Views
No views yet
| Property | Value |
|---|---|
| Base model | meta-llama/Llama-3.2-1B-Instruct |
| Training variant | C — Multi-turn dialogue (200 multi-turn episodes with iterative self-improvement) |
| Fine-tuning method | QLoRA (r=32-64, alpha=64-128) |
| Random seed | 123 |
| Training data | SRD v2 curated dataset |
1{
2 "run_name": "srd_C_llama-1B_seed123",
3 "model_name": "meta-llama/Llama-3.2-1B-Instruct",
4 "model_key": "llama-1B",
5 "variant": "C",
6 "seed": 123,
7 "data_path": "data/variant_c_v2/train.jsonl",
8 "data_size": 200,
9 "compute_dtype": "torch.bfloat16",
10 "lora_r": 32,
11 "method": "qlora_4bit"
12}1@article{ouyang2026srd,
2 title={Self-Reference Depth: A Unified Framework for Intelligence Across Biological and Artificial Systems},
3 author={Ouyang, Shumiao},
4 year={2026}
5}