Hacker Newsnew | past | comments | ask | show | jobs | submit | more sbohacek's commentslogin

The 2022 Calf Canyon/Hermits Peak Fire in New Mexico was caused by controlled burns that got out of control. One part of the fire was from a controlled burn that was completed in January and then "reignited" in April!

In any case, some regions might have frequent episodes that are hot, dry, and/or windy so that techniques that easily worked a hundred years ago are no longer useful.


I have not tried, but I assumed that API access to Anthropic's Claude is available through AWS Bedrock.


Which I think is the case now but the beta access to Claude 2 has had a signup on their site for months. I am not as interested in having to go through AWS Bedrock before even experience the potential performance of the API. I give a lot of praise to OpenAI for how quickly they are both scaling and releasing.


I have not been able to get good results either, but I have not tried it in the past year. I also tried many of the architectures in Darts. I have found that fairly straightforward architectures work well. That is, I can iterate on my own design for my own specific data (with all its specific covariates) and get better results than I could with Darts or Prophet.


I don't think so. NSF Small Business Technology Transfer (STTR) and NSF Small Business Innovation Research (SBIR) grants include getting products to market. But this is a regular research grant from NSF Secure and Trustworthy Cyberspace (SaTC)[1].

The SaTC does have a Transition to Practice (TTP) option. However, this research is CORE (see the text "CORE" in the project title [2]). The objective is to write research papers.

[1] https://www.nsf.gov/pubs/2022/nsf22517/nsf22517.htm [2] https://www.nsf.gov/awardsearch/showAward?AWD_ID=2245223&His...


Holy shit, so 1.2 mil to develop this and write some papers? This is some oceans eleven shit right here lmfao.

That’s amazing!


Thanks for the correction!


Be sure that you select GPT-4 from the drop down list of models. For each new chat, it reverts back to default GPT-3.5.


I don't see any dropdowns for such a thing.


I'm currently grading an open book test as you describe. It turns out that someone put their attempt at answers on chegg.com shortly after I posted the test. The temptation to use chegg is too great for students to resist. When chegg has the wrong solution (which is often the case), students will doubt themselves and will go with the wrong chegg answer.

To be clear, the only goal of chegg.com is to help students cheat. The world would be a better place if chegg and its copies did not exist.


My solution to this is to use version control and have them record an explanation of their work. If they copy from chegg they also have to forge a commit history, as well as explain the code line by line. I’d like to see them do that without learning anything.


This is a common thing in computer art e.g. demoscene competitions too.


What about seeding the system and putting your own (bad) answers up there ahead of time?


The university I work for is unable to employ and pay students that are not in the country. As a result, training students online while they are outside of the country is not possible for graduate students that do not have families to support them. This is especially ridiculous since the first one or two years of a PhD is mostly in class training. Yes, this is xenophobia. The purpose of the visa is to train foreign students, without a visa, training cannot happen (except for the wealthy students).


While this is anecdotal, it indicates that people with no symptoms can have lung damage (pneumonia), but not necessarily long-term damage:

https://www.nytimes.com/2020/04/20/opinion/coronavirus-testi...


Having pneumonia is not having no symptoms. That would be on the low end of severe in the case category. Right?


Instead of using GPS, consider using the WiFi base stations. Specifically, each location can be characterized by the set of WiFi base stations a phone can detect. GPS is useful while outdoors, but virus transmission is somewhat difficult outdoors. Indoors, a conference room on the third floor and the 40th floor will have the same GPS coordinates, but a phone in each location will detect a different set of WIFI based stations. This paper shows how WiFi base stations can be used https://arxiv.org/pdf/1610.04730.pdf.

I might not understand methods to achieve privacy, but here are some thoughts. 1. The data could be stored more safely with something like Intel SGX, where only the application can access the data. In this scenario, the carrier (or healthcare worker), uploads the carrier’s path into SGX-based database. Then, individual users that are concerned about their risk could use the app to upload their location paths into the SGX-based system and learn if they are at risk as a simple yes/no. (I have never built an SGX application, so I might be mistaken on its abilities.) 2. I don’t think this is possible: “The solution is a ‘pull’ model where users can download encrypted location information about carriers” If the application is on my device, I can decompile it and get the decryption key or use other methods to dump the carriers’ location data to disk. 3. It seems that the user’s data is also stored on the device. This data is then at risk of being stolen by malicious applications. Instead, the location data can be encrypted with a public key that can only be decrypted on the SGX-protected servers.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: