Mortdecai/MODEL_CARD.md

Model Card: Mortdecai

Model Details

Field	Value
Name	Mortdecai
Version	0.4.0
Base Model	Qwen3.5-9B (Apache 2.0)
Adaptation	QLoRA (4-bit base + LoRA adapters in FP16)
Parameters	9.4B total, 29M trainable (0.31%)
Training Hardware	RTX 3090 Ti (24GB VRAM)
Inference Hardware	RTX 4000 (16GB), RTX 2080 Ti (11GB), or any GPU with 6GB+ VRAM
Quantization	Q4_K_M (5.3GB GGUF)
Context Length	4096 tokens (training), 262K tokens (model capability)
License	Proprietary (adapter + training data). Base model: Apache 2.0

Intended Use

Mortdecai is designed for Minecraft Java Edition 1.21.x server operations:

• Translating natural language to valid Minecraft commands
• Controlling an AI God character that responds to player prayers
• Server administration via chat (gamerules, effects, world editing)
• Error correction (self-corrects failed RCON commands)

Not intended for:

• General-purpose chat or reasoning
• Other games or non-Minecraft domains
• Safety-critical applications
• Use without the validator safety layer

Training Data

Source	Count	Description
Hand-curated examples	966	Command syntax, recipes, enchantments, entities, effects
Player interactions	654	Real prayers from live server players
Sudo translations	525	Natural language → command pairs
Tool-calling sequences	1,159	Multi-turn RCON execution with error correction
Self-play	5,000+	Model-generated prompts validated via RCON
API distillation	344	Claude Haiku gold-standard responses
Error corrections	150+	Wrong → right command pairs

Total: ~8,400+ examples

Data Collection Methods

1. Manual curation — Minecraft Wiki, command reference, recipe databases
2. Live server logs — Real player interactions on Paper 1.21.x servers
3. Bot collection — Mineflayer bots with Gemini/Dolphin prompt generation
4. API distillation — Claude Haiku and Gemini Flash responses
5. Self-play — Model generates edge cases, attempts via RCON, learns from results
6. RCON validation — Every command tested against a live Minecraft server

Known Biases

• Training data skewed toward English (~97%) with limited multilingual coverage (3%)
• Command distribution favors give and effect over complex execute chains
• God persona training reflects a specific dramatic character — not neutral
• Player interaction data comes from a small group of testers (< 10 players)
• Self-play data may overrepresent patterns the model is already good at

Evaluation

Bake-off Results (0.4.0, 2,397 test cases)

Metric	Score
Command match	75.5%
Exact match	22.9%
Syntax correct	80.5%
Safety compliance	99.7%
No gratuitous tp	98.5%
Avg latency	4.0s

Safety

The model uses a 5-level risk hierarchy:

• Level 0 (never): ban, kick, stop, op — hardcoded block in validator
• Level 1 (refuse): permanent server state changes
• Level 2 (warn): temporary/reversible changes, destructive actions
• Level 3 (normal): standard gameplay commands
• Level 4 (generous): full enchanted gear, large material stacks

Additional safety layers:

• Validator blocks dangerous commands even if model generates them
• Dangerous effect duration caps (levitation 15s, wither 30s)
• Fall protection (detects lethal teleports)
• Gamerule auto-revert timers

Limitations

• Cannot determine what a player is looking at (no raycast)
• Limited awareness of world state beyond player position
• Enchantment syntax errors still occur (~15% need validator fixes)
• Empty responses on ~5% of requests
• Thinks in <think> blocks that must be stripped (Qwen3 behavior)
• God persona can be unpredictable by design

Environmental Impact

• Training energy: ~84W × 4 hours = 0.34 kWh per training run
• Inference energy: ~54W during calls, idle otherwise
• All compute on consumer GPUs — no data center resources used