the model's preference for position, and a letter remapping sums the probabilities over each option's letters. Both help on the harder and more ambiguous questions.

After testing several models I had lying around, I settled on gemma-4-26B-A4B as my daily driver. On public JevBench, Google's QAT build scores 207/231 against Jev 1.13's 200 (not statistically significant yet; I'll be submitting it for evaluation on the withheld items soon). Served from vLLM with the NVFP4 build, it scores 204 with a median response time of about 56 ms on my laptop 5090. For folks without that much VRAM to spare, the smaller Gemma 4 models also perform well: E4B fits in about 5 GB, though it does suffer on the harder queries. Qwen also works (my colleague has reported good results with Qwen3.8), but on my setup I found it a bit slow for my use case, and the benefits weren't strong enough to outweigh the slowdown.

Because Gemma 4 is multimodal, Lichen also extends the query format so images can be part of the state (served from vLLM), with answers in about 190 ms. The results are good but not perfect. On 125 Wikimedia Commons photos that I checked by hand, it got every one right. On 98 road-sign photos drawn at random from Commons, it named the right sign 92 times when it had to pick one. When I added "another sign, or no road sign" as an option, it dropped to 73: it tends to decide that a small, distant or vandalized sign isn't there, and it's confident about those misses. The confidence on images isn't well calibrated in either direction, and a separate temperature for images didn't fix it. The README has the numbers and a couple of the failures.

I've been using this as a partial replacement for the backbone of our company's context graph for a few days now, and it's performed excellently so far. Hopefully folks will find some use for it.

(Sidenote: Big thanks to my friend and colleague Jacob M for his vLLM support patch)

Replies jeff_ciesielski · 1d
Open on HN
Loading the discussion…

Domain filters

Stories from these domains are hidden from every list. Subdomains match too: blocking substack.com also hides danluu.substack.com.

    About YAVCHN

    YAVCHN is a reader for Hacker News and Lobsters, with articles and discussions in separate windows or Classic pages.

    Created by Paul Parks and built with PUDL.

    YAVCHN source code on GitHub

    Help

    Keyboard

    j / k
    Move down and up the story list. The arrow keys scroll whatever has focus.
    Enter
    Read the marked story in the article reader.
    ]
    Read the next story in the same article-reader applet. Back returns to the previous story.
    p
    Pin or unpin the marked story, which keeps it in Pinned.
    n / N
    Move to the next or previous top-level comment in the window in front.
    c
    Collapse or expand that comment.
    f
    Hide or show the story list.
    Esc
    Close a menu or this help.
    Access key m
    Go to the menu bar. Most browsers take it with Alt on Windows and Linux, and Safari with Control and Option.
    ?
    Show this help.

    Windows

    Each story opens in a window holding its article above its discussion; drag the bar between them to share the room differently. A window can be moved by its title bar, resized from any edge, snapped to a half or a corner by dragging it there, maximised, or minimised to the bar at the foot of the page. Use Window > New reader window to open an empty reader, or a story row's Open in new reader window button to compare articles. Docked readers keep their articles when you select another story from the sidebar. Minimized readers can be restored and reused for their site. A window's Next story link reads on down the list in the same window.

    A link in a comment or an article to another Hacker News or Lobsters thread opens that thread in a window too. A link to a single HN comment opens the comment above its replies.

    While a story's window is in front, the Story and Discussion menus in the menu bar hold its commands: pinning, Next story, sorting, collapsing every thread, jumping to the first new comment. Each window also remembers where you were in its article and discussion, so a reload, or Back to a story that Next took you past, finds your place again. Closing a window forgets it.

    The whole arrangement lives in the address, so a bookmark or a shared link brings it back, and Back undoes the last change. Moving between Hacker News, Lobsters, their lists, Pinned and Find changes only the list, and leaves the windows open.

    The list

    The pin at the start of a row keeps the story in Pinned, and the cross at its end hides it. Pinned can be narrowed by words in the title, site or author, by source, and to the stories you haven't opened yet, and ordered by when you pinned them, by points or by comments; the filters are part of the address, so a filtered view can be bookmarked. Scroll past the end of the list to load more. Domain filters, in the View menu, hide every story from a site.

    Browsing view

    View > Windowed and View > Classic select the browsing view and save your default in this browser. Window view reuses a reader for each feed. Classic view opens stories and applets as pages. Open as a page and Open in a window are one-off actions that do not change your saved default. Direct page links always open as pages.

    Applets

    The Applets menu in the menu bar holds three tools, each a window of its own. Replies to me takes your Hacker News user name and lists the replies to your last thirty comments and stories, checking again every three minutes while it is open, and marking what is new since you last marked them read. Look up a user opens a profile on Hacker News or Lobsters, with their submissions and recent comments, as a commenter's name in any discussion does; the bar at the top of a profile looks up someone else in the same window, and Back returns to the one before. Who is hiring? filters the posts of HN's monthly hiring threads by the words you type.

    They read only what the sites publish to everyone, so none of them asks for a login, and your user name stays in this browser.

    Find

    Find takes any link and lists every time it was submitted to Hacker News and Lobsters, so you can read each discussion of it.

    About

    YAVCHN never sees your Hacker News or Lobsters login. The discussion is fetched from each site's public API; to vote or reply, follow the link above the discussion, or the arrow beside a comment, to the source's own site. Pins, hidden stories, filters and layout are kept in this browser only.

    Open source: github.com/paulmooreparks/yavchn. Built with PUDL.