Bartlomiej Lewandowski

Karma: 4

[Question] SOLAR model paper questions

Bartlomiej Lewandowski29 Dec 2023 20:34 UTC

1 point

0 comments1 min readLW link

Bartlomiej Lewandowski 29 Mar 2023 13:28 UTC
2 points
1
on: Why no major LLMs with memory?
I think there has been a lot of research in the past in this space. The first thing that popped into my mind was https://huggingface.co/docs/transformers/model_doc/rag
Currently, there are some approaches using langchain to persist the history of a conversation into an embeddings database, and retrieve the relevant parts performing a similar query / task.

Bartlomiej Lewandowski 14 Mar 2023 22:48 UTC
3 points
1
on: GPT can write Quines now (GPT-4)
OpenAI has hired a lot of software engineers to code simple tasks, maybe these quines were a part of the fine tuning set?

Bartlomiej Lewandowski 9 Dec 2021 4:54 UTC
2 points
on: Deepmind’s Gopher—more powerful than GPT-3
How is retro different from https://ai.facebook.com/blog/retrieval-augmented-generation-streamlining-the-creation-of-intelligent-natural-language-processing-models/ ?

Bartlomiej Lewandowski 18 Jul 2021 11:19 UTC
1 point
on: A closer look at chess scalings (into the past)
Isn’t ELO a reference metric that changes with time? I would assume that 2800 ELO in the 90s is a different level to today’s 2800. Can we still make the same conclusions with that in mind?