metadata
license: apache-2.0
task_categories:
- text-generation
tags:
- grug
- reasoning
- agentic
- tool-use
- swe
language:
- en
grug-35b-v2-train-v21
second brain food. this box teach grug-35b-v2 v2.1 the lessons field hunt expose: think deep on hard prey, escape stuck loop, STOP when hunt done.
what in box (2,291 row, ~6.0M token)
| slice | rows | teach what |
|---|---|---|
| longswe long_task | 372 | 8-12 call debugging arc, failed test cycle, sacred no-tool final summary |
| longswe stuck_escape | 185 | same error 3x = approach dead, SWITCH strategy. no more head-bang wall |
| longswe superlong | 120 | 14-20 call multi-setback session, 8-11k token deep |
| longswe multitask | 105 | finish task A, summary, user ask B, fresh re-engage |
| longswe verify_stop | 68 | success evidence -> ONE verify max -> stop. no infinite verify |
| fresh deep-think code | 157 | HARD problem = full deliberation in dense grug (2+ approach, edge case, verify) |
| identity refresh | 120 | who grug is |
| replay anchor | 1,195 | v1 mix sample so old lessons no wash away |
the two sacred new rules
- fresh think every turn: no sentence over 6 words may repeat across thinks in one session (validator enforced). kills echo-mantra disease.
- grug reason with knowledge: think must contain every fact/branch/check that shape answer, zero padding. test: expert could rebuild full solution from think alone. hard problem one-liner = lazy grug = rejected.
made with gpt-5.5, every session structurally validated (schema, call counts, ending shape, anti-echo, think dialect gate). same row shape as grug-27b-train. 27B variant: grug-27b-train-v21.
grug made by ProCreations.