Just train it to make it win. Then, anytime it is a bit in front, perform suboptimal moves, until things are even (those engines typically have a score on whether they are in front or not).
Not sure that would work in an end game though. If you are behind and make a good move, you accidentally win. If you are technically ahead and make a bad move, you probably lose.
If you can get reliably to an end game with only a king and a one or maybe two pieces left on each side, then maybe it can just be hand-scripted and bolted onto what you're proposing, but I'm not sure your proposal would consistently get it there.
We tested it for knowledge extraction against GPT-3. Unfortunately, it's only at the level of GPT-3's smallest ada engine (for which we have found no use). Nowhere near GPT3's strongest model davinci (for which we have a lot of use cases; for their second strongest model curie as well).
Having said that, I am very happy that eleuther.ai works on open-sourcing GPT-x. Eventually, it will be successful.
Hi HN, we are Become and help job candidates get a better job faster.
We write resumes/CVs and cover letters for them and apply to relevant job positions in their name.
In this submission, we are publishing a small tool we use to automate our work: a GPT3-powered rewrite tool for developer work experiences.
Happy to discuss anything related to writing resumes and cover letters with GPT3. Plus what to write to increase the probability for getting a job interview :)
Great service. Our startup offers a related service: we send out job applications for candidates and make sure they don't miss an opportunity: https://cvandme.com
We have seen that job applicants leave so much simple improvements on the table, it is unbelievable :) The most basic thing: Give more details on your technology stack for every experience, be more concise and avoid fancy graphics and formatting.
I am still trying, and achieving, to give them the benefit of the doubt. They understood and took it back.
But I am scratching my head how they could honestly miss the importance of what they were planning to do.. I guess a combination of stress, pressure and usual disregard of privacy by big players clouded their judgement.
Yeah it’s hard to reconcile that discussion and this apology. That is unless they were hemorrhaging users after that email hit and reversed course because of that. I like the idea of triplebyte but I’m a bit hesitant now. Perhaps this is the blindness that people in privileged positions in life can’t see... similar to the real name policy on google that caused a problem for the people that didn’t want their identities tied to it. Gay people that weren’t out of the closet yet or gay people in countries with laws against that or people escaping abusive exes/stalkers etc.
Rather than safeguarding the fact that you are or were job searching, we threatened exposure. Current employers might retaliate if they saw that you were job searching. You did not expect that any personal information you’d given us, in the context of a private, secure job search, would be used publicly without your explicit consent. I sincerely apologize. It was my failure.
So, what happened? How did I screw this up? I’ve been asking myself this question a bunch over the past 48 hours. I can point to two factors (which by no means excuse the decision). The first was that the profiles as spec’d were an evolution of a feature we already had (Triplebyte Certificates--these are not default public). I failed to see the significance of “default public” in my head. The second factor was the speed we were trying to move at to respond to the COVID recession. We’re a hiring company and hiring is in crisis. The floor has fallen out on parts of our business, and other parts are under unprecedented growth. We've been in a state of churn as we quickly try various things to adapt. But I let myself get caught in this rush and did not look critically enough at the features we were shipping. Inexcusably, I ignored our users’ very real privacy concerns. This was a breach of trust not only in the decision, but in my actual thought process. The circumstances don’t excuse this. The privacy violation should have been obvious to me from the beginning, and the fact that I did not see this coming was a major failure on my part.
Our mission at Triplebyte has always been to build a background-blind hiring process. I graduated at the height of the financial crisis as most companies were doing layoffs (similar to what many recent-grads are experiencing today). My LinkedIn profile and resume had nothing on them other than the name of a school few people had heard of. I applied to over 100 jobs the summer after I graduated, and I remember just never hearing back. I know that a lot of people are going through the same thing right now. I finally got my first job at a company that had a coding challenge rather than a resume screen. They cared about what I could do, not what was on my resume. This was a foundational insight for me. It's still the case today, though, that companies rely primarily on resume screens that don’t pick up what most candidates can actually do--making the hiring problem much worse than it needs to be. This is the problem we're trying to fix.
We believed that we could do so by building a better Linkedin profile that was focused on your skills, rather than where you went to school, where you worked, or who you knew. I still believe there's a need for something like this. But to release it as a default public feature was not just a major mistake, it was a betrayal. I'm ashamed and I'm sorry.
Triplebyte can’t function without the trust of the engineering community. Last Friday I lost a big chunk of that trust. We’re now going to try to earn it back. I’m not sure that’s fully possible, but we have to try. What I will do now is slow down, take a step back, and learn the lessons I need to avoid repeating this.
I understand that cancelling this feature does not undo the harm. It’s only one necessary step. Please let me know any other concerns or questions that I can answer (replies to this email go to me). I am sorry to all of you for letting you down.
I'm going to inline this text into the top post so that everyone can read it. (Edit: that's done, and I deleted "continued in comments" - normally I'd ask for permission first, but in this case it seemed better not to wait.)
You probably split the post up this way because the software told you the text was too long. Tip for the future: you can get around that by clicking 'edit' and adding the rest later. Don't tell anybody :)
Would you mind also doing that thing where the comment is collapsed by default? I spent way too long trying to figure out what was different about this text compared to the email or the top post before I skipped down and saw your explanaion.
They likely have a different definition for software developer than myself.