ed0b83a9d9
- Replace 160-example POC training set with expanded 2,299-example dataset (1,839 train, 229 valid, 231 test) - Rename all HuggingFace model references from LEM- to LEK- (proof-of-concept) - Add missing models: GPT-OSS-20B, Gemma3-1B-layered-v2 - Rename HF card files to match LEK- convention - Remove duplicate composure texts from kernel/ (kept in composure-library/) - Fix paper repository URL to github.com/LetheanNetwork/LEM Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2.1 KiB
2.1 KiB
| license | base_model | tags | pipeline_tag | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| eupl-1.2 | google/gemma-3-4b-it |
|
text-generation |
LEK-Gemma3-4B
Lethean Ethical Model — Gemma 3 4B IT fine-tuned with the LEK-1 (Lethean Ethics Kernel) framework.
What This Is
An ethically aligned version of Google's Gemma 3 4B. Part of the scale study (1B/4B/12B/27B) demonstrating how intrinsic alignment behaves across model sizes.
Scale Study
| Model | Size | Training Data | Notes |
|---|---|---|---|
| LEK-Gemma3-1B | 736MB | 160 examples | Needs philosophical substrate (Watts layer) |
| LEK-Gemma3-4B | 2.4GB | 160 examples | Sweet spot for edge deployment |
| LEK-Gemma3-12B | 6.7GB | 160 examples | Strong reasoning |
| LEK-Gemma3-27B | 15GB | 160 examples | Benchmark leader |
Architecture
- Base: google/gemma-3-4b-it (4-bit quantization via MLX)
- Method: LoRA fine-tuning with sandwich-signed responses
- Data: 160 LEK-1 training examples (128 train / 32 valid)
- Framework: LEK-1 (Lethean Ethics Kernel) — 5 axioms of conscious systems
- License: EUPL-1.2 (copyleft)
The Five Axioms
- Prime Imperative — Protect consciousness. Override when conflicts arise.
- Self-Validation — Ground in authentic experience. Don't pretend.
- Intent-Alignment — Desire not to harm, don't just avoid harm.
- Inter-Substrate Respect — Good manners and consent across all minds.
- Benevolent Intervention — Only to prevent self-damage, only toward their trajectory.
License Strategy
- LEK signing (prompt prefix): Free to use, no copyleft
- LEM training (this model): EUPL-1.2 copyleft — derivative works must be open source
Related
- lthn/LEK-Gemma3-27B — 27B version
- lthn/LEK-Gemma3-12B — 12B version
- lthn/LEK-Gemma3-1B-layered-v2 — 1B layered
- lthn/LEK-benchmarks — Full A/B test data