Embed Notice

HTML Code

<blockquote style="position: relative; padding-left: 55px;"><section><a href="https://subs4social.xyz/notes/a0tv7fhjci">kuteboiCoder (kuteboicoder@subs4social.xyz)'s status on Thursday, 21-Nov-2024 10:27:28 JST</a><a href="https://subs4social.xyz/@KuteboiCoder" title="kuteboicoder@subs4social.xyz"><img src="https://gnusocial.jp/theme/gnusocialjp/default-avatar-stream.png" width="48" height="48" alt="kuteboiCoder" style="position: absolute; left: 0; top: 0;">kuteboiCoder</a><div><a href="https://masto.ai/@napocornejo/113517004417367705" rel="in-reply-to">in reply to</a><ul><li><li><a href="https://gnusocial.jp/user/298602" title="napocornejo@masto.ai">Napoleon Cornejo</a></li></ul></div></section><article><p><a href="https://masto.ai/@napocornejo">@napocornejo@masto.ai</a> <br><br>Yesterday I played with <a href="https://subs4social.xyz/tags/phi3" rel="tag">#phi3</a> <a href="https://subs4social.xyz/tags/LLM" rel="tag">#LLM</a> - actually it's a state-of-the-art <a href="https://subs4social.xyz/tags/SLM" rel="tag">#SLM</a> - on a cloud <a href="https://subs4social.xyz/tags/GPU" rel="tag">#GPU</a>.  It runs faster than mistral-nemo on the same GPU while still giving encyclopedic answers.<br><br>I haven't compared phi3 against a mid-sized or large <a href="https://subs4social.xyz/tags/Mistral" rel="tag">#Mistral</a> <a href="https://subs4social.xyz/tags/Mixtral" rel="tag">#Mixtral</a> model.  When I do, it just might surprise me.<br><br><a href="https://blob.cat/users/icedquinn">@icedquinn@blob.cat</a></p></article><footer><a rel="bookmark" href="https://gnusocial.jp/conversation/4039236#notice-7893244">In conversation</a><time datetime="2024-11-21T10:27:28+09:00" title="Thursday, 21-Nov-2024 10:27:28 JST">about 15 hours ago</time> <span>from <span><a href="https://subs4social.xyz/notes/a0tv7fhjci" rel="external" title="Sent from subs4social.xyz via ActivityPub">subs4social.xyz</a></span></span><a href="https://subs4social.xyz/notes/a0tv7fhjci">permalink</a></footer></blockquote>

Corresponding Notice

Embed this notice
kuteboiCoder (kuteboicoder@subs4social.xyz)'s status on Thursday, 21-Nov-2024 10:27:28 JSTkuteboiCoder
in reply to
- iced depresso
- Napoleon Cornejo
@napocornejo@masto.ai

Yesterday I played with #phi3 #LLM - actually it's a state-of-the-art #SLM - on a cloud #GPU. It runs faster than mistral-nemo on the same GPU while still giving encyclopedic answers.

I haven't compared phi3 against a mid-sized or large #Mistral #Mixtral model. When I do, it just might surprise me.

@icedquinn@blob.cat
In conversationabout 15 hours ago from subs4social.xyzpermalink

Public

Embed Notice

HTML Code

Corresponding Notice