News for hackers. Stuff that matters.

A comment on Show HN: How I Topped the HuggingFace Open LLM Leaderboard on Two Gaming GPUs. Parent comment

Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by Balinares on Tuesday March 10, 2026 @12:09PM (#47325175)
The idea that there may be a cognitive lingua franca hiding in the layers is fascinating and gives me hope for a neat idea: pluggable knowledge banks.

MoE notwithstanding, a model trained on the whole Internet and a few hundred thousands stolen books carries way more knowledge than is actually needed for any given workflow. It would be great if we could ship slimmed down models into which we'd plug the knowledge banks useful for today's work, and only those.

It would also mean that you could keep a model's knowledge fresh without retraining the whole of it.

[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by gitpusher on Tuesday March 10, 2026 @01:14PM (#47326102)
> pluggable knowledge banks.

plugs in knowledge bank LLM: ... I know kung fu.

[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by dormento on Tuesday March 10, 2026 @01:32PM (#47326336)
This is interesting. Would this mean less space for hallucination as well (depending on the breadth of knowledge applied to a specific task)?
[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by pennomi on Tuesday March 10, 2026 @03:00PM (#47327457)
Agreed, I suspect that LLMs in the future will have separate (possibly standardized) decoding/encoding layers that plug into logic layers.
[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by ay on Tuesday March 10, 2026 @09:31PM (#47330862)
Isn’t that what LoRA does ?
[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by CuriouslyC on Tuesday March 10, 2026 @10:36PM (#47331202)
LoRAs are better at steering models to produce correct answers from their data set than imparting new knowledge.
[ Parent | on HN ]
Re:Show HN: How I Topped the HuggingFace Open LLM Leaderboar... by kgeist on Thursday March 19, 2026 @04:59AM (#47436627)
https://arxiv.org/abs/2603.01097

>Overall, our findings position LoRA as the complementary axis of memory alongside RAG and ICL, offering distinct advantages.

[ Parent | on HN ]