Hacker News new | past | comments | ask | show | jobs | submit | pheme1's comments login

Another idea: an online json editor


Are these not already phishing sites of some kind?

I used to have some co-workers who would dump json docs containing sensitive information into these sites all the time, and despite showing them how to format stuff in VS Code.


Wouldn't be hard to adapt this to be literally one of your VS Code extensions.

https://github.com/microsoft/vscode-extension-samples/blob/m...


There's seems to have a lot of news crawler APIs lately, a quite google and found these:

[1] https://currentsapi.services/en/product/price

[2] https://aylien.com/news-api/

[3] https://www.cityfalcon.com/products/api/financial-news



What does the normalized news means? Able to query from a predefined RSS feed? Because I was expecting news coverage from different partisan source pointing to the same news event.


I wish to see some kind of encryption and token access for each module. A quick glance of the setup.py, it seems to rely on appdirs to store user data unencrypted?


appdirs is just a module to find the config directory in a portable way, the data can be stored anywhere you point the config to, but you're right that there is no encryption involved.

Note that it's meant to be running on your own computer, and it's using the filesystem to access the data, no network interactions involved. Ultimately it's about how you're protecting the data on your disk and whether you trust youself with it. Of course, my code has to be trusted too, but it's at least possible to run it in a sandbox/use Docker, etc.

The modules run untrusted Python code, so if you keep your token on the disk, they can potentially steal it token too, unless you use some elaborate authentication system.


Based on your paper, it seems you didn't compare to other model based evaluation method such as Frechet Embedding Distance (FED). I would certainly like to see correlation study between each model evaluation method.


Thanks for your feedback, while we couldn't compare with all model-based methods out there, YiSi (August 2019) and BERTScore which was just presented at ICLR2020 this week (April 27th 2020) are very strong methods to compare against and reflect state-of-the-art. All comparisons are welcome!


As someone who thinker with Text to Speech (TTS), I can say this apply to TTS as well. Good model such as Tacotron2 rarely scale beyond clean ( good text and speech alignmen ) large ( > 12 hours ) datasets.


Just sharing my friend experience, she had to end her exchange study early from Europe (due to recent escalation) and back to Taiwan. My friend however, wasn't a Taiwanese and require a place for her 14 days self isolation. She ended up finding an Airbnb willing to host her during the self isolation period (hotel usually have a central ventilation which isn't optimal for self isolation).


> I can tell you that the industry is much more appreciative of each other's competitors

Totally agree, back in 2012 Intel, TSMC, Samsung pool together roughly 5 billions of investment [1] in ASML to manufacture the EUV machines needed to manufacture 5nm today.

[1] https://www.asml.com/en/news/press-releases/samsung-joins-as...


A global recession is expected to happen this year due to coronavirus [1,2]. So it reminded me of this "trend" during the 2008 recession.

[1] https://www.marketwatch.com/story/global-recession-is-expect...

[2] https://www.cnbc.com/2020/03/17/morgan-stanley-a-global-rece...


A similar discussion can be found here [1] as well which also got to front page as well. The content is strikingly similar : detail OCR introduction and goes to API introduction.

[1]: https://news.ycombinator.com/item?id=21626025


Consider applying for YC's W25 batch! Applications are open till Nov 12.

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

Search: