Mortdecai
9c2c9a2310
Distilled Training Data (1,203 examples): - 341 initial gold (plugins, enchantments, builds, effects, god, errors) - 165 buildings + pipeline (100 structures built on dev, 65 request→query→act) - 24 safety-aware (worldborder, safe tp, intentional harm, gamemode checks) - 17 advanced logic (decanonized items, redstone gates, iterative builds) - 12 redstone mastery (NOT/OR/AND/XOR/RS-latch/T-flip-flop/comparator/clock) - 7 circuit verification and diagnosis - 1 compact comparator gates - 10 redstone methodology (build→test→save→recall→learn from mistakes) - 8 player journal usage - 29 creative+uncommon+pipeline+god with full tool chains Player Journal System: - agent/tools/player_journal.py — per-player text files (1-10 lines) - journal.read + journal.write tool schemas added - Cross-contaminated: God and Sudo share same journal per player - Includes sentiment, relationship, builds, preferences, skill level Redstone Engineering: - agent/prompts/redstone_rules.md — baked-in wall torch, dedicated lead, repeater rules - Learned from 4 iterations of 8-switch circuit: wall_torch on back face, not top - T-junction bypass prevention: dedicated lead wire between merge and NOT block - RCON limitation: can build circuits but cannot test them (lever toggle doesn't propagate) Training Data Cleaning: - 466 @s→@p fixes, 10 template commands removed - 12 outdated refusals replaced with correct plugin commands - Data de-duped across all sources Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Mortdecai
A 9B parameter language model fine-tuned for Minecraft server operations. Translates natural language to commands, controls an AI God character, manages plugins, writes mcfunction scripts, and learns from its mistakes.
Base model: Qwen3.5-9B | Current version: 0.5.0 | Quantization: Q4_K_M (5.6GB)
Training Progress
| Version | Base Model | Training Examples | Loss | Key Addition |
|---|---|---|---|---|
| 0.1.0 | Qwen3-8B | 500 | 2.10 | Seed data only |
| 0.2.0 | Qwen3-8B | 1,200 | 1.45 | +entities, +mobs |
| 0.3.0 | Qwen3-8B | 2,100 | 0.82 | +error correction |
| 0.4.0 | Qwen3.5-9B | 3,175 | 0.35 | +tool-calling, base model upgrade |
| 0.5.0 | Qwen3.5-9B | 4,358 | 0.16 | +plugins, +memory, +scripts |
Bake-off: 0.5.0 vs 0.4.0
| Category | 0.4.0 | 0.5.0 | Change |
|---|---|---|---|
| Enchantments | 20% | 67% | +47% |
| EssentialsX | 0% | 60% | +60% |
| Effects | 0% | 25% | +25% |
| Basic commands | 75% | 75% | — |
| Teleport | 100% | 100% | — |
| Overall | 45.2% | 46.8% | +1.6% |
Architecture
17 tools across 5 categories:
| Category | Tools |
|---|---|
| Execution | rcon.execute |
| Knowledge | minecraft.wiki_lookup, plugin.docs_lookup, minecraft.changelog_lookup, paper.docs_lookup |
| World | world.player_info, world.server_state, world.nearby_entities |
| Memory | memory.read, memory.write |
| Scripts | script.write, script.validate, script.execute, script.read, script.list, script.delete, script.schedule |
Plugins: FastAsyncWorldEdit, WorldGuard, CoreProtect, EssentialsX, Vault, LuckPerms
Training Data
~20,000+ examples from:
- Hand-curated seed data (3,196)
- Tool-calling sequences with 17 tools (1,430)
- IGLU build dataset — Microsoft Research (4,656)
- RCON-validated plugin examples (104)
- Exploration self-play with wiki grounding (150)
- Self-play across 3 GPUs (2,900+)
- Live server audit from wolf bots + real players (8,000+)
Infrastructure
| GPU | Role |
|---|---|
| RTX 3090 Ti (24GB) | Training + self-play |
| RTX 2080 Ti (11GB) | Exploration self-play |
| Quadro RTX 4000 (8GB) | Production inference — 3 MC servers |
| GTX 1660 Super (6GB) | Prompt generation |
GPU Scheduler: gpu.sethpc.xyz — preset-based job scheduler with live monitoring
Links
- Play:
minecraft.mortdec.ai - Model card: MODEL_CARD.md
- Domain: mortdec.ai