News for hackers. Stuff that matters.
Re:AI benchmarks are a bad joke – and LLM makers are the one... by Dylan16807 on Saturday November 08, 2025 @12:53PM (#45858520)
On some level this makes sense, but on the other hand LLMs already have perfect recall of thousands of symbols built into them, which is what pencil and paper gives to a human test taker.
Re:AI benchmarks are a bad joke – and LLM makers are the one... by zamadatix on Saturday November 08, 2025 @02:11PM (#45859110)
If only context recall was actually perfect! The data is certainly stored well, accurately accessing the right part... maybe worse than a human :D.
Re:AI benchmarks are a bad joke – and LLM makers are the one... by Dylan16807 on Saturday November 08, 2025 @02:15PM (#45859146)
If you're not doing clever hacks for very long windows, I thought a basic design fed in the entire window and it's up to the weights to use it properly.