| | Humanity has designed an AGI architecture – it is just prohibitively expensive (twitter.com/victortaelin) |
|
1 point by tosh 67 days ago | past | 1 comment
|
|
| | VictorTaelin: SupGen is a coding AI runs on 1 core CPU can prove theorems (twitter.com/victortaelin) |
|
38 points by hyperbrainer 3 months ago | past | 5 comments
|
|
| | Since Agda->JS is unmaintained, I'm now using Sonnet-3.5 as the compiler (twitter.com/victortaelin) |
|
2 points by LightMachine 7 months ago | past
|
|
| | Discovering algorithms by enumerating terms in Haskell (twitter.com/victortaelin) |
|
143 points by agomez314 9 months ago | past | 48 comments
|
|
| | $100k bet: AGI by end of 2025 (twitter.com/victortaelin) |
|
2 points by kvpan 10 months ago | past | 1 comment
|
|
| | Bend: High-level language that runs on GPUs (twitter.com/victortaelin) |
|
7 points by drak0n1c 11 months ago | past | 1 comment
|
|
| | HVM2: Run Python-like programs in 10000's of GPU cores (twitter.com/victortaelin) |
|
1 point by davikr on April 19, 2024 | past
|
|
| | A::B (twitter.com/victortaelin) |
|
7 points by pierre on April 12, 2024 | past | 1 comment
|
|
| | "Impossible" GPT Challenge Beaten in Hours (twitter.com/victortaelin) |
|
5 points by bratao on April 7, 2024 | past | 1 comment
|
|
| | [flagged] LLMs can't simulate a Turing machine reliably (twitter.com/victortaelin) |
|
56 points by Michelangelo11 on April 7, 2024 | past | 66 comments
|
|
| | A simple puzzle GPTs will NEVER solve (twitter.com/victortaelin) |
|
3 points by tosh on April 6, 2024 | past | 2 comments
|
|
| | Claude-3 emulates Pokémon Red (twitter.com/victortaelin) |
|
3 points by LightMachine on March 31, 2024 | past
|
|
| | PSA: GPT-4 Is Obsolete (twitter.com/victortaelin) |
|
2 points by LightMachine on March 14, 2024 | past
|
|
| | We can solve 3-SAT "quickly" via superposed booleans on interaction nets (twitter.com/victortaelin) |
|
1 point by LightMachine on Jan 9, 2024 | past
|
|
| | HVM updates: simplifications, finally runs on GPUs, 80x speedup on RTX 4090 (twitter.com/victortaelin) |
|
91 points by LightMachine on Oct 7, 2023 | past | 29 comments
|
|
| | A Lock-Free Interaction Combinator Evaluator to Fix LoL' Client Performance (twitter.com/victortaelin) |
|
2 points by LightMachine on Sept 8, 2023 | past
|
|
| | wrong url :( (twitter.com/victortaelin) |
|
1 point by LightMachine on July 28, 2023 | past
|
|
| | Increase GPT4's context size by asking it to compress your prompts (twitter.com/victortaelin) |
|
40 points by stolenmerch on April 5, 2023 | past | 14 comments
|
|
| | Increase GPT-4's effective context size by abbreviation-based prompt compression (twitter.com/victortaelin) |
|
3 points by lnyan on April 3, 2023 | past
|
|
| | GPT-4 emergent behavior, release not endorsed by red team (twitter.com/victortaelin) |
|
2 points by lawrencechen on March 15, 2023 | past | 1 comment
|
|
| | Higher Order Company (twitter.com/victortaelin) |
|
1 point by LightMachine on Jan 4, 2023 | past
|
|
| | HVM, the parallel runtime: a summary of its history and next steps (twitter.com/victortaelin) |
|
6 points by LightMachine on Nov 4, 2022 | past
|
|
