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…
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?
0 comments
No comments yet.
Related stories
- If you configure a program to run in Windows 2000 compatibility mode, then it is also vulnerable to Windows 2000 security issuesblogs.msdn.microsoft.comLobsters · 4 points · about 9 years ago
- Hacker News · 1 points · 5 days ago
- Hacker News · 2 points · 7 days ago
- Hacker News · 4 points · 11 days ago
- Hacker News · 4 points · 13 days ago
- DEV Community · 8 points · 23 days ago