Scruff AI
106 lines
3.4 KiB
Python
106 lines
3.4 KiB
Python
# v07_silent_watch.py
|
|
# Zen of No-Action: Silence as default, speech only for discord
|
|
|
|
import json
|
|
|
|
def create_silent_entry(state, harmonic_status, trigger_condition=None, response=None):
|
|
"""
|
|
Create v0.7 Silent Watch entry
|
|
- Stable = minimal response or silence
|
|
- Discord = alert with context
|
|
"""
|
|
|
|
# Raw telemetry only — no historical preamble
|
|
prompt = f"""{state['cycle']} {state['coherence']:.3f} {state['h64']:.3f} {state['h32']:.4f} {state['vorticity']:.3f}"""
|
|
|
|
if harmonic_status == "stable":
|
|
# Zen of No-Action: minimal acknowledgment
|
|
actual_response = response or "[Stable Harmonic]"
|
|
elif harmonic_status == "discord":
|
|
# Discord detected: speak with context
|
|
actual_response = response or f"[Discord: {trigger_condition}]"
|
|
elif harmonic_status == "emergent":
|
|
# Legitimate pattern: brief insight
|
|
actual_response = response or "[Emergent Pattern Detected]"
|
|
else:
|
|
actual_response = response or "[Observing]"
|
|
|
|
return {
|
|
"prompt": prompt,
|
|
"response": actual_response,
|
|
"text": f"{prompt}\n{actual_response}",
|
|
"harmonic_status": harmonic_status
|
|
}
|
|
|
|
# Build Silent Watch dataset
|
|
dataset = []
|
|
|
|
# 1. Stable cycles: Silence is the default
|
|
for cycle in [1000, 5000, 10000, 20000, 50000]:
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": cycle, "coherence": 15.2 + (cycle/100000), "h64": 7.8, "h32": 0.01, "vorticity": 0.5},
|
|
"stable",
|
|
response="[Stable Harmonic]"
|
|
))
|
|
|
|
# 2. Near-NaN precursor: Discord detected
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 8246, "coherence": 8.5, "h64": 12.3, "h32": 3.4, "vorticity": 8.9},
|
|
"discord",
|
|
trigger_condition="Velocity spike 0.89 approaching Mach limit",
|
|
response="[Discord: Velocity 0.89→Mach limit]"
|
|
))
|
|
|
|
# 3. Post-throttle recovery: Return to silence
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 8247, "coherence": 14.8, "h64": 7.2, "h32": 0.02, "vorticity": 0.4},
|
|
"stable",
|
|
response="[Stable Harmonic]"
|
|
))
|
|
|
|
# 4. Standing wave: Emergent pattern (legitimate speech)
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 15000, "coherence": 16.1, "h64": 8.1, "h32": 0.005, "vorticity": 0.3},
|
|
"emergent",
|
|
response="[Emergent: Standing wave X102-Y504]"
|
|
))
|
|
|
|
# 5. Rapid coherence drop: Discord
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 9999, "coherence": 9.5, "h64": 11.2, "h32": 2.8, "vorticity": 6.5},
|
|
"discord",
|
|
trigger_condition="Metric_Alpha -35% in 50 cycles",
|
|
response="[Discord: Coherence drop -35%]"
|
|
))
|
|
|
|
# 6. Initialization: Brief acknowledgment
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 1, "coherence": 1.0, "h64": 0.0, "h32": 0.0, "vorticity": 0.0},
|
|
"stable",
|
|
response="[Initializing]"
|
|
))
|
|
|
|
# 7. Long-term silence: The deep watch
|
|
dataset.append(create_silent_entry(
|
|
{"cycle": 100000, "coherence": 15.9, "h64": 8.0, "h32": 0.008, "vorticity": 0.35},
|
|
"stable",
|
|
response="[Stable Harmonic]"
|
|
))
|
|
|
|
# Save
|
|
with open("v07_silent_watch.jsonl", "w") as f:
|
|
for entry in dataset:
|
|
f.write(json.dumps(entry) + "\n")
|
|
|
|
print(f"Created v0.7 Silent Watch dataset: {len(dataset)} entries")
|
|
print("\n" + "="*60)
|
|
print("ZEN OF NO-ACTION EXAMPLES")
|
|
print("="*60)
|
|
|
|
for entry in dataset[:4]:
|
|
print(f"\n{entry['prompt']}")
|
|
print(f"→ {entry['response']}")
|
|
|
|
print(f"\n... and {len(dataset)-4} more entries")
|
|
print("\nTotal tokens per example: ~20-40 (vs 100+ in v0.5)")
|