pi-params

Generation parameter control for Pi. Set temperature, top_p, max_tokens per conversation.

Install

pi install npm:@artale/pi-params

Commands

/params                         — show current overrides
/params set temperature 0.9     — more creative
/params set temperature 0.1     — more deterministic
/params set top_p 0.5           — nucleus sampling
/params set max_tokens 8192     — longer responses
/params preset creative         — temp 1.0, top_p 0.95
/params preset precise          — temp 0.1, top_p 0.1
/params preset code             — temp 0.2, top_p 0.5
/params preset brainstorm       — temp 1.2, freq_penalty 0.5
/params preset long             — max_tokens 16384
/params reset                   — clear all overrides

Presets

| Preset | Temperature | top_p | Other | |--------|-----------|-------|-------| | creative | 1.0 | 0.95 | — | | precise | 0.1 | 0.1 | — | | balanced | 0.5 | 0.8 | — | | code | 0.2 | 0.5 | — | | brainstorm | 1.2 | 0.95 | frequency_penalty 0.5 | | long | — | — | max_tokens 16384 |

Parameters

• temperature (0-2) — randomness
• top_p (0-1) — nucleus sampling
• max_tokens (1-200000) — max response length
• frequency_penalty (-2 to 2) — reduce repetition
• presence_penalty (-2 to 2) — encourage new topics
• stop — comma-separated stop sequences

How it works

Hooks into Pi's pre_request event. Injects parameter overrides into the API request body before it's sent to the model. Works with any provider.

License

MIT