dreaminvm

Happy to see this type of work that is truly open source and commercially usable. Is this the entire corpus or a subset? Do you intend to release any new iterations?

I've been thinking of starting similar efforts at another BigCorp by hosting a UL2 or GPT-J instance.

Here's a recent release of fine-tuning Flan-UL2 on instructions (alpaca). https://medium.com/vmware-data-ml-blog/lora-finetunning-of-u...

Highlighting the deception from the media. It would be like calling Aspirin a dog pain reliever or the Hepatitis vaccine a dog vaccine which while true is meant to mislead and likely framed the people trying alternative therapeutics like Ivermectin as insane. Ivermectin is an anti-parasitic which has been prescribed 3.7 billion times to people. Does it help with Covid? Current data seems to imply it does not.

I agree with your points on ethically testing and prescribing experimental drugs.

The intent of my comment is to highlight the media's focus on one side of wrong decisions which only further divides the public.

* For airplanes, I am pointing out the requirement for masking currently even when masking is not required anywhere in California/New York (2 of the most restrictive states for COVID throughout the pandemic) for indoor. It's pure theatre to have people show up to the airport never wearing a mask and then putting it on (ineffectively no less) and then taking it off immediately when they leave the airport.

* Ivermectin is an anti-parasitic which has been prescribed 3.7 billion times (and won a nobel prize). Does it help with Covid? Current data seems to imply it does not. My callout is on the media calling its a horse dewormer (which while true is meant to mislead).

* On virus mutations, see my comment on flip-flopping.

My point is that they've received basically no criticism for many of their mistakes. Saying vaccines will prevent the spread of COVID-19 makes no sense if they've not proven so ever, why make a massive policy decision which likely contributed to further spread (many restrictions were lifted due to their announcement, only to be put back later). Or as another example, they continued to require masks for children even as many countries removed that requirement long ago with data showing that it was safe to do so. Or that they mistakenly counted non-Covid related deaths (72k deaths) in their mortality data.

I call out flip-flopping as it reduces what little confidence the public has in their policy decisions and further empowers the fringe in discussions of ineffective treatments. No one is saying they should know on Day 1 of anything. That's a nonsense expectation. I do however expect them to communicate better and to educate the public on the evolution of the disease rather than rushing to any decisions until the data is clear.

Not defending Ivermectin, but what do people have against trying alternative therapeutics for COVID? I rarely see articles bashing the "other side" for being wrong on many COVID decisions:

CDC flip-flopping on masks

CDC flip-flopping on how vaccine prevents spreads

Airplanes being the safest with high quality filters, yet they are the only place where masking is still required (which even the airlines disagree with)

CNN calling Ivermectin a horse dewormer, when Ivermectin has been administered to a large percentage of the global population

... the list goes on and on

*edit for formatting

If not now, then when?

UK studies show IFR for COVID for those vaccinated/immunized is now lower than seasonal flu [1]. We never wore masks during the flu and I am not sure why anyone would advocate this continuing. Moreover if someone wants to wear a mask, nothing stops them from doing that, but don't force it on everyone ages 2 and up.

https://www.ft.com/content/e26c93a0-90e7-4dec-a796-3e25e94bc... [1]

Solid results with exceptional margins and a healthy FY21 outlook. I wonder if markets just have unrealistic expectations (from the craziness of GME/AMC and cannbis stocks) or people are just taking profits on a ~400% run up since IPO.

They cherry-pick rare cases and use their fleet to get more examples of these situations. This seems like the right approach given more miles following the same car in a straight line is pretty useless.

My takeaway from the presentation is that Tesla will perform better than other companies in this space (although I don't know enough about Waymo to comment) due to the following:

-You want a large dataset (Tesla and many companies have this and can simulate) -You want a varied/diverse dataset (Tesla and many companies have this and can simulate)--the point here is simulations for simple cases work (you can only simulate when you know), but for complex ones are close to the difficulty of actual FSD -You want a real dataset (Tesla is the only company who can say this and can say they have data on how X00Ks of drivers will handle these situations)

I think people underestimate the potential benefits of being a good partner to the Dept of Defense. While facial recognition maybe a rounding error in Amazon's revenue, there are large cloud infrastructure contracts that could be won with the DoD. This sort of partnership on FR could build trust and position Amazon as a willing (and front-running) partner for the DoD (and its many challenges in modernizing data centers and migrating to the cloud).

I am not sure that's a fair statement. The other manufacturers (most, but especially in the luxury segments) build higher quality cars with much lower rates of error. Tesla enthusiasts (much like early adopters of things) are forgiving and overlook a lot of imperfections (fit and finish on teslas vary greatly -see Youtube reviews). This has been the primary reason holding me back for purchasing one.

Tesla's existing accomplishments and ability to meet tough goals is admirable, but their first to market advantage is starting to fade away as more companies bring to market competitive cars without the quality issues (however few they might be pumping out for now).

Link to paper for those that are curious: https://arxiv.org/abs/1801.06146

Not FAANG, but in SV.

* What do you do?

-Product management

* What is your job title?

-Senior Product Manager

* What is your total comp?

-To be super clear:

  -Total is ~$250k
    -150K base salary
    -50K in RSU (basically equivalent to cash for public company)
    -7-10k from ESPP
    -40k in bonus

-N/A

* Where do you live (assuming you don't work from home)?

-Bay Area

* How long did it take you to get here?

-Hired after internship. Its been 3.5 years at current company.

* How did you get here (networking/raw technical skill/job board/dumb luck, etc.)?

-Started as an engineer and quickly took the lead on a high visibility project. Lots of politics in between.

Total comp in the 3.5 years: year 1: 145k year 2: 195k year 3: 225k year 4: expected to be 250k+

Amazingly this summarizes my teenage years with Runescape.

Not only did I pick up botting and development of a private server, but I learn a lot about how virtual economies worked and how to maximize value buying small quantities and sell in bulk in Varrock.

Was the rumored $30B offer from Google ever serious? This news only popped up recently with SNAP stock nosediving before lockup expiry for employees.

Another reason for turning down a buyout situation is the terms. They (Evan and the VCs) would not have such favorable terms as the IPO where they received substantial stock grants, liquidation preferences and retain absolute control over the company without requiring a large (any) financial interest.

Seems like a keynote that was thrown together last minute. Full of awkwardness and lacking any substance.

I think the OP's point was that even someone who is as powerful and wealthy as Thiel could get ostracized in a community of his peers by taking an unpopular position on a political issue, hence the average Joe could do much worse.

Because he is not the most likable CEO and not everyone is looking for a payday.

I am merely pointing out that his security costs are negligible to him/Snap Inc based on their valuation.

The only point I am arguing here is that the popular vote for any election is meaningless (except an electoral tie) when the POTUS is decided by the electoral colleges.

If the popular vote decided the POTUS, then Trump voters in traditionally Democratic states (think Pacific NorthWest) might have turned out in droves and vice-versa for Hilary.

If Snap Inc. IPOs at their expected 25B value, Spiegel will be valued at nearly 5 billion (21% ownership), so ~1 million in security per year is not too far-fetched.

I really don't see the value in reiterating this statistic over and over again.

The POTUS is decided by the electoral vote. If the process was by the popular vote, the candidates would have campaigned very differently and the voter turnout could be much higher.

I could not agree more. People really need to let go of the provocative candidate Trump (the stock market did today) and focus on the impact that's coming.

As much as his path to the nomination and then the presidency is a surprise, his call for unity was more presidential than he's ever been. I think it would be short-sighted to immediately discount his presidency as a failure without seeing the people and actions that he puts into place.

I've been using Google Finance for a long time and did not know about the "e" (extended trading) feature. Thanks!

Even if on average most citizens are peaceful, it only takes one ambitious lunatic to tip the scale in the wrong direction [cue Trump/Putin/Kim].

While I don't necessarily agree with the large US military spending, I feel better knowing we have more national defense than necessary rather than not enough/just enough.

The reported number from Cisco is 5500 (7%) per http://www.wsj.com/articles/cisco-announces-plans-to-cut-5-5....

The 7% number is more of a norm for Cisco at the end of each fiscal year. I wonder why the big gap between the Reuters source and Cisco's own reports.

Not sure if you are trolling, but why would Postgres want to support mysqldump? Postgres has pg_dump and MySQL has mysqldump. If you are comparing the capabilities of pg_dump vs mysqldump, that's a different discussion.

People were also skeptical about Facebook turning a profit and have a PE less than 100. Now Facebook is literally printing 2B per quarter with a more sane PE (~60) as growth continues to accelerate.

Only time will tell, but if Amazon continues bring users into their Prime ecosystem and AWS maintains its position, there's plenty of room to increase profits.

If they can bring down shipping costs either through their own trucks or drones, then the eCommerce business is quite a bit more profitable than it is now. Along with continued dominance in the public cloud business, Amazon investors are going to get a really nice payout.

This is a common occurrence with Groupon.

If you call the shop and mention using Groupon, most of the time they ask you not to purchase the Groupon. They still honor the discount, but avoid a 50% take by Groupon.

Are you saying 150k salaries for new grads or total comp?

ASAIK, Goog/Uber/FB starting salaries are closer to 110k base with $40-50k RSUs vesting per year (refreshes each year), performance bonus 10-15% salary and sometimes a generous signing bonus.