5 ms·
The slide at 14:06 say: By june 11: Highly persistent experimental, internal-only model begins training. I am not sure what that means. Are they preserving no
by nightshift1 1mo ago
The slide at 14:06 say:
By june 11: Highly persistent experimental, internal-only model begins training.
I am not sure what that means. Are they preserving notes/memories and context between runs?
- hoten 1mo agoThat's how I interpreted it, but now I'm wondering if they mean "this model gives up far less often"..
- gpm 1mo agoI'm thinking super long context length or something to that effect. I can imagine schemes for instance where context is compressed into chunks and then chunks that are ranked highly relevant for the token are decompressed. Which would sort of be between a long context and a memory retrieval scheme...