Datasetpublic

zkaedi/zkaedi-prime-constitutions

ZKAEDI PRIME Constitutions Generated by Qwen2.5-72B-Instruct-AWQ via ZKAEDI PRIME recursive Hamiltonian field dynamics. Schema Column Type Description run_id str Run folder name e.g. run_001_20260312_184001 timestamp str UTC timestamp tier int 1=canonical (η≈0.4,β≈0.1,σ≈0.05) · 2=structured eta/gamma/beta/sigma/seed float PRIME field parameters v1_tokens/v2_tokens/critique_tokens int Generation token counts field_trace str Full 100-step… See the full description on the dataset page: https://huggingface.co/datasets/zkaedi/zkaedi-prime-constitutions.

sourceHugging Faceupdated 6mo agoView on Hugging Face

0likes2.3kdownloads

Dataset Card

ZKAEDI PRIME Constitutions

Generated by Qwen2.5-72B-Instruct-AWQ via ZKAEDI PRIME recursive Hamiltonian field dynamics.

Schema

Column	Type	Description
`run_id`	str	Run folder name e.g. `run_001_20260312_184001`
`timestamp`	str	UTC timestamp
`tier`	int	1=canonical (η≈0.4,β≈0.1,σ≈0.05) · 2=structured
`eta/gamma/beta/sigma/seed`	float	PRIME field parameters
`v1_tokens/v2_tokens/critique_tokens`	int	Generation token counts
`field_trace`	str	Full 100-step μ/σ/max/min trace
`constitution_v1`	str	Initial Qwen2.5-72B generation
`critique`	str	Self-critique of v1
`constitution_v2`	str	Refined generation post-critique
`phase_stats`	str	JSON of labelled phase statistics
`sd_min/sd_max/sd_range`	float	Field variability signals
`has_v2_improvement`	bool	v2 meaningfully longer than v1

Quality Tiers

Tier	Criteria	Count
1 — Canonical	η≈0.4, β<0.15, σ<0.08 (H100 A100 runs)	1
2 — Structured	β≥0.5 or non-tier3 params	133
~~3 — Dropped~~	β<0.4 AND σ>0.20 (decoupled noise)	filtered

Total: 134 records — train 127 / test 7

Pipeline

PRIME field evolution (32×32, depth=100)
  → field_trace (100-step μ/σ/max/min)
    → Qwen2.5-72B → constitution_v1
      → Qwen2.5-72B self-critique
        → Qwen2.5-72B → constitution_v2
          → telemetry.json

License

MIT