HN user
Reasoning models are just wrappers over the base model. It was pretty obvious it wasn’t actually reasoning but rather just refining the results using some kind of reasoning like heuristic. At least that’s what I assumed when they were released and you couldn’t modify the system prompt.
Link to example workflow: https://github.com/edelauna/stock-predictor/blob/prod/.githu...
I wonder what the break even point would be to just have switched to use JVM.
How could you be a human being and not be small D disgruntled?
I know it always gets shutdown, but I’m comfortable with it being done by the US/CIA[1], especially since Satoshi stopped sending emails once Gavin Andresen met with CIA and « took over » the project.
I anticipate it’ll be unclassified at some point, and then really take off, but not before the US exhausts the ability to try and trace malicious transactions they might be focused on. Also the US owning a bunch of the original bitcoins is in line with owning a bunch of the lower IPv4 addresses which haven’t seem much, if any, activity.
[1] https://www.reddit.com/r/CryptoCurrency/comments/mr780k/sato...
Yeah, see GPT 3.5 vs GPT 4 pricing.
I thought this is what GPT 4 was, it uses a boosting algorithm over GPT 3.5?
I think you’re forgetting about the use case where the LLM returns something partially correct to a discerning expert, who is still able to use the response, but does not bother with a message like “btw I had to do X to make your suggestions usable”.
Does go have something similar to spring, or is there a framework baked into the language
I’ve not had any success to make responses deterministic with these settings. I’m even beginning to suspect historic conversations via API are used to influence future responses, so I’m not sure if it’ll truly be possible.
What’s a quality attribute requirement, and how are they usually documented and prioritized?
So where is the secret key stored for signing the JWT? In the front end as well?
Edit: Oh I found it here: https://postgrest.org/en/stable/how-tos/sql-user-management....
That’s a pretty neat design. Also an interesting attack surface
Truth has always been hard to figure out. An old mentor once told me to be skeptical of everything I hear and only trust half of what I see. When I asked about which half to believe, he said that’s something I needed to figure out.
>> slowest part of software delivery is testing
In my experience the slowest part has been marking a feature as done. I loved working at places with QA. I could assign tickets to QA once the PR was up.
Now I gotta build in that I’ll be bumping PRs for review for approximately 30-50% of the time I’m working on a feature.
I feel like I’m missing something about this chatGPT craze… to me the product is chatGPT, and it’s mostly fleshed out (would like to see the ability to upload an image, as part of a conversation), I don’t see the appeal for paying 100x more for every new iteration for a marginal reduction in misinformation.
To me the target user of chatGPT needs to have some level of expertise in the domain their asking a question, as based on how gpt works, chatGPT will eventually produce an output with misinformation, therefore someone familiar with the subject matter would have a higher chance at catching the nuance in a response and be able to continue their conversation.
It’s wild, I actually downgraded back to gpt 3.5 cuz gpt 4 just wasn’t worth the money.
It was written in 2021 apparently but also has
last updated 1993-02-18
at the bottom of the page. Is this letter a form of satire?
What’s a profiler and how do I start using one?
If they’re language specific, then for typescript, Ruby, and Kotlin.
This is largely based on my experience (which may be incomplete), but static code analysis tools in our CI pipeline don’t add comments to a PR, rather generate a report and will sometimes annotate the affected coded.
Annotations are not tracked historically so any new ci run will regenerate the annotations. I thought this was better for chat gpt reviews since they’re non-deterministic and each CI run could result in different types of feedback.
That's great, thanks for the feedback, will update the prompt to be more specific.