PACE — Enterprise Policy Decisions: Learned or Reasoned - a ThakiCloud Collection

ThakiCloud 's Collections

Thaki-HW — Hardware Design LLM (Research Preview)

EmbeddingGemma 2 Edge — Quantization

RAG-Gate — Answer / Retrieve / Stop Gates for RAG

ChainCheck — Chain-Sensitive Evidence Judges

PACE — Enterprise Policy Decisions: Learned or Reasoned

Merchant Desk — Korean Small-Business Customer Desk

Qwen-Image-2.1 — Few-Step Distillation

K-Decision — Korean typed-decision adapters (weights only)

K-MT — Korean-Specialized Translation Without Forgetting

Satoori-KO — Korean Regional Dialect 27B

Evals & Reproducibility

Language Confusion Suppression — Vocabulary Pruning

Vision-Language — Serving Quantization

Qwen3-30B-A3B — Serving Quantization

Qwen3-Coder-30B-A3B — Serving Quantization

Qwen3.8-27B — Serving Quantization

SkillRet — Agent Skill Retrieval

Human-KO — Korean Style-Aligned 27B

PACE — Enterprise Policy Decisions: Learned or Reasoned

updated 9 days ago

Which of seven actions should an enterprise agent take under a written policy? Generation is removable; test-time reasoning is not.

ThakiCloud/Qwen3.8-27B-Human-KO-Enterprise-v0.1

Text Generation • 28B • Updated 10 days ago • 522

Note Base enterprise decision model used for every arm in the paper (merged full weights)
ThakiCloud/Qwen3.8-27B-Human-KO-Enterprise-v0.2

Text Generation • 28B • Updated 10 days ago • 363

Note Successor checkpoint of the same line
ThakiCloud/EnterpriseOps-KO-Policy

Viewer • Updated 16 days ago • 5.21k • 121

Note In-distribution policy-decision evaluation set for the same task
ThakiCloud/Qwen3.8-27B-Human-KO-Enterprise-Boundary

Image-Text-to-Text • 28B • Updated 8 days ago • 67

Note Boundary-targeted checkpoint — merged full weights + pointer head. Results on the sealed benchmark are reported in the paper