Post B6Lorl8DclWUkzmcgi by niconiconi@mk.absturztau.be
(DIR) More posts by niconiconi@mk.absturztau.be
(DIR) Post #B6Lorl8DclWUkzmcgi by niconiconi@mk.absturztau.be
0 likes, 0 repeats
It's surprising that the LLM programming hype is greater than ever, but no one has held a Top LLM Core War Championship where the LLMs write code to kill each other in Core War, like what TCEC does for chess engines. Must be an amusing show. I don't know anything about Core War or LLM agents, otherwise I would do it to fill the gap.
(DIR) Post #B6LpbcZ7nxJ3VjEkpk by wakame@tech.lgbt
0 likes, 0 repeats
@niconiconi "We would need a Terabyte GPU memory, because otherwise the models wouldn't fit."
(DIR) Post #B6LpbckT7mNe4uXoW0 by niconiconi@mk.absturztau.be
0 likes, 0 repeats
@wakame@tech.lgbt Not unlike early computer chess championships of which you need a semiconductor fab or Bell Labs' budget to join.
(DIR) Post #B6LpyYOJUVaNDz7CPA by wakame@tech.lgbt
0 likes, 0 repeats
@niconiconi TBH, my one-liner would defeat all those models if "generating the next token" was counted as an instruction.