TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Google calls Gemma 3 the most powerful AI model you can run on one GPU

127 点作者 gmays2 个月前

15 条评论

cwoolfe2 个月前
Apparently it can also pray. Seriously, I asked it for biblical advice about a tough situation today and it said it was praying for me. XD
评论 #43428763 未加载
评论 #43429056 未加载
评论 #43428864 未加载
评论 #43429405 未加载
评论 #43428937 未加载
评论 #43429376 未加载
评论 #43429911 未加载
timmg2 个月前
I&#x27;m wondering how small of a model can be &quot;generally intelligent&quot; (as in LLM intelligent, not AGI). Like there must be a size too small to hold &quot;all the information&quot; in.<p>And I also wonder at what point we&#x27;ll see specialized small models. Like if I want help coding, it&#x27;s probably ok if the model doesn&#x27;t know who directed &quot;Jaws&quot;. I suspect that is the future: many small, specialized models.<p>But maybe training compute will just get to the point where we can run a full-featured model on our desktop (or phone)?
评论 #43429766 未加载
评论 #43430096 未加载
LeoPanthera2 个月前
Maybe Llama 3.3 70B doesn&#x27;t count as running on &quot;one GPU&quot;, but it certainly runs just fine on one Mac, and in my tests it&#x27;s far better at holding onto concepts over a longer conversation than Gemma 3 is, which starts getting confused after about 4000 tokens.
pram2 个月前
Gemma 3 is a lot better at writing for sure, compared to 2, but the big improvement is I can actually use a 32k+ context window and not have it start flipping out with random garbage.
grej2 个月前
It lasted until Mistral released 3.1 Small a week later. Such is the pace of AI...
评论 #43428912 未加载
评论 #43430139 未加载
CamperBob22 个月前
Technically, the 1.58-bit Unsloth quant of DeepSeek R1 runs on a single GPU+128GB of system RAM. It performs amazingly well, but you&#x27;d better not be in a hurry.
评论 #43429180 未加载
评论 #43428094 未加载
评论 #43429836 未加载
评论 #43428182 未加载
m00x2 个月前
I found Mistral Small 3.1, which released slightly after Gemma 3, much better.<p>Much fewer refusals, more accurate, less babbling, generally better, but especially at coding.
ChrisArchitect2 个月前
Google post from last week: <a href="https:&#x2F;&#x2F;blog.google&#x2F;technology&#x2F;developers&#x2F;gemma-3&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.google&#x2F;technology&#x2F;developers&#x2F;gemma-3&#x2F;</a>
pretoriusdre2 个月前
My instinct is that it would be cheaper overall to buy API credits when needed, compared with buying a top-of-the-line GPU which sits idle for most of the day. That also opens up access to larger models.
评论 #43431053 未加载
评论 #43433859 未加载
评论 #43431063 未加载
odysseus2 个月前
Does it run on the severed floor?
评论 #43429051 未加载
williamDafoe2 个月前
Does anyone use GoogleAI? For an AI Company with an AI Ceo using AI language translation, I think their actual GPT products are all terrible and have a terrible rep. And who wants their private conversation shipped back to google for spying?
评论 #43429697 未加载
评论 #43429640 未加载
评论 #43433180 未加载
RandyRanderson2 个月前
So says Gemma 3.
ForTheKidz2 个月前
...until this coming tuesday? ...let&#x27;s talk value.<p>EDIT: I do feel like a fool, thank you.
评论 #43428225 未加载
zeroq2 个月前
I call it the biggest bs since I had my supper.
impure2 个月前
It’s a 27B model, I highly doubt that.
评论 #43428149 未加载
评论 #43429579 未加载
评论 #43429211 未加载