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。

OpenAI API

516 点作者 gdb将近 5 年前

33 条评论

wildermuthn将近 5 年前
In one of their examples, they note “They saw ratings hover around 60% with their original, in-house tech — this improved by 7-8% with GPT-2 — and is now in the 80-90% range with the API.”<p>Bloomberg reports the API is based on GPT-3 and “other language models”.<p>If that’s true, this is a big deal, and it epitomizes OpenAI’s namesake. The largest NLP models require vast corporate resources to train, let alone put into production. Offering the largest model ever trained (with near-Turing results for some tasks) is a democratization of technology that would otherwise have been restricted to well-funded organizations.<p>Although the devil will be in the details of pricing and performance, this is a step worthy of respect. And it bodes well for the future.
评论 #23492762 未加载
评论 #23498026 未加载
评论 #23491829 未加载
评论 #23492961 未加载
andyljones将近 5 年前
Concrete numbers from the various pullouts:<p>&gt; They saw ratings hover around 60% with their original, in-house tech — this improved by 7-8% with GPT-2 — and is now in the 80-90% range with the API.<p>&gt; The F1 score of its crisis classifier went up from .76 to .86, and the accuracy went up to 96%.<p>&gt; With OpenAI, Algolia was able to answer complex natural language questions accurately 4x as often as it was using BERT.<p>I think the most informative are the first two, but the most _important_ is the final comparison with BERT (a Google model). I am, uh, a little worried about how fast things will progress if language models go from a fun lil research problem to a killer app for your cloud platform. $10m per training run isn&#x27;t much in the face of a $100bn gigatech R&amp;D budget.
评论 #23490591 未加载
minimaxir将近 5 年前
Since the demos on this page use zero-shot learning and the used model has a 2020-05-03 timestamp, that implies this API is using some form of GPT-3: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=23345379" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=23345379</a> (EDIT: the accompanying blog post confirms that: <a href="https:&#x2F;&#x2F;openai.com&#x2F;blog&#x2F;openai-api&#x2F;" rel="nofollow">https:&#x2F;&#x2F;openai.com&#x2F;blog&#x2F;openai-api&#x2F;</a> )<p>Recently, OpenAI set the GPT-3 GitHub repo to read-only: <a href="https:&#x2F;&#x2F;github.com&#x2F;openai&#x2F;gpt-3" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;openai&#x2F;gpt-3</a><p>Taken together, this seems to imply that GPT-3 was more intended for a SaaS such as this, and it&#x27;s less likely that it will be open-sourced like GPT-2 was.
评论 #23490881 未加载
评论 #23494990 未加载
zitterbewegung将近 5 年前
Looks like OpenAI is going head to head with huggingface.<p>This makes a lot of sense and it seems they are telegraphing to monetize what they have been doing. It also seems like this is why they don&#x27;t release their models in a timely manner.
评论 #23490604 未加载
mcrider将近 5 年前
Whoa -- Speech to bash commands? That&#x27;s a pretty novel idea to me with my limited awareness of NLP. I could see this same idea in a lot of technical applications -- Provisioning cloud infrastructure, creating a database query.. Very cool!
评论 #23490864 未加载
评论 #23496825 未加载
评论 #23490995 未加载
typon将近 5 年前
This is incredible. I can&#x27;t tell how much this is cherry-picked examples vs. revolutionary new tech.
评论 #23491778 未加载
say_it_as_it_is将近 5 年前
OpenAI started off wide-eyed and idealistic but it made the mistake of taking on investors for a non-profit mission. A non-profit requires sponsors, not investors. Investors have a fiduciary responsibility to maximize profits, not achieve social missions of open AI for all.
评论 #23492260 未加载
m_ke将近 5 年前
I guess Sama plans on manufacturing growth metrics by forcing YC companies to pretend that they&#x27;re using this.<p>Generic machine learning APIs are a shitty business to get into unless you plan on hiring a huge sales team and selling to dinosaurs or doing a ton of custom consulting work, which doesn&#x27;t scale the way VCs like it to. Anybody who will have enough know how to use their API properly can jus grab an open source model and tune it on their own data.<p>If they plan on commercializing things they should focus on building real products.
评论 #23490207 未加载
评论 #23492505 未加载
评论 #23490168 未加载
评论 #23490167 未加载
eggsnbacon1将近 5 年前
OpenAI started as a non-profit, went for-profit. Still owned by the big players.... Something isn&#x27;t right.<p>Is OpenAI just a submarine so the tech giants can do unethical research without taking blame??? Its textbook misdirection, nonprofit and &quot;Open&quot; in the name, hero-esque mission statement. How do you make the mental leap from &quot;we&#x27;re non-profit and we won&#x27;t release things too dangerous&quot; to &quot;JK we&#x27;re for-profit and now that GPT is good enough to use its for sale!!&quot;. You don&#x27;t. This was the plan the whole time.<p>GPT and facial recognition used for shady shit? Blame OpenAI. Not the consortium of tech giants that directly own it. It may just be a conspiracy theory but something smells very rotten to me. Like OpenAI is a simple front so big names can dodge culpability for their research.
评论 #23492186 未加载
评论 #23491451 未加载
评论 #23492035 未加载
评论 #23493885 未加载
评论 #23492448 未加载
评论 #23492210 未加载
评论 #23493624 未加载
评论 #23492442 未加载
评论 #23491486 未加载
评论 #23491879 未加载
gumby将近 5 年前
I miss the opposite: the old openAI gym and other testbeds. I still don’t know why they shut those down.<p>What alternatives do people like?
nutanc将近 5 年前
Why are there no live examples on the page. All I see is video presentations and some cached API response.<p>Is it a confidence problem? Are the OpenAI folks not confident on a single use case? Or did I miss the live demo somewhere?
评论 #23496157 未加载
sytelus将近 5 年前
Natural language search is approximately $100B business. This might be first AI application that changes the search landscape from 1990s and finally puts an end to the question “where is money in AI?”.
krallistic将近 5 年前
I wonder if there are any legal complications in the transition from a non-profit to a regular company (especially from a tax perspective)
d_burfoot将近 5 年前
In NLP there is a very clear and powerful new paradigm: train a HUGE language model using vast amounts of raw text. Then to solve the problem of interest, either fine-tune the model by training on your specific dataset (usually quite small), or 0&#x2F;1-shot the learning somehow.<p>The crucial question is : is this paradigm viable for OTHER types of data?<p>My hypothesis is YES. If you train a HUGE image model using vast quantities of raw images, you will then be able to REUSE that model to work for specific computer vision problems, either by fine-tuning or 0&#x2F;1-shotting.<p>I&#x27;m especially optimistic that this paradigm will work for image streams from autonomous vehicles. Classic supervised learning has proved to be difficult if not impossible to get to work for AV vision, so the new paradigm could be a game-changer.
评论 #23494983 未加载
评论 #23496537 未加载
sytse将近 5 年前
An API that will try to answer any natural language question is a mind blowing idea. This is a universal thinking interface more than an application programming one.
dalys将近 5 年前
I just sent in a request to join the waiting list, for the company I work at, Kognity. The potential for this in the EdTech field is mindblowingly amazing!<p>There are a few good examples of educational help on the list but it&#x27;s really only scratching the surface.<p>I&#x27;m really excited and hope Kognity and EdTech in general can use this for even more value-full (both for students and teachers) tasks soon.
owenshen24将近 5 年前
Seems potentially more simple to get up and running then the Azure and Google Cloud alternatives which seemed involved when I last tried them.
kamikazehosaki将近 5 年前
OpenAI seems like a completely disingenuous organization. They have some of the best talent in Machine Learning, but the leadership seems completely clueless.<p>1) (on cluelessness) If Sama&#x2F;GDB were as smart as they claim to be, would they not have realized it is impossible to run a non profit research lab which is effectively trying &quot;to compete&quot; with DeepMind.<p>2) (on disingenuity) The original openAI charter made OpenAI an organization that was trying to save the world from nefarious actors and uses of AI. Who were such users? To me it seemed like, entities with vastly superior compute resources who were using the latest AI technologies for presumably profit oriented goals. There are few organizations in the world like that, namely FAANG, and their international counterparts. Originally OpenAI sounded incredibly appealing to me, and a lot of us here. But if their leadership had more forethought, they would perhaps not have made this promise. But given the press, and the money they accrued, it has now become impossible to go back on this charter. So the only way to get themselves out of the whole they dug into was by making it into a for profit research lab. And by commercializing perhaps a more superior version of the tools Microsoft, Google and the other large AI organizations are commercializing, is OpenAI any different from them?<p>How do we know OpenAI will not be the bad actor that is going to abuse AI given their self interest?<p>All we have is their charter to go by. But given how they are constantly &quot;re-inventing&quot; their organizational structure, what grounds do we have to trust them?<p>Do we perhaps need a new Open OpenAI? One that we can actually trust? One that is actually transparent with their research process? One that actually releases their code, and papers and has no interest in commercializing that? Oh, that&#x27;s right, we already have that -- research labs at AI focused schools like MIT, Stanford, BAIR and CMU.<p>I am quite wary of this organization, and I would encourage other HN readers to think more careful about what they are doing here.
评论 #23491441 未加载
LockAndLol将近 5 年前
It&#x27;d be great if OpenAI also introduced CAPTCHA. I&#x27;d be much more willing and understanding to resolve those than anything Google makes.
agakshat将近 5 年前
It’s been a long time coming, but I am curious to see how OpenAI’s research output is directed and impacted by market forces.
nick_araph将近 5 年前
It seems like a step towards OpenAI becoming something like a utility provider for AI capabilities
lerax将近 5 年前
Natural Language Shell seems fun
alphagrep12345将近 5 年前
Interesting to see this. Is this similar to Google and Azure&#x27;s ML apis?
mcemilg将近 5 年前
From AGI to money machine...
zmitri将近 5 年前
What happened to working on AI for the good of humanity, including AGI, and making sure it didn’t fall into the hands of bad actors? Wasn’t that the original aspiration? Now this reads like next generation Intercom&#x2F;Olark tools.
danielscrubs将近 5 年前
It seems it has translation. How does it compare to Google Translate?
dmvaldman将近 5 年前
AGI in text is &lt; 3yrs away.
评论 #23491853 未加载
评论 #23532810 未加载
评论 #23491856 未加载
ericlewis将近 5 年前
Exciting!
historyremade将近 5 年前
&quot;Powered by Azure&quot; -- Elon clearly distrust Amazon.
评论 #23490228 未加载
Grimm1将近 5 年前
Awesome! Just signed onto the wait list.
jaimex2将近 5 年前
&quot;OpenAI technology, just an HTTPS call away&quot;<p>&#x27;an&#x27; is only mean to proceed a vowel. Should say<p>&quot;OpenAI technology, just a HTTPS call away&quot;
评论 #23495198 未加载
评论 #23495220 未加载
organicfigs将近 5 年前
On a side note, has anyone noticed a lack of diversity on the group photo on their careers page: <a href="https:&#x2F;&#x2F;openai.com&#x2F;content&#x2F;images&#x2F;2020&#x2F;04&#x2F;openai-offsite-july-2019.jpg" rel="nofollow">https:&#x2F;&#x2F;openai.com&#x2F;content&#x2F;images&#x2F;2020&#x2F;04&#x2F;openai-offsite-jul...</a><p>I remember coming across it not too long ago and felt unwelcomed&#x2F;disappointed.
评论 #23494337 未加载
评论 #23494284 未加载
评论 #23494319 未加载
评论 #23494328 未加载
brainless将近 5 年前
This is what I submitted for beta list:<p>I want to create a software that can generate new code given business case hints, by studying existing open source code and their documentation.<p>I know this is vague, but sounds like what we eventually want for ourselves right?
评论 #23490707 未加载
评论 #23490255 未加载