* Items of interest 2025-05-04
(Credit to Daniel Faggella for the picture:
https://x.com/danfaggella/status/1882745287959294249)
1. Qwen3-8B can fit into an 8GB VRAM GPU. With thinking enabled you can have your own LLM instance with robust tool calling capabilities.
https://huggingface.co/Qwen/Qwen3-8B
Further context: you can get an R9 580 for like 50 quid.
2. "Huawei has embarked on an unprecedented effort to develop every part of the AI supply chain domestically from wafer fabrication equipment to model building. We have never seen one company attempt to do everything before."
https://archive.is/cF1Ar
3. Google/gemini-2.5-pro has beaten Pokemon Blue (albeit with heavy scaffolding) in the span of 37 days.
https://x.com/googleaidevs/status/1918458458279469183
4. Strategies to thrive (more like survive) as AIs get better:
https://www.youtube.com/watch?v=A_fOHpBqj50
(Credit to Daniel Faggella for the picture:
https://x.com/danfaggella/status/1882745287959294249)
1. Qwen3-8B can fit into an 8GB VRAM GPU. With thinking enabled you can have your own LLM instance with robust tool calling capabilities.
https://huggingface.co/Qwen/Qwen3-8B
Further context: you can get an R9 580 for like 50 quid.
2. "Huawei has embarked on an unprecedented effort to develop every part of the AI supply chain domestically from wafer fabrication equipment to model building. We have never seen one company attempt to do everything before."
https://archive.is/cF1Ar
3. Google/gemini-2.5-pro has beaten Pokemon Blue (albeit with heavy scaffolding) in the span of 37 days.
https://x.com/googleaidevs/status/1918458458279469183
4. Strategies to thrive (more like survive) as AIs get better:
https://www.youtube.com/watch?v=A_fOHpBqj50
🔥1
https://telegra.ph/Logical-structuring-of-text-caches-reasoning-05-04 can't find a better place to microblog anonymously
Telegraph
Logical structuring of text caches reasoning
More important than *observed* intelligence in LLMs is their sociopathy and the double standard on textual interfaces. Whereas impatient managers and "non-technical" people have consistently asked to complicate interfaces from a text-based context (think…
🌭1
Forwarded from /g/‘s Tech Memes (krust)
This media is not supported in your browser
VIEW IN TELEGRAM
🌭1
Last schizophrenics online post about boblore was 31st of October... Guess what also happened at that time...
🌭1
This is nuanced, believe me. Connect the dots. Follow the money. Once you see it, you'll shit bricks... more like brocks.
🌭1
Forwarded from Axis of Ordinary
Links for 2025-05-05
AI
1. On the generalization of language models from in-context learning and finetuning: a controlled study https://arxiv.org/abs/2505.00661
2. Novel AI model inspired by neural dynamics from the brain https://news.mit.edu/2025/novel-ai-model-inspired-neural-dynamics-from-brain-0502
3. Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers https://physics.allen-zhu.com/part-4-architecture-design/part-4-1
4. Nikolay Savinov predicts that the industry is going to achieve near perfect retrieval across 1-2M context length 'quite soon', and that soon afterwards a 10M token context window will become the norm. https://www.youtube.com/watch?v=NHMJ9mqKeMQ
5. What's going on with AI progress and trends? (As of 5/2025) https://www.lesswrong.com/posts/v7LtZx6Qk5e9s7zj3/what-s-going-on-with-ai-progress-and-trends-as-of-5-2025
6. Waymo robotaxis are safer than human drivers https://growsf.org/news/2025-05-02-waymo-safety/
7. Around 60% of students reported using AI themselves, while they estimated that nearly 90% of their peers use AI. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5232910
8. “I Recorded Everything I Said for Three Months. AI Has Replaced My Memory.” https://www.wsj.com/tech/personal-tech/ai-personal-assistant-wearable-tech-impressions-28156b57 [no paywall: https://archive.is/xac6J]
9. This Chart Might Keep You From Worrying About AI’s Energy Use https://spectrum.ieee.org/ai-energy
10. Will nuclear energy power the AI boom? https://thebaffler.com/latest/project-ludicrous-northwood [no paywall: https://archive.is/m8KCB]
11. Jensen: "First thing to understand: 50% of the world's AI researchers are Chinese." https://www.youtube.com/live/E2o9O0EVouA?si=RZdkLpin-k5C8kGZ&t=594
12. What if AI just keeps getting smarter? https://www.lesswrong.com/posts/MCaqKAfSn345MCz7o/ra-x-controlai-video-what-if-ai-just-keeps-getting-smarter
13. Where’s my ten minute AGI? – if AIs are actually able to perform most tasks on 1-hour task horizons, why don’t we see more real-world task automation? https://epochai.substack.com/p/wheres-my-ten-minute-agi
14. T2I-R1: Reinforcing Image Generation with Collaborative Semantic-level and Token-level CoT https://arxiv.org/abs/2505.00703
15. Beyond the Last Answer: Your Reasoning Trace Uncovers More than You Think https://arxiv.org/abs/2504.20708
Miscellaneous
1. Novel High Resolution 3D Printing Method for Metals and Ceramics https://www.youtube.com/watch?v=kLgPW2672s4
2. Non-linear Ethnic Niches: The emerging Western caste system https://substack.com/home/post/p-162313414
3. Mathematician solves algebra’s oldest problem using intriguing new number sequences: “This is a dramatic revision of a basic chapter in algebra.” https://www.unsw.edu.au/newsroom/news/2025/05/mathematician-solves-algebras-oldest-problem-using-intriguing-new-number-sequences
AI
1. On the generalization of language models from in-context learning and finetuning: a controlled study https://arxiv.org/abs/2505.00661
2. Novel AI model inspired by neural dynamics from the brain https://news.mit.edu/2025/novel-ai-model-inspired-neural-dynamics-from-brain-0502
3. Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers https://physics.allen-zhu.com/part-4-architecture-design/part-4-1
4. Nikolay Savinov predicts that the industry is going to achieve near perfect retrieval across 1-2M context length 'quite soon', and that soon afterwards a 10M token context window will become the norm. https://www.youtube.com/watch?v=NHMJ9mqKeMQ
5. What's going on with AI progress and trends? (As of 5/2025) https://www.lesswrong.com/posts/v7LtZx6Qk5e9s7zj3/what-s-going-on-with-ai-progress-and-trends-as-of-5-2025
6. Waymo robotaxis are safer than human drivers https://growsf.org/news/2025-05-02-waymo-safety/
7. Around 60% of students reported using AI themselves, while they estimated that nearly 90% of their peers use AI. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5232910
8. “I Recorded Everything I Said for Three Months. AI Has Replaced My Memory.” https://www.wsj.com/tech/personal-tech/ai-personal-assistant-wearable-tech-impressions-28156b57 [no paywall: https://archive.is/xac6J]
9. This Chart Might Keep You From Worrying About AI’s Energy Use https://spectrum.ieee.org/ai-energy
10. Will nuclear energy power the AI boom? https://thebaffler.com/latest/project-ludicrous-northwood [no paywall: https://archive.is/m8KCB]
11. Jensen: "First thing to understand: 50% of the world's AI researchers are Chinese." https://www.youtube.com/live/E2o9O0EVouA?si=RZdkLpin-k5C8kGZ&t=594
12. What if AI just keeps getting smarter? https://www.lesswrong.com/posts/MCaqKAfSn345MCz7o/ra-x-controlai-video-what-if-ai-just-keeps-getting-smarter
13. Where’s my ten minute AGI? – if AIs are actually able to perform most tasks on 1-hour task horizons, why don’t we see more real-world task automation? https://epochai.substack.com/p/wheres-my-ten-minute-agi
14. T2I-R1: Reinforcing Image Generation with Collaborative Semantic-level and Token-level CoT https://arxiv.org/abs/2505.00703
15. Beyond the Last Answer: Your Reasoning Trace Uncovers More than You Think https://arxiv.org/abs/2504.20708
Miscellaneous
1. Novel High Resolution 3D Printing Method for Metals and Ceramics https://www.youtube.com/watch?v=kLgPW2672s4
2. Non-linear Ethnic Niches: The emerging Western caste system https://substack.com/home/post/p-162313414
3. Mathematician solves algebra’s oldest problem using intriguing new number sequences: “This is a dramatic revision of a basic chapter in algebra.” https://www.unsw.edu.au/newsroom/news/2025/05/mathematician-solves-algebras-oldest-problem-using-intriguing-new-number-sequences
🤮1
Forwarded from The Cube (LΩⁿī§ W∆ī№)
This media is not supported in your browser
VIEW IN TELEGRAM
🌚1
The magical garden of the gnomes
Someone tag the daily poor
Somewhat agree but they won't process the payment to Shiloh, just to let everyone know of this speculation.
And that's the optimistic outlook.
And that's the optimistic outlook.
👍1🌭1
Forwarded from Jonathan Bowden Unofficial Archive
This media is not supported in your browser
VIEW IN TELEGRAM
"You've got to be more radical than that..."
🌭1