toad.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
Mastodon server operated by David Troy, a tech pioneer and investigative journalist addressing threats to democracy. Thoughtful participation and discussion welcome.

Administered by:

Server stats:

291
active users

#openwebui

0 posts0 participants0 posts today
MetaDarko<p>Which model is your favorite?</p><p>I keep experimenting, but the real choice… is never easy.</p><p>And most of all, how do you use them?</p><p><a href="https://sigmoid.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://sigmoid.social/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLM</span></a> <a href="https://sigmoid.social/tags/OpenSource" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenSource</span></a> <a href="https://sigmoid.social/tags/Gemma3" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Gemma3</span></a> <a href="https://sigmoid.social/tags/ChatGPT" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ChatGPT</span></a> <a href="https://sigmoid.social/tags/GPT" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GPT</span></a> <a href="https://sigmoid.social/tags/Qwen" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Qwen</span></a> <a href="https://sigmoid.social/tags/DeepSeek" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DeepSeek</span></a> <a href="https://sigmoid.social/tags/AICommunity" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AICommunity</span></a> <a href="https://sigmoid.social/tags/Ollama" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Ollama</span></a> <a href="https://sigmoid.social/tags/OpenWebUI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenWebUI</span></a> <a href="https://sigmoid.social/tags/huggingface" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>huggingface</span></a> <a href="https://sigmoid.social/tags/API" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>API</span></a></p>
nonlinear<p>I decided to take a step back and craft an intake prompt for my projects, be them personal, hobbies, or work. </p><p>I want to be more a measure twice cut once kinda guy, only jumping to execution when I have a clear picture of the environment and scope constraints. of any project.</p><p>I'm currently learning advanced <a href="https://social.praxis.nyc/tags/prompt" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>prompt</span></a> engineering on <a href="https://social.praxis.nyc/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a> </p><p><a href="https://docs.openwebui.com/features/workspace/prompts/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">docs.openwebui.com/features/wo</span><span class="invisible">rkspace/prompts/</span></a></p>
Bill<p>I have set up a disconnected local chatbot using open webui, and I have fed it all of the manuals and maps and diagrams of the Sempf Compound. I can ask it things like "which breaker should I turn off to replace the floor outlets in the living room" and it answers correctly.</p><p>I'm afraid to talk about it here because of the general hatred of GenAI. Probably should set up an alt on a techbro instance or something. My filters would probably block everyone so I'm not sure that would work.</p><p><a href="https://infosec.exchange/tags/genai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>genai</span></a> <a href="https://infosec.exchange/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a></p>
CupcakeRob<p>ik heb een vim boek in <a href="https://gts.cupcakerob.nl/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a> gegooid en ollama is nu 5 uur lang aan het denken over de vraag "how do I exit vim"</p>
Rich Tong ❤️🤷‍♂️<p>ai: Open WebUI -1 problems, more ram, $200 plans and ComfyUI URLs<br>OK, some miscellaneous but valuable comments in the AI journey</p><p>Open WebUI model parameters redux</p><p>The context length and other parameters of Open WebUI are really low, like 2K context windows. This doesn't make any sense in the era of AI coding. Some defaults help a bit, which we've covered be<br><a href="https://tongfamily.com/2025/08/04/ai-open-webui-1-problems-more-ram-200-plans-and-comfyui-urls/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">tongfamily.com/2025/08/04/ai-o</span><span class="invisible">pen-webui-1-problems-more-ram-200-plans-and-comfyui-urls/</span></a><br><a href="https://mastodon.social/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://mastodon.social/tags/comfyui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>comfyui</span></a> <a href="https://mastodon.social/tags/Geek" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Geek</span></a> <a href="https://mastodon.social/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a></p>
Blabla Linux<p><a href="https://mastodon.blablalinux.be/tags/Blinko" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Blinko</span></a> passe en version 1.2.0 (<a href="https://github.com/blinkospace/blinko/releases/tag/1.2.0" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/blinkospace/blinko/</span><span class="invisible">releases/tag/1.2.0</span></a>), <a href="https://mastodon.blablalinux.be/tags/OpenWebUI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenWebUI</span></a> passe en version 0.6.17 (<a href="https://github.com/open-webui/open-webui/releases/tag/v0.6.17" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/open-webui/open-web</span><span class="invisible">ui/releases/tag/v0.6.17</span></a>), et <a href="https://mastodon.blablalinux.be/tags/PhotoPrism" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>PhotoPrism</span></a> en version 250707-d28b3101e (<a href="https://github.com/photoprism/photoprism/releases/tag/250707-d28b3101e" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/photoprism/photopri</span><span class="invisible">sm/releases/tag/250707-d28b3101e</span></a>) 😎</p>
nonlinear<p>I proposed a feature for <a href="https://social.praxis.nyc/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a>: voice differentiation for agents (they call them models, somehow)</p><p>Possibilities: QR codes, AR links, bookmarks for elderly users — voice-first UX without barriers.</p><p>📌 <a href="https://github.com/open-webui/open-webui/issues/15534" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/open-webui/open-web</span><span class="invisible">ui/issues/15534</span></a></p><p><a href="https://social.praxis.nyc/tags/sovereignAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sovereignAI</span></a> <a href="https://social.praxis.nyc/tags/usability" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>usability</span></a> <a href="https://social.praxis.nyc/tags/accessibility" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>accessibility</span></a> <a href="https://social.praxis.nyc/tags/voice" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>voice</span></a> <a href="https://social.praxis.nyc/tags/cyborg" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>cyborg</span></a></p>
nonlinear<p>I proposed a feature for <a href="https://social.praxis.nyc/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a>: a URL param to launch directly into Voice Mode, no clicks needed.<br>Possibilities: QR codes, AR links, bookmarks for elderly users — voice-first UX without barriers.</p><p>📌 <a href="https://github.com/open-webui/open-webui/issues/15531" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/open-webui/open-web</span><span class="invisible">ui/issues/15531</span></a></p><p><a href="https://social.praxis.nyc/tags/sovereignAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sovereignAI</span></a> <a href="https://social.praxis.nyc/tags/usability" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>usability</span></a> <a href="https://social.praxis.nyc/tags/accessibility" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>accessibility</span></a> <a href="https://social.praxis.nyc/tags/voice" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>voice</span></a> <a href="https://social.praxis.nyc/tags/cyborg" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>cyborg</span></a></p>
Geekmaster 👽:system76:<p>I have also heard positive feedback on <a href="https://ioc.exchange/tags/OpenWebUI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenWebUI</span></a> - which brings a <a href="https://ioc.exchange/tags/GUI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GUI</span></a> to ollama. Haven't tried it yet (it requires <a href="https://ioc.exchange/tags/docker" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>docker</span></a> to be setup) but my colleagues in the space are raving about it.</p>
Peter B.<p>For my <a href="https://mastodon.social/tags/FIAF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>FIAF</span></a> presentation &amp; workshop in Bologna end of the week, I've put together a clean "<a href="https://mastodon.social/tags/PlugAndPlayAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>PlugAndPlayAI</span></a>" setup:</p><p><a href="https://mastodon.social/tags/USB" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>USB</span></a> <a href="https://mastodon.social/tags/LiveDistro" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LiveDistro</span></a>: <a href="https://mastodon.social/tags/MXLinux" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MXLinux</span></a> <a href="https://mastodon.social/tags/GNU" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GNU</span></a> <a href="https://mastodon.social/tags/Debian" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Debian</span></a> with <a href="https://mastodon.social/tags/Ollama" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Ollama</span></a>, <a href="https://mastodon.social/tags/LLava" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLava</span></a> and <a href="https://mastodon.social/tags/OpenWebUI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenWebUI</span></a> in <a href="https://mastodon.social/tags/Docker" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Docker</span></a> and <a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLM</span></a> files on another USB3 <a href="https://mastodon.social/tags/nvme" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>nvme</span></a> </p><p>Fast like a local install. Given enough RAM and CPU/GPU.</p><p>Scary how easy it is to do that.</p><p>Bonus: Creates self-replicating <a href="https://mastodon.social/tags/ISO" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ISO</span></a> boot images of itself (in current configuration) in less than 5 minutes. 🤯 😎 .</p><p>Wow.</p>
nonlinear<p>I went to the limit for what <a href="https://social.praxis.nyc/tags/openwebui" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>openwebui</span></a> offers to my <a href="https://social.praxis.nyc/tags/sovereignAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sovereignAI</span></a>, NGL thing is nifty but kinda shoddy sometimes... </p><p>Like a voice control so sensitive that ANY movement pauses answers, defeating the whole point of using voice while walking, in a cafe, cleaning, etc.</p><p>Reddit community is kinda ineffective, so now I'm writing straight on repo, as issues</p><p><a href="https://github.com/open-webui/open-webui/issues/14500" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/open-webui/open-web</span><span class="invisible">ui/issues/14500</span></a></p>

From OpenWebUI:

> 🦻 Screen Reader Support for Response Messages
All chat responses are now fully compatible with screen readers, making the platform more inclusive and accessible for everyone.

Well um, I mean, I don't use a hearing aid... But um, that's... Good... I'll see if it works.

Continued thread

The prompt I sent to gemma3:27b locally via #openWebUI was as follows (wrapped in a shell script to populate $topic): "Write one paragraph in markdown format about \"'$topic'\", in the voice of a wikipedia author. For any word in the response that would be a link to its own article, use markdown formatted links to link to an appropriate url. The base url for any path should always be http://localhost:3000/sandbox"

Replied in thread

@Gina If I had to choose, I'd probably go with #Ollama (which has been mentioned several times already). It's licensed under the MIT license and the models are about as close to open source as you can get. When I play with LLMs, it's what I use. Locally run and with an API that could be used to integrate with other stuff. I also have #OpenWebUI to make things prettier. Both can run locally, though OpenWebUI can integrate with cloud LLMs too. Of course, tomorrow everything could change.

Сегодня маленький приятный восторг - и опять с нейросетями.

Коллега последние два с чем-то дня пердолился с одной внешней системой, которая хотела очень специфическую подпись запросов.

У системы есть референсный микросервис на Python и невалидные примеры кода на C#.

Он пытался спрашивать ChatGPT, но результат получался неправильный. Кстати, такой же как в дотнете.

Я решил попробовать помочь. Предложил потыкать Claude Sonnet. Но у него нет туда доступа.

Тогда пошёл копаться сам. Нашёл неконсистентности между тем как делается подпись в Python и Go.

Рабочая гипотеза была сформирована.

Пошёл в ChatGPT, попросил написать с учётом своих находок и получил диалог вида:

Я: Напищи мне порт на Go
GPT: Вот
Я: Но тут совсем не то, смотри в чем разница
GPT: Тогда вот
Я: Но это не работает. Смотри какая проблема и ошибка.
GPT: Ну все правильно, ты передашь не то в метод, вот как надо
GPT: *По сути откатывает код назад*
Я: Ебать спасибо нахуй

Меня это тяни-толкание заёбывает, я иду и покупаю кредиты на OpenRouter за крипту.

Выписываю ключ API и подключаю его к своему @openwebui чтобы кроме локальных моделей можно было подключить облачную тяжелую артиллерию когда надо.

Иду в Claude Sonnet.

Я: надо портировать на Go
Claude: Вот
Claude: *пишет код со всеми теми нюансами, которые нашёл и я когда смотрел разницу в биндингах к криптобиблиотекам*
Я: Неплохо, но вот в этом месте стандартной либы Go происходит ошибка.
Claude: И правда. Вот что надо исправить.

И выдаёт полностью рабочий код абсолютно идентичный референсному микросервису, который выдаёт правильные подписи за счёт правильных костылей имитирующих то, что творится в питоне.

Примерно за 5 центов в сумме.

Короче выводы сегодняшнего дня:

- Если вам не помогает GPT с кодом - обязательно сходите к Claude
- Я буду добиваться корпоративного OpenWebUI с доступом к разным ассистентам в компании

#dev#log#work

Deepseek Model R1 läuft lokal auf dem Mac Studio M3 Ultra
Das Large Language Modell R1 vom Chinesischen Startup Deepseek lässt sich ohne Probleme lokal auf einem vollausgebauten Mac Studio mit M3 Ultra installieren und betreiben. Aber nicht nur das: Es steht dann auch auf iPhone, iPads und Mac im gleichen Netzwerk zur
apfeltalk.de/magazin/news/deep
#KI #News #AI #GenerativeKI #KI #KnstlicheIntelligenz #Ollama #OpenWebUI

Apfeltalk Magazin · Deepseek Model R1 läuft lokal auf dem Mac Studio M3 UltraLernen Sie die Vorteile der lokalen Nutzung von Ollama auf dem Mac Studio mit M3 Ultra kennen und geniessen Sie schnelle Antworten.

Testing Open WebUi with Gemma:3 on my proxmox mini PC in a LXC. My hardware is limited, 12th Gen Intel Core i5-12450H so I’m only using the 1b (28 token/s) and 4b (11 token/s) version for now.

Image description is functioning, but it is slow; it takes 30 seconds to generate this text with the 4b version and 16G allocated for the LXC.

Next step, trying this on my Mac M1.

The sun has returned!

I mean, it has always been there. Relatively speaking in the same place it was over the last few months. But the Earth’s tilt is such that as it revolves around the nearest star the portion on which I live (the northern hemisphere) is getting ever so slightly closer and faces it just long enough that temperatures are beginning to warm.

And I’m a very happy person […]

https://cdevroe.com/2025/03/13/diversions-9/