mm_maybe

OMG a conspiracy by the DNC to elect an actual registered Democrat instead of an independent, how shocking /s

seriously, what did you expect?

the bernie bros have been out in force since the election shouting down anyone who disagrees with their vision of a populist party run by an old, out of touch white man. why does that sound familiar /s

I hate to say this, but it's too fucking late for any sort of electoral solution, at least on a national level. At this point we need to develop local capacity to secure our progressive enclaves in liberal states and big metropolitan areas. This means rooting out the conservative factions within our police and firing bad actors, as well as training community-based self-defense forces. We also need to build more effective local resilience and mutual aid networks so that the Federal government can't dangle emergency aid over us when pandemics and climate-related disasters recur, as they frequently will. We need politicians who will defend our territory, boundaries and values, and fight to discontinue the tremendous subsidy we pay to MAGA territories. We need more self-sufficient economies that trade primarily with healthy, rights-respecting economies in Canada and Europe. It's a heavy lift and unfortunately I don't see anyone up to the task right now, as even Bernie, Warren, and AOC are stubborn institutionalists who will continue trying to work within the system even as it works against them and their constituents.

this is learning completely the wrong lesson. it has been well-known for a long time and very well demonstrated that smaller models trained on better-curated data can outperform larger ones trained using brute force "scaling". this idea that "bigger is better" needs to die, quickly, or else we're headed towards not only an AI winter but an even worse climate catastrophe as the energy requirements of AI inference on huge models obliterate progress on decarbonization overall.

those are all classification problems, which is a fundamentally different kind of problem with less open-ended solutions, so it's not surprising that they are easier to train and deploy.

I live in a rural part of a deep blue Northeast state and have been thinking about this a lot. Most of my surrounding area is predictably liberal college towns but the town next door to me is very MAGA and I have to drive through it to get to the highway. Honestly, I want to know what it takes to get those people to leave so that we can secure and expand a safe haven here...

I really wish it were easier to fine-tune and run inference on GPT-J-6B as well... that was a gem of a base model for research purposes, and for a hot minute circa Dolly there were finally some signs it would become more feasible to run locally. But all the effort going into llama.cpp and GGUF kinda left GPT-J behind. GPT4All used to support it, I think, but last I checked the documentation had huge holes as to how exactly that's done.

ok but... the Democrats in swing states (hell, any state) who said "I'm staying home because Kamala courts Republicans, even though I know Trump is Hitler" are absolutely as shitty as the people who voted for Trump

there's a lot of racist and misogynistic democrats

One of the reasons I love StarCoder, even for non-coding tasks. Trained only on Github means no "instruction finetuning" bullshit ChatGPT-speak.

oh great, so we can look forward to another horrifyingly mismanaged pandemic!

absolutely batshit crazy how nobody at all mentioned that not laid it at Trump's doorstep during the campaign

The Internet in general, but especially social media, is a cesspool of AI slop, bots, algorithm manipulation, and less exotic/high-tech forms of disinformation. How about instead of telling our politicians to meet us there we get the fuck off of it and meet each other in the streets. They can join our party if they want but we're starting it.