41–50 of 61 posts
Didn't do a great job on this AP News article on coronavirus - only four bullet points, two of them repeated: https://apnews.com/545af824f44a22f7559c74679a4f1f53 .
The Basic Summarizer has its restrictions. Try Advanced summarizer. It will give better results.
I love the idea. The implementation did not produce the expected results ( article shown on HN - https://amp-economist-com.cdn.ampproject.org/c/s/amp.economi... ). That said. Keep at it. It seems like a viable and valuable service.
On March 9th America’s government awarded a trio of firms $39.7m to design “microreactors” that can supply a few megawatts of power to remote military bases, and be moved quickly by road, rail, sea and air. The idea of small reactors is as old as nuclear power itself.
In July 1951, five months before a reactor in Idaho became the first in the world to produce usable electricity through fission, America began building USS Nautilus, a nuclear-powered submarine.
A report by the army in 2018 said that Holos, a prototype mobile nuclear reactor, would be 62% cheaper than using liquid fuel.
NASA is developing smaller “Kilopower” reactors for space missions, designed to power small lunar outposts.
I’ve noticed an increase in services like this lately. What gives? Is there some sort of ML serverless offering made available on GCP?
This is v2 of the service that I launched last year. I am not aware of any ML serverless offering that does text summarization.
Didn't do a great job on this AP News article on coronavirus - only four bullet points, two of them repeated: https://apnews.com/545af824f44a22f7559c74679a4f1f53 .
Most people have had mild to moderate illness and recovered, but the virus is more serious for those who are older or have other health problems.
The risk of virus transmission from food servers is the same risk as transmission from other infected people, but “one of the concerns in that food servers, like others facing stark choices about insurance and paychecks, may be pressured to work even if they are sick,” she said.
Tests have found high amounts of virus in the throats and noses of people a couple days before they show symptoms. Flu kills about 0.1% of those it infects, so the new virus seems about 10 times more lethal, the National Institutes of Health’s Dr. Anthony Fauci told Congress last week.
The death rate has been higher among people with other health problems -- more than 10% for those with heart disease, for example.
I’ve noticed an increase in services like this lately. What gives? Is there some sort of ML serverless offering made available on GCP?
Not exactly “serverless” but I built something similar with AWS SageMaker, which has elastic inference abilities. it’s rather fast to spin up and down. Also, when it comes to summarization- you don’t really need to infer each run, you can throw up a pretty simple caching system. Which means repeat requests are far cheaper and faster. I used cloudflare workers as a proxy / caching layer with KV in front of an AWS lamb…
It sounds like a joke, I tried to reduce a technical manual but it didn't... well... I don't really know how to expect anything. This however: https://www.latimes.com/world-nation/story/2020-03-13/china-... worked really well.
It isn't really built for manuals etc. It is mainly for blog posts or news articles.
1. From the articles I tried, the summaries seem to be very basic. They don't seem to capture the essential points of the articles it is trying to summarize.
2. I tried the 'advanced summarizer' too. Here again, it seems to have the same problem. Worse, it seems to skip parts of the article, especially if they are beyond a certain length.
3. The landing page is nice. But the product seems to be targeted towards people who want to share a summary of a random blog post rather than try and save time reading the article.
In my opinion, SMMRY, as mentioned in another comment and which I've used since whenever, seems like a much better product. Additionally, SMMRY also gives you the capability of expanding the number of lines of the summary in case you've found the article interesting and want to read a few more details of it, rather than the full thing.
SMMRY: https://smmry.com/
I may try my hand at writing one with the reqs outlined in the top comment just for a fun coding project.
I tried this on three of my own articles. It worked very poorly. There’s a lot of room for improvement before asking people to pay for it IMO. I may try my hand at writing one with the reqs outlined in the top comment just for a fun coding project.
Nice landing page. If you Google "summarizer", you will find dozens of similar services for free. The mechanism behind it is very simple. A couple a years ago I built one from scratch in about 2 hours, then I accidentally deleted it and rewrote it in 15 minutes. Here's how most of them work: 1. Split the text into words 2. Rank each word based on how many times it appears in the text. For example, a word that appears…
The method you described is a part of our algorithms but more steps are needed to make it give meaningful results and make sure it works on different kinds of articles.