If you're an American, the difference is simple. With the Israel conflict, you won't get drafted, sent into a jungle, get napalmed, and sent home in a body bag.
Hopefully it's helpful for OP. I think a great thing about a forum like this is if they use it to learn something from those with more experiences. So if the question is asked in good faith - good for them. If you're on IG or TikTok, you might not get this perspective.
I think when there are 9 steps, a video does make it a bit clearer. I skimmed them and it wasn't totally clear how it worked. Two browser windows, add triple backticks to the code when pasting it, etc.
I'm 65. I've had a bad head reaction to even fairly small amounts of spicy food - basically, a day of brain fog - ever since I was in my early 50s. I was fine before that - I never liked massively spicy food, but I really enjoyed a bit of spice. No one's ever heard of this. It might be related to eustachian tube dysfunction, but it's peculiar and I have no idea how to treat it. I've had to give up Indian food and Szechuan food, and there are places I'm afraid to travel to because I can't eat the food.
Yeah, MCAS can be more mild and present as non-specific symptoms. I wouldn't take it as the gospel truth, but there's some interesting research (and treatment protocols) by Afrin, and unusual non-specific symptoms are one of the main issues he struggled with developing MCAS treatments.
I wish I were watching this with friends and alcohol, because it's remarkable how bad it is. Like scenes don't even cut together properly, and the dialog recording is uneven. Stuff that real projects usually do so well that you don't notice.
How is your relationship and communication? You say it's great, etc - but having to drop hints rather than simply discuss it implies that maybe the relationship has more issues than you've described?
I think if you want to prove anything, you'd have to show a token interpretation that would only be possible with your technique, not an LLM response that could be a hallucination. With a stochastic algorithm, you're also going to get different answers. I don't know if you asked 8B or 70B, but this was the response when I just asked the exact same thing to Llama 3 70B Instruct:
I know he said the 70B model was still improving but they had to stop to get ready for Llama 4, but did he ever say explicitly whether the 8B model had asymptoted or if training loss was still going down?
Would be interested in hearing more about the process. I usually brew decaf, but would be interested in maybe half or quarter caff blends to get some different tastes in there.
I don't really drink dark roasts but I use the same basic approach for anything from light to medium. It is a lower temperature brew overall, but can get hotter towards the end.
Thanks! I'm very curious on the temperature. I brew green tea, and the temperature makes a huge difference - with Japanese teas, people often use way too high a temperature.
Youtube does not care what the content is unfortunately... if the combination of the channel name and/or the profile avatar is deemed too close to an existing channel, and is not made very obvious that is specifically a fan channel, it could be swept up in this policy (which is no doubt at least semi-automated, if not fully)
But apparently the appeal was denied. This is where Google royally screws their audience. I understand if it auto-flags stuff and then you have to jump through manual hoops to get it back - but unless you know someone at Google to advocate for you, you're pretty much screwed.
Resident Alien. The first season was just brilliant fun. And then the second season just screeched off the rails with entire episodes that seemed like a different team or something. Tried to push through to see if it recaptured, but finally tapped out. Never tried the third season.
I just started drinking coffee this week, and I was wondering if it's normal for your urine to suddenly smell like vienna sausages. This has been happening for the few days since I started drinking coffee. Thanks!
That can occasionally happen when people eat meat and can't break down certain compounds. Never heard it related to coffee, though. Are you eating a lot of meat, or could you have some kind of infection (UTI, etc)?
I wouldn't think a week is very long, but also you'd need more consistent linking. Not sure the best way to remember that, but all of those should link to NETWORK, not various versions (I believe you can change it so it displays NETWORKING but links to NETWORK).
Do you think AutoGen is the best way to do this, or something like Kobold? I haven't done much local stuff (limited by hardware) so it seems like AutoGen would allow me to do the LLM calls through API like OpenAI or Anthropic, but still give it access to a local system (either sandboxed or connected to the nuclear launch codes).
I don't have much experience with Kobold except doing some samples, so I don't know the ins and outs of what it can do and what's possible, but correct me if I'm wrong—it's more of a general framework to communicate with all kinds of LLMs, right?
Thanks for that detail! I've only used the API through individual Python apps and such, but some sort of agent structure sounds incredibly promising and much more flexible. I'll have to dive into AutoGen a bit and learn the workflow since obviously that's geared specifically toward agents.
No way to tell anymore. A few times I've thought it was film and looked it up to find it was digital. The only one I can think of offhand in recent times where I didn't already know what it was shot on but looked like beautifully shot film and it was - was Shoplifters.
In all fairness to the articles - plenty of stuff back then was confused with contradicting statements being released. They did not have the benefit of a full trial with discovery and testimony.
Wow, I watched most of the trial but must have missed her saying that. I remember in the ABC interview Baldwin says Halyna told him to point the gun at the camera, obviously that wasn’t confirmed by Joel or the script supervisor. It will be interesting to see what comes out during Baldwins trial.
I definitely was doing way too much agitation for awhile - so I had to grind very coarse to balance that out but didn't realize what I was doing. Also doing decaf is even more finicky.
I’m a maths and science teacher and I’m using obsidian to slowly create a “wiki” of sorts where each note is a concept (or small group of closely related concepts) with notes on common misconceptions, the key mathematical concepts, and places that have really interesting questions and or textbook sections where questions can be found.
How do you use it for recipes? I was thinking of doing this. In Evernote I just have usually links to the page, but that's not always super useful. I did web clippings at one point, but that was too much extra info.
Thanks. I think I'll try doing that as the Evernote ones just have too much extraneous information. I want to know ingredients, instructions, and a few tags like country of origin. I don't need the seven paragraphs about their grandmother's cooking. :)
That we would all have amazing video communicators in our pockets and on our desks and they would be free to use to call anyone in the world…and instead we'd send a text message. ¯\_(ツ)_/¯
This looks great. Can you give more details about how long it took and what framework or language you were using? When you tried GPT4, was it after the 4/9 update or before?
End to end I’d say about 4 hours. Of which half the time i just spent on styling and which map would look good. I used python flask for the backend, html css js for the front end, and the mapbox api.
I’m saying my grandma just came to me and told me about her stocks and asked me what I should do with them. So I asked reddit because reddit has always solved my problems before and I just wanted other peoples opinion. But everybody is saying I told her to dump 100k into riot when I was still a child😂 like what 10th grader knows about RIOT.
I feel like people back then wanted peace more, sure they had some extreme activist, yet the majority of people wanted peace and no war.
Some of that may be true, but it's rarely that simple.
If you're an American, the difference is simple. With the Israel conflict, you won't get drafted, sent into a jungle, get napalmed, and sent home in a body bag.
Hopefully it's helpful for OP. I think a great thing about a forum like this is if they use it to learn something from those with more experiences. So if the question is asked in good faith - good for them. If you're on IG or TikTok, you might not get this perspective.
we need lmsys chat to add something like this
This would be great. I don't see why LMSYS couldn't do it? They wouldn't have to do it for every model, but just add a few quants to the testing.
I’m really lost as to what’s confusing about the very few steps required to test it…
I think when there are 9 steps, a video does make it a bit clearer. I skimmed them and it wasn't totally clear how it worked. Two browser windows, add triple backticks to the code when pasting it, etc.
I'm 65. I've had a bad head reaction to even fairly small amounts of spicy food - basically, a day of brain fog - ever since I was in my early 50s. I was fine before that - I never liked massively spicy food, but I really enjoyed a bit of spice. No one's ever heard of this. It might be related to eustachian tube dysfunction, but it's peculiar and I have no idea how to treat it. I've had to give up Indian food and Szechuan food, and there are places I'm afraid to travel to because I can't eat the food.
Possibly related to mast cell activation syndrome? Many people have it triggered by spicy foods (among other things).
Could be, but the symptoms aren't really like anaphylaxis symptoms. It's just a foggy feeling in the head. I'll look into it.
Yeah, MCAS can be more mild and present as non-specific symptoms. I wouldn't take it as the gospel truth, but there's some interesting research (and treatment protocols) by Afrin, and unusual non-specific symptoms are one of the main issues he struggled with developing MCAS treatments.
It's one of the worst movies I've ever watched. It's honestly worth watching purely because of how bad it is.
I wish I were watching this with friends and alcohol, because it's remarkable how bad it is. Like scenes don't even cut together properly, and the dialog recording is uneven. Stuff that real projects usually do so well that you don't notice.
How is your relationship and communication? You say it's great, etc - but having to drop hints rather than simply discuss it implies that maybe the relationship has more issues than you've described?
He seemed to think the Simpsons was part of a conspiracy to normalize absurdity and got the Frank Grimes ending for it.
Good ol' Grimey. How is he anyways?
I think if you want to prove anything, you'd have to show a token interpretation that would only be possible with your technique, not an LLM response that could be a hallucination. With a stochastic algorithm, you're also going to get different answers. I don't know if you asked 8B or 70B, but this was the response when I just asked the exact same thing to Llama 3 70B Instruct:
I know he said the 70B model was still improving but they had to stop to get ready for Llama 4, but did he ever say explicitly whether the 8B model had asymptoted or if training loss was still going down?
stop it, you're gonna make me c
If only we had something that could predict the next token. I guess we'll never know.
Would be interested in hearing more about the process. I usually brew decaf, but would be interested in maybe half or quarter caff blends to get some different tastes in there.
I don't really drink dark roasts but I use the same basic approach for anything from light to medium. It is a lower temperature brew overall, but can get hotter towards the end.
Thanks! I'm very curious on the temperature. I brew green tea, and the temperature makes a huge difference - with Japanese teas, people often use way too high a temperature.
Youtube does not care what the content is unfortunately... if the combination of the channel name and/or the profile avatar is deemed too close to an existing channel, and is not made very obvious that is specifically a fan channel, it could be swept up in this policy (which is no doubt at least semi-automated, if not fully)
But apparently the appeal was denied. This is where Google royally screws their audience. I understand if it auto-flags stuff and then you have to jump through manual hoops to get it back - but unless you know someone at Google to advocate for you, you're pretty much screwed.
I think that is offensive: to kabuki theater, holes in the ground, and even ducks. None of those deserve to be compared to our rotten politicians.
Resident Alien. The first season was just brilliant fun. And then the second season just screeched off the rails with entire episodes that seemed like a different team or something. Tried to push through to see if it recaptured, but finally tapped out. Never tried the third season.
Probably an unpopular opinion, but Stranger Things. Tried twice to get into the 3rd season, but it was just too much teenage drama.
Totally agree. First season was amazing. Second season was still good. By the third season it just felt like a completely different show.
I just started drinking coffee this week, and I was wondering if it's normal for your urine to suddenly smell like vienna sausages. This has been happening for the few days since I started drinking coffee. Thanks!
That can occasionally happen when people eat meat and can't break down certain compounds. Never heard it related to coffee, though. Are you eating a lot of meat, or could you have some kind of infection (UTI, etc)?
I wouldn't think a week is very long, but also you'd need more consistent linking. Not sure the best way to remember that, but all of those should link to NETWORK, not various versions (I believe you can change it so it displays NETWORKING but links to NETWORK).
I like doing this also :D
Do you think AutoGen is the best way to do this, or something like Kobold? I haven't done much local stuff (limited by hardware) so it seems like AutoGen would allow me to do the LLM calls through API like OpenAI or Anthropic, but still give it access to a local system (either sandboxed or connected to the nuclear launch codes).
I don't have much experience with Kobold except doing some samples, so I don't know the ins and outs of what it can do and what's possible, but correct me if I'm wrong—it's more of a general framework to communicate with all kinds of LLMs, right?
Thanks for that detail! I've only used the API through individual Python apps and such, but some sort of agent structure sounds incredibly promising and much more flexible. I'll have to dive into AutoGen a bit and learn the workflow since obviously that's geared specifically toward agents.
No way to tell anymore. A few times I've thought it was film and looked it up to find it was digital. The only one I can think of offhand in recent times where I didn't already know what it was shot on but looked like beautifully shot film and it was - was Shoplifters.
In all fairness to the articles - plenty of stuff back then was confused with contradicting statements being released. They did not have the benefit of a full trial with discovery and testimony.
Wow, I watched most of the trial but must have missed her saying that. I remember in the ABC interview Baldwin says Halyna told him to point the gun at the camera, obviously that wasn’t confirmed by Joel or the script supervisor. It will be interesting to see what comes out during Baldwins trial.
Yep! It's amazing what came out in the press versus the trial.
Similar here with the C2 and seems like i'm in the middle of the process you've been through.
I definitely was doing way too much agitation for awhile - so I had to grind very coarse to balance that out but didn't realize what I was doing. Also doing decaf is even more finicky.
I like reading a lot, but my Japanese has always been stuck at a mediocre level for a variety of reasons. Congrats on making amazing progress.
I’m a maths and science teacher and I’m using obsidian to slowly create a “wiki” of sorts where each note is a concept (or small group of closely related concepts) with notes on common misconceptions, the key mathematical concepts, and places that have really interesting questions and or textbook sections where questions can be found.
How do you use it for recipes? I was thinking of doing this. In Evernote I just have usually links to the page, but that's not always super useful. I did web clippings at one point, but that was too much extra info.
Most of them are my own so I just write it down. If they are from a webpage. A simple copy paste then format to my liking is enough.
Thanks. I think I'll try doing that as the Evernote ones just have too much extraneous information. I want to know ingredients, instructions, and a few tags like country of origin. I don't need the seven paragraphs about their grandmother's cooking. :)
We don't really understand some of that, beyond certain genetic components.
That we would all have amazing video communicators in our pockets and on our desks and they would be free to use to call anyone in the world…and instead we'd send a text message. ¯\_(ツ)_/¯
This looks great. Can you give more details about how long it took and what framework or language you were using? When you tried GPT4, was it after the 4/9 update or before?
End to end I’d say about 4 hours. Of which half the time i just spent on styling and which map would look good. I used python flask for the backend, html css js for the front end, and the mapbox api.
Really cool.
I’m saying my grandma just came to me and told me about her stocks and asked me what I should do with them. So I asked reddit because reddit has always solved my problems before and I just wanted other peoples opinion. But everybody is saying I told her to dump 100k into riot when I was still a child😂 like what 10th grader knows about RIOT.
You're doing great. Reddit is often very helpful - and simultaneously consistently dickish. So seems like Reddit delivered as usual.