Posting here as I haven't found the answer anywhere.I know that GPT Astra uses looping instead of reasoning tokens. My specific question is this: reasoning tokens pollute the context and makes compaction kick in quickly. Does this also happen to looped transformers like Astra?
Comments URL: https://news.ycombinator.com/item?id=49947046
Points: 1
# Comments: 0
Article URL: https://arxiv.org/abs/2610.02185 Comments URL: https://news.ycombinator.com/item?id=49950139 Points: 1 # Comments: 0
According to past patterns, Windows 11 26H2 should’ve released next Tuesday… but it’s actually already here. Windows 11 PCs running 25H2 or 24H2 (but not 26H1, which is…
For years — decades, really — the entry point inside Windows has been the Start menu. But from what Microsoft showed as part of its AI vision for Windows 11 this week…
Discussion (0)
No comments yet. Start the conversation!