Differences
This shows you the differences between two versions of the page.
| Both sides previous revisionPrevious revisionNext revision | Previous revision |
| linux64 [2026/11/02 21:29] – [Using pi.dev as ai harness] bengt | linux64 [2026/11/07 19:26] (current) – [Using pi.dev as ai harness] bengt |
|---|
| In this https://github.com/earendil-works/pi/issues/2020 thinkingFormat changes to qwen or maybe qwen-chat-template. | In this https://github.com/earendil-works/pi/issues/2020 thinkingFormat changes to qwen or maybe qwen-chat-template. |
| |
| I decided to install a cpu compactor "pi install npm:@sting8k/pi-vcc" (plain npm:pi-vcc is abandoned and buggy) to avoid running out of context window while compacting. Means that I can use the full server side context window client side. (I first tried npm:pi-compact but then it sometimes answered in Chinese and had a remember tool that I do not want.) Find compactors here: https://pi.dev/packages?name=compactor | A also did the following to make pi behave: |
| | - Install a cpu compactor "pi install npm:@sting8k/pi-vcc" (plain npm:pi-vcc is abandoned and buggy). This means that I can use the full server side context window client side. |
| | - Use "/settings" in pi and configure "HTTP idle timeout" to disabled. For local models with long output this can aovid "Error: Terminated". |
| | - You can also configure theme in "/settings" to match your terminal colors. |
| | - It is possible to control if all pre-entered steering prompts should be sent at once with "/settings " and "Steering mode" all. |
| |
| But this was suggested by the LLM to control output better: | Editing the system promt was also suggested to control output better: |
| <code> | <code> |
| $ cat ~/.pi/agent/APPEND_SYSTEM.md | $ cat ~/.pi/agent/APPEND_SYSTEM.md |