The pace at which AI/ML research is being published is phenomenal. I could see that some wins would be possible just by having a Sonnet or 4o level read the faster/more and combine ideas that haven't been combined before. I would just be concerned about the synthetic of its own papers being able to lead itself astray if they weren't edited by a human ML researcher? Seems like it could produce helpful stuff right now, but I would just want to not mess up our own datasets of ML "research".
You are right on! Teamwork outside of paid work is underrated. So many solo projects/goals/quests stall when a person with a different skill set could've made all the difference and helped bring it to completion. I think in-person community is best for this, although Internet strangers can certainly become friends and do fun projects together.
I actually think that your landing page is on the right track. At least, if I'm your target customer. As someone who has hours and hours of long-form video, this page looks like I can take the videos that I have recorded and mince them into smaller self-contained pieces very easily. For perspective, I am a paid user of Descript and have played with Riverside.fm. This tool intrigues me (I've already signed up and am experimenting) because it looks like it can even auto-detect where B-Roll should go. If this is true, then no other tool really does this that I'm aware and is kind of a killer feature. You also seem adjacent to [AutoPod](https://www.autopod.fm/)
Congrats on the launch! I had a similar issue as the comment above. I put in the prompt "Celtic symphonic rock" (which seems to work on Suno.ai) and some lyrics. The output ended up being just readings of the lyrics without any music, except some artifact-level whispering of music when the voice was silent. Would definitely love to see some demos of what it can produce!
That's very funny. It's like the time that I was in a pub in Cambridge and the gentleman* next to me said, "Y'know, the US and the UK, they go way back. The US used to be a colony of the UK actually." and I laughed, like, "Yes, and we built our entire national identity off of ending our relationship." He was dumbfounded. He was surprised that I even knew details about it. Which makes me laugh just typing this because of how thoroughly ingrained that information is to many Americans.
*Who, to be fair, had actually been educated at the "other place". (I joke as someone who did not go to college.)
Went to Prague last year and saw the very windows that at least one defenstration used. It was so high up and the window sills and leaves were combined about 5ft wide. So it's not like they just knocked someone through glass. They had to really push them off the ledge. Really sad how some hotheads triggered some of the wars in Europe.
If they really thought the guys were so evil, they should have tried them, not defenestrated them. They knew better.
My understanding of the 1618 defenestration is that war was essentially inevitable by that point.
The Holy Roman Empire had always been fragile—more of a loose alliance of princedoms than a proper kingdom or empire—and yet less than one year into his reign Ferdinand II started rolling back the strong religious protections that a large number of his vassals had begun to take for granted. There was no way that Ferdinand's approach to ruling the quarrelsome and religiously diverse German states wouldn't have led to war, even if these specific hotheads had kept their cool in that specific building on that specific day.
If it was the 1618 one (that's the one you can see from the courtyard), they wrestled him out the window and then beat his hands with their scabbards as he hung on to the sill for dear life. Fortunately he landed in a pile of horse manure and suffered only minor injuries.
FWIW, the horse manure story was probably invented as a reaction to the Catholic explanation that they were saved by divine providence. The whole event is so steeped in religious propaganda from both sides that I don't think we can conclusively say what happened except that both did in fact survive.
Very true. What gets even weirder (I'm kind of a 30 years war nerd) was how dynastic concerns always ended up confusing the confessional tensions, so that you had German protestant houses fighting for the Emperor to improve their position vs. the Palatinate, and Catholic France allying with Calvinist Sweden to expand both of their influence in northern Germany.
But then weirdly the Ottoman Sultan was kind of low-key supporting the Habsburgs because his more immediate concern was Bethlen Gabor and if Vienna didn't have to worry about the Ottomans it would have a greater force-in-readiness to deter Bethlen. It's just such a fascinating time with so much of the modern world being laid down by people who were only thinking about immediate concerns.
Wedgewood and Wilson (Wilson's 30 years war book is absolutely amazing, if you haven't read it) agree, though they both spend an entire chapter warning you that Schiller is excellent literature and absolutely irresponsible history.
Oh sure. The way I see it, I just don't buy into "other people murdering for lesser reasons makes my murder ok." I think they had systems of due process, knew due process was important, and decided to yeet people anyway. But yes, I realize people are a product of their times, etc. etc. And I don't think this is a controversial position within Czechia, from what I encountered?
The cathlic government was literally going in and doing ethnic and religious cleansing and basically was willing to systematically kill anybody that would follow the party line.
I see in your bio that you live in or are from Prague. Since you and I see it differently, I would sincerely like to read more up on this event and the history leading up to the 30 years war. If you have anything that you usually recommend, I would gladly take that recommendation.
They did try them. That was why they threw them out the window.
I’m not sure a more level headed trial of the catholics in Prague would’ve prevented the war for long. That particular power struggle could’ve easily boiled over for any reason.
What LLM are you hoping to use. Have you considered using HelixML? If I am reading you right, the primary concern is compute costs, not human-time costs?
We are finding there is a trade-off between model performance and hosting costs post-training. The optimal outcome is where we have a model that performs well on next-token prediction (and some other in-house tasks we've defined) that ultimately results in a model that we can host on the lowest-cost hosting provider rather than be locked in. I think we'd only go the proprietary model route if the model really was that much better. We're just trying to save our selves weeks/months of benchmarking time/costs if there was already an established option in this space.
That said, I think that dvt's comment is helpful about RAG likely being what you need rather than fine-tuning, but wanted to offer something if you know that's what you need.
To your point, I had a season of extremely hard work (for me) where I was on two timezones (one in Europe and one in the US) and ended up in a situation where, due to severe jet lag, I would go to bed at the right time for the timezone I was in, sleep for 12hrs, and wake up *more exhausted* than when I went to sleep. Very demoralizing, but this continued for weeks. Eventually I decided to take a sleep-cation. I went to a friend's house who generously offered a private room. I prepared a couple of meals in the fridge and proceeded to sleep as much as I could for the next three days. I probably slept more than 20 hrs per day, although I didn't necessarily have very long stretches. I only needed like 4-5 meals over that time. It really took me out of a negative spiral and got me back on track. If you actually can sleep, I've seen this approach help people "reset" on their sleep and then get more regular benefits from their sleep hygiene.
reply