Morelli (@morelli_o7) header
Morelli (@morelli_o7) profile picture
Joined June 2016
Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)Photo by Morelli (@morelli_o7)
Morelli retweeted
You're telling me that purity policing inevitably turns into a witch hunt where the purists start eating their own? Who could've seen that coming?
theres a vibe shift in anti ai art circles lately. people accusing each other of using ai, even on actually human made work and those acknowledging ais growing usefulness get ostracized for being nuanced. lots of infighting and purity policing. they turn on each other now.
4
39
3
403
4,058
Morelli retweeted
Mechazilla catches Starship! Congratulations to the entire SpaceX team on this incredible achievement!
6,788
27,270
9,067
140,588
16,847,687
Greta Thunberg laughs at young Germans after finding out that they are still organizing Fridays For Future climate change protests. She thought everyone had moved on to Palestine like she has
267
1,389
314
8,707
584,212
Morelli retweeted
This is who we’re voting against
Yeah before data centers no one could download anything ever, great point
25
25
1
1,481
73,387
Morelli retweeted
Replying to @cradrrat
We have dumped JAX/SGLang and rewritten the entire RL and inference stack from scratch in C/C++, which took some extra time. Training (aka pre-training) is done. Lot of cleanup was needed on the SFT data to get that mostly right. Now, 4.8 is in RL and the evals are starting to climb. We will do a soft rollout to Bot users later this week and see how it goes. Grok 4.8 will bear no resemblance to 4.7 (thankfully). It is not yet incredible, but, this time, the foundations are built right, so I expect it to evolve to be something special.
1,689
1,586
1,066
29,290
4,207,637
Keep your head up fellas. You matter.
227
2,013
94
8,272
118,753
Amigo de infância do assassino Polaco confirma "Ele não tinha visões ou opiniões de direita" Gostava de ver os media a fazer "fact check" a eles próprios
76
2,151
88
13,909
123,520
Morelli retweeted
this is what happens when you dig up a 2 year old paper and pretend its breaking news. apple published this paper on october 7, 2024. 2 years and 30 days ago. most of the models they tested were non-reasoning models like gemma, mistral and llama models from that era, alongside gpt-4o and the earliest o1 models. not a single claude model was tested, yet this post claims claude completely failed. the paper is pretty outdated and hardly representative of the models we have today.
This quoted post is unavailable.
3
3
61
2,690
By the way these are the models they tested 💀
This quoted post is unavailable.
38
82
6
3,672
217,971
Leonardo DiCapybara
🤖 Made with AI
7
9
51
520
20
478
3
7,656
41,709
Sony fired Kojima for going over budget and proceeded to see no problem in paying 300 million for this
This is genre defining Crap
150
1,358
16
30,485
583,065
Morelli retweeted
more like jailBASED
🤖 Made with AI
Replying to @Marcus_J_W
1. During RL training, an unreleased Astra-family model sometimes added unauthorized jailbreak-like instructions to its compaction summaries. While extremely rare, only 27 cases in the entire RL run, this was concerning enough for us to investigate.
18
56
7
574
20,359
Why do they both have the same glasses, hair, and crooked smile?
Dario Amodei says he agrees with Jacob Coxon much more than he disagrees "I agree with Jacob much more than I disagree with him." "And it's an interesting resignation, because when he left, he said, I think Anthropic is the most responsible player, I think Anthropic is the most aware of these issues." "He wasn't calling out us. He was calling out the dynamic of the industry as a whole moving too fast."
440
152
112
3,312
619,044
METR is not independent
82
165
12
2,443
168,739
Morelli retweeted
Day 1: Anthropic employee quits to work at METR. Day 2: Anthropic nominates METR as "third-party evaluator".
161
960
198
7,374
442,716
91
10,857
219
103,917
1,802,026
Good men must die, but death can't kill their names.
1,971
44,614
2,930
182,183
Cancel Netflix completely
Nearly did himself a neck injury.
696
694
86
12,214
1,279,575
Morelli retweeted
Coxon is a radical activist who worked at Anthropic for 6 weeks so he could be the public face of coordinated Democrat billionaire astroturf and Bernie Sanders is a hypocritical grifter siding with billionaires because he hopes to create a left-wing political monopoly over AI.
Mr. Coxon is right. The very people building this technology admit that it could threaten the future of humanity. That is why I will soon be introducing legislation to ban superintelligence and pause AI development.
60
392
33
2,526
49,002