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。

Understanding, using, and finetuning Gemma

118 点作者 rasbt大约 1 年前

5 条评论

brucethemoose2大约 1 年前
What are HNers looking for in this article? The architectural differences, or how to run/finetune it?
评论 #39492797 未加载
评论 #39491887 未加载
lopkeny12ko大约 1 年前
Gemma, despite being developed by a company worth billions of dollars, is a phenomonally poor model.<p>I tried the open source release yesterday. I started with the input string &quot;hello&quot; and it responded &quot;I am a new user to this forum and I am looking for 100000000000000...&quot; with zeros repeating forever.<p>Ok, cool I guess. Looks like I&#x27;ll be sticking with GPT-4.
评论 #39494404 未加载
评论 #39494431 未加载
brunooliv大约 1 年前
Anyone who uses these models for more than 10 min will immediately realize that they&#x27;re really, really bad compared to other free, OSS models. Even Phi-2 was giving me &quot;on par&quot; results except that its a model of a different league.<p>Many models are being released now, which is good to keep OpenAI on their toes and not mess up, but, truth be told, I&#x27;ve yet to see _any_ OSS model that I can run on my machine being as good as ChatGPT 3 (not 3.5, not 4, but the original one from when everyone went crazy).<p>My hopes for consumer hardware ChatGPT-3.5 within 2024 probably lie with what Meta will keep building upon.<p>Google was great, once. Now, they&#x27;re a mere bystander in the larger scheme of things. I think that&#x27;s a good thing. Everything in the world is cyclic and ephemeral and Google enjoyed their time while it lasted, but, newer and better things are and will, keep on coming.<p>PS: Completely unrelated, but, gmail is now the only Google product I actively use. I don&#x27;t, genuinely, remember the last time I did a Google Search... When I need to do my own digging I use Phind these days.<p>Times are changing and that&#x27;s great for tech and future generations joining the field and workforce!
评论 #39492947 未加载
评论 #39492882 未加载
评论 #39492865 未加载
评论 #39492899 未加载
评论 #39494432 未加载
Solvency大约 1 年前
Can we just stop talking about Gemini&#x2F;Gemma for at least two years before it&#x27;s improved? In fact, the two-year mark is rather strategic recommendation, because I guarantee it&#x27;ll become vaporware by then anyway with Google&#x27;s track record. It&#x27;s outrageously poorly performing.
评论 #39492864 未加载
behnamoh大约 1 年前
Gemma (and Gemini) are heavily nerfed. Why are they on the news lately?<p>Also, Gemma is a +9B model. I think it&#x27;s not okay that Google compared it with Mistral and Llama 2 (7B) models.<p>Google also took llama.cpp and used it in one of their Github repos without giving credit. Again, not cool.<p>All this hype seems to be backed by Google to boost their models whereas in practice, the models are not that good.<p>Google also made a big claim about Gemini 1.5 1M context window, but at the end of their article they said they&#x27;ll limit it to 128K. So all that 1M flex was for nothing?<p>Not to mention their absurd approach in alignment in image creation.
评论 #39491911 未加载
评论 #39492352 未加载
评论 #39491957 未加载
评论 #39492858 未加载
评论 #39492152 未加载
评论 #39493139 未加载
评论 #39491866 未加载
评论 #39492645 未加载
评论 #39493212 未加载
评论 #39492128 未加载