-
5b28002001
0.6.0 training session: Oracle Bot, RL combat, Mind's Eye, multilingual pipeline
master
Seth
2026-03-22 20:22:50 -04:00
-
baab24f8b1
feat(oracle): HTML5 Canvas frontend — Mind's Eye viewport
Seth
2026-03-22 04:13:00 -04:00
-
61cdf70ebc
feat(oracle): express + websocket server with trace/command endpoints
Seth
2026-03-22 04:10:14 -04:00
-
7e9acad658
feat(oracle): world state abstraction layer
Seth
2026-03-22 04:08:45 -04:00
-
3510f0f571
feat(oracle): scaffold project + mineflayer spectator bot
Seth
2026-03-22 04:07:06 -04:00
-
924f16b9da
22-tool architecture: log.query, user.ask, journal system deployed
Mortdecai
2026-03-21 21:04:01 -04:00
-
9c2c9a2310
1200+ distilled gold examples, journal system, redstone mastery, safety awareness
Mortdecai
2026-03-21 20:50:52 -04:00
-
d9acb653fe
Fix chart labels, add version history table to README
Mortdecai
2026-03-21 15:48:35 -04:00
-
b6fbfac2ae
Add README with training progress chart and bake-off results
Mortdecai
2026-03-21 15:31:39 -04:00
-
f5118505b1
0.5.0 bake-off results, knowledge lookup tools, training progress chart
Mortdecai
2026-03-21 15:28:09 -04:00
-
da8f557219
GPU scheduler, 14-tool architecture, plugin deployment, event dispatcher
Mortdecai
2026-03-21 03:14:45 -04:00
-
434589d098
Prompt pipeline: 1660 generates, bigger GPUs process via Mortdecai
Mortdecai
2026-03-21 00:08:48 -04:00
-
3c1cbfce39
Shared player memory system + 39 training examples
Mortdecai
2026-03-20 23:37:32 -04:00
-
8158178a56
Shared player memory system + whitelist migration to CT 650
Mortdecai
2026-03-20 23:28:04 -04:00
-
84036d39ca
revert_after in model output + 20 training examples
Mortdecai
2026-03-20 23:25:20 -04:00
-
06b082bd21
0.5.0 pre-training: 9,444 examples, prod pattern fixes
Seth
2026-03-20 21:48:54 -04:00
-
bd65f4a84c
Add LICENSE, MODEL_CARD, requirements, CONTRIBUTING
Seth
2026-03-20 21:43:21 -04:00
-
f39809eaca
Semver rename: v1-v5 → 0.1.0-0.5.0 across all files
Seth
2026-03-20 21:37:14 -04:00
-
a03c0a8087
17 radius-aware kill examples: context determines blast radius
Seth
2026-03-20 21:27:20 -04:00
-
634f0137bb
10 entity targeting examples: THE zombie vs ALL zombies
Seth
2026-03-20 21:25:03 -04:00
-
5c71976a34
22 distance scale examples: 1 block to 30 million
Seth
2026-03-20 21:23:11 -04:00
-
b6e5874a11
45 new examples: chaos events, fireball/projectile mechanics, distance concepts
Seth
2026-03-20 21:20:30 -04:00
-
0f043384e5
Self-play: --api-key for authenticated gateway connections
Seth
2026-03-20 19:40:01 -04:00
-
aa5400e31e
12 multi-step dependency training examples
Seth
2026-03-20 18:43:03 -04:00
-
ead16fd429
Persistent RCON connections — fixes server crash from connection spam
Seth
2026-03-20 18:24:44 -04:00
-
67179f75ad
Self-play data + mortdecai-sites container + Grafana 3-GPU dashboard
Seth
2026-03-20 08:06:51 -04:00
-
25918b5b66
Self-play: 50 rounds, 0.1s sleep, max GPU utilization
Seth
2026-03-20 07:36:01 -04:00
-
dcc40a0bf8
Mortdecai v4 bake-off: 75.5% cmd match, 99.7% safety, 4.0s avg
Seth
2026-03-20 05:55:14 -04:00
-
027b835286
Session final: bakeoff fix, branding fonts, 3-GPU parallel self-play
Seth
2026-03-20 00:56:45 -04:00
-
3580d350b4
Parallel 3-GPU self-play: all tiers run simultaneously
Seth
2026-03-20 00:55:24 -04:00
-
de14f4a1c8
3-GPU overnight self-play: 3090 Ti + 2080 Ti + RTX 4000
Seth
2026-03-20 00:54:29 -04:00
-
9ef5ab5aa4
PLAN.md complete update — v4 deployed, all session work documented
Seth
2026-03-20 00:49:57 -04:00
-
7ae9a499fa
26 death/environment training examples, Mortdecai v4 deployed
Seth
2026-03-20 00:26:50 -04:00
-
d7138b3514
33 fall safety + suffocation training examples, fall damage test data
Seth
2026-03-20 00:07:36 -04:00
-
98d035439d
PLAN.md complete rewrite — Mortdecai project status, TODOs, risk hierarchy
Seth
2026-03-19 23:45:03 -04:00
-
4fc94170e4
Gamerule revert timers, drop/height training, revert_after field for v5
Seth
2026-03-19 23:42:22 -04:00
-
edfc365c5f
Dangerous effect caps: levitation 15s, wither 30s, poison 60s, nausea 30s
Seth
2026-03-19 23:35:57 -04:00
-
b85b1a6725
40 risk hierarchy examples: L0 blocked, L1 permanent, L2 temporary, injections
Seth
2026-03-19 23:30:46 -04:00
-
fbf6974af3
49 gamerule + invincibility training examples
Seth
2026-03-19 23:27:26 -04:00
-
7a31e500e4
Qwen3.5-9B on prod, Gemini 2.5 Flash for dev, error correction, branding
Seth
2026-03-19 23:09:27 -04:00
-
b75a737c11
7 enchantment syntax error examples: count order, typos, old NBT
Seth
2026-03-19 22:20:33 -04:00
-
a3d139e04f
Mortdecai v4 pre-training: /no_think, dedup, 3,369 examples
Seth
2026-03-19 20:15:00 -04:00
-
910d7b4ca7
Qwen3.5-9B bake-off results, model named Mortdecai
Seth
2026-03-19 19:46:00 -04:00
-
9abf9238c5
3-tier self-play: command drills, self-critique, adversarial
Seth
2026-03-19 19:39:33 -04:00
-
c947fc3fa9
Self-play loop, Qwen3.5-9B bake-off: 70% base accuracy
Seth
2026-03-19 19:35:57 -04:00
-
d31cdb21fd
1,833 training examples: entities, execute chains, multiplayer, advanced, redstone, biomes, errors
Seth
2026-03-19 19:22:32 -04:00
-
750cf15c79
1,542 seed + 1,159 tool-calling examples, async processing, validator tracking
Seth
2026-03-19 19:03:30 -04:00
-
ee764cd22a
Tool-calling training: 1,159 multi-turn examples with error correction
Seth
2026-03-19 18:49:08 -04:00
-
4e83da39fd
Quantity boundaries: item tier caps, tone-based scaling, 32 training examples
Seth
2026-03-19 18:22:26 -04:00
-
e780aef8c6
v3 model trained (1,308 examples, loss 0.55), API cascade, context update
Seth
2026-03-19 04:52:04 -04:00
-
234f2722db
v3 training dataset: 1,308 examples with risk_level + distilled data
Seth
2026-03-18 22:51:17 -04:00
-
e28836106f
Risk_level in all 644 examples + model outputs risk classification
Seth
2026-03-18 22:35:50 -04:00
-
0083e80aca
Persistent Haiku cost tracking, Sethian whitelist web app
Seth
2026-03-18 22:29:19 -04:00
-
0473eb0b50
Minecraft knowledge corpus, recipe trees, GitHub scraper, 644 examples
Seth
2026-03-18 20:33:09 -04:00
-
65ee146043
Swarm bots, RCON validation, Haiku distillation complete
Seth
2026-03-18 19:18:19 -04:00
-
961f53ea7d
God Soul document, Claude distillation pipeline, soul-driven prompts
Seth
2026-03-18 18:28:21 -04:00
-
62419976e5
361 training examples, default to 1 epoch
Seth
2026-03-18 18:03:33 -04:00
-
17a2a95f56
Add multilingual prompts (3%) — 12 languages from Qwen3 supported set
Seth
2026-03-18 18:02:54 -04:00
-
13debc8a59
Add audit log ingestion pipeline with language/leak filtering
Seth
2026-03-18 17:58:52 -04:00
-
7b9e4a9517
Dolphin-Mistral offensive prompts (5%), survival bot, cost-triggered POS printer
Seth
2026-03-18 17:54:17 -04:00
-
029bd28a58
Gemini-powered prayer bots, POS cost printer, first LoRA training run
Seth
2026-03-18 17:36:08 -04:00
-
142e4fd3c4
Fix training script: bf16 for Ampere GPU, add system prompts to training data
Seth
2026-03-18 16:26:47 -04:00
-
78031d16c0
Risk gradient (0-5), updated system prompts, 233 examples
Seth
2026-03-18 16:14:54 -04:00
-
9d789d2524
Three-tier constraint model, mode-aware eval, boundary examples, playtest tooling
Seth
2026-03-18 15:57:01 -04:00
-
38b9a02e45
Phase 2: eval harness, 182 examples, live bake-off, playtest infrastructure
Seth
2026-03-18 13:38:12 -04:00
-
eaa9e0c26b
Update PLAN.md with bake-off decisions and hardware assignments
Seth
2026-03-18 10:41:47 -04:00
-
48b627d498
Add LoRA training scripts and fix bake-off token budget
Seth
2026-03-18 10:40:18 -04:00
-
6fbab8045c
Add bake-off results summary (7 models, 31 examples)
Seth
2026-03-18 09:03:40 -04:00
-
7da28c8800
Add model bake-off harness and base model research
Seth
2026-03-18 08:54:11 -04:00
-
e00d454b19
Add baseline assistant with tools, guardrails, and system prompts (Phase 1.4)
Seth
2026-03-18 02:12:20 -04:00
-
77efac0283
Add knowledge corpus: 14 command references, server context, and TF-IDF search index (Phase 1.3)
Seth
2026-03-18 02:01:12 -04:00
-
827850b8d7
Initial project scaffold: dataset schema, 31 seed training examples, Mineflayer bot framework, and 7-phase roadmap
Seth
2026-03-18 01:51:28 -04:00