851–860 of 1001 posts
This is incredible. In April I used the standard GPT-4 model via ChatGPT to help me reverse engineer the binary bluetooth protocol used by my kitchen fan to integrate it into Home Assistant. It was helpful in a rubber duck way, but could not determine the pattern used to transmit the remaining runtime of the fan in a certain mode. Initial prompt here [0] I pasted the same prompt into o1-preview and o1-mini and both c…
If I pay for the chain of thought, I want to see the chain of thought. Simple. How would I know if it happened at all? Trust OpenAI? LOL
The real irony is how closed "Open"AI is... but that's not news.
o1 Maybe they should spend some of their billions on marketing people. Gpt4o was a stretch. Wtf is o1
I asked it "I was watching a show and in the subtitles an umlaut u was rendered as 1/4, i.e. a single character that said 1/4. Why would this happen?"
and it gave a pretty thorough explanation of exactly which encoding issue was to blame.
https://chatgpt.com/share/66e37145-72bc-800a-be7b-f7c76471a1...
Earlier quoted context omitted.
> Hopefully the open source community will figure out how to effectively match and replicate what OpenAI is doing. No need for that, there is a Reflection 70B model that does the exact same thing - with chains of thought being separated from the "final answer" via custom 'tag' tokens.
That reflection model is in no way comparable to whatever OpenAI is doing.
First shot, I gave it a medium-difficulty math problem, something I actually wanted the answer to (derive the KL divergence between two Laplace distributions). It thought for a long time, and still got it wrong, producing a plausible but wrong answer. After some prodding, it revised itself and then got it wrong again. I still feel that I can't rely on these systems.
4o was struggling so I gave up. Tried o1 on it, and after trying nearly 15 prompts back and forth helping it along the way we're still far from correct. It's hard to tell if it's much better, but at least my intuition from this feel like this is pretty incremental.
Some practical notes from digging around in their documentation: In order to get access to this, you need to be on their tier 5 level, which requires $1,000 total paid and 30+ days since first successful payment. Pricing is $15.00 / 1M input tokens and $60.00 / 1M output tokens. Context window is 128k token, max output is 32,768 tokens. There is also a mini version with double the maximum output tokens (65,536 tokens…
Tier 5 level required for _API access_. ChatGPT Plus users, for example, also have access to the o1 models.
Earlier quoted context omitted.
> Did the 80% accuracy test results take 10 seconds of compute? 10 minutes? 10 hours? 10 days? It's impossible to say with the data they've given us. The gist of the answer is hiding in plain sight: it took so long, on an exponential cost function, that they couldn't afford to explore any further. The better their max demonstrated accuracy, the more impressive this report is. So why stop where they did? Why omit actu…
See also: them still sitting on Sora seven months after announcing it. They've never given any indication whatsoever of how much compute it uses, so it may be impossible to release in its current state without charging an exorbitant amount of money per generation. We do know from people who have used it that it takes between 10 and 20 minutes to render a shot, but how much hardware is being tied up during that time i…
Earlier quoted context omitted.
The calculator didn’t eliminate math majors. Excel and accounting software didn’t eliminate accountants and CPAs. These are all just tools. I spend very little of my overall time at work actually coding. It’s a nice treat when I get a day where that’s all I do. From my limited work with Copilot so far, the user still needs to know what they’re doing. I have 0 faith a product owner, without a coding background, can us…
Accounting mechanization is a good example of how unpredictable it can be. Initially there were armies of "accountants" (what we now call bookkeepers), mostly doing basic tasks of collecting data and making it fit something useful. When mechanization appeared, the profession split into bookkeeping and accounting. Bookkeeping became a job for women as it was more boring and could be paid lower salaries (we're in the 1…
Aren't LLMs much more limited on the amount of output tokens than input tokens? For example, GPT-4o seems to support only up to 16 K output tokens. I'm not completely sure what the reason is, but I wonder how that interacts with Chain-of-Thought reasoning.
Not really. There's no fundamental difference between input and output tokens technically. The internal model space is exactly the same after evaluating some given set of token, no matter which of them were produced by the prompter or the model. The 16k output token limit is just an arbitrary limit in the chatgpt interface.
It is a hard limit in the API too, although frankly I have never seen an API output go over 700 tokens.