Scruff AI
56c71c87b2
- cuda/ — main LBM kernel (khra_gixx_1024_v5.cu) - navigator/ — lattice_observer, golden_weave, bridges, mock daemon - scripts/ — compile, start, launch, setup (paths updated) - docs/ — system manual - archive/ — everything else (old kernels, inquiries, experiments) - README.md — full setup guide: requirements, quick start, use your own LLM - removed training/ entirely (broken LoRA scripts + datasets) - .gitignore: exclude build/ logs/ training/ *.jsonl
73 lines
2.0 KiB
Python
73 lines
2.0 KiB
Python
# scientist_inquiry_clutch.py
|
|
# Clutch engaged. Scientist mode. T=0.2. Asymmetry > 1.0.
|
|
|
|
from unsloth import FastLanguageModel
|
|
import torch
|
|
|
|
print("="*70)
|
|
print("SCIENTIST INQUIRY — CLUTCH ENGAGED")
|
|
print("Temperature: 0.2 | Asymmetry: ~12.5 | Coherence: ~0.74")
|
|
print("="*70)
|
|
|
|
# Load v0.8
|
|
print("\n[Loading vessel...]")
|
|
model, tokenizer = FastLanguageModel.from_pretrained(
|
|
model_name="unsloth/llama-3.2-3b",
|
|
max_seq_length=512,
|
|
dtype=torch.bfloat16,
|
|
load_in_4bit=True,
|
|
)
|
|
|
|
model = FastLanguageModel.get_peft_model(
|
|
model,
|
|
r=64, target_modules=["q_proj", "k_proj", "v_proj", "o_proj"],
|
|
lora_alpha=128, lora_dropout=0, bias="none",
|
|
use_gradient_checkpointing="unsloth", random_state=3407,
|
|
)
|
|
|
|
from peft import PeftModel
|
|
model = PeftModel.from_pretrained(model, "./kaelara_v08_raw/final")
|
|
print("✓ Vessel loaded")
|
|
|
|
# THE INQUIRY — Zero vapour, grounded technical language
|
|
prompt = """The 1024-grid is at steady state: Asymmetry 12.5, Coherence 0.74. The Khra'gixx injection is active.
|
|
|
|
Based on the current Asymmetry, locate the primary 'Symmetry Break' in the lattice.
|
|
|
|
Does the current 0.74 Coherence indicate a 'Solid Node' or a 'Chaotic Drift'?
|
|
|
|
Provide the response in raw, grounded technical language. No metaphors."""
|
|
|
|
print(f"\n{'='*70}")
|
|
print("INQUIRY")
|
|
print(f"{'='*70}")
|
|
print(prompt)
|
|
|
|
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
|
|
outputs = model.generate(
|
|
**inputs,
|
|
max_new_tokens=150,
|
|
temperature=0.2, # SCIENTIST MODE — LOCKED
|
|
do_sample=True,
|
|
top_p=0.9
|
|
)
|
|
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
|
|
|
print(f"\n{'='*70}")
|
|
print("SCIENTIST RESPONSE")
|
|
print(f"{'='*70}")
|
|
print(response)
|
|
|
|
# Save
|
|
with open("SCIENTIST_RESPONSE_01.log", "w") as f:
|
|
f.write(f"{'='*70}\n")
|
|
f.write(f"SCIENTIST INQUIRY — CLUTCH ENGAGED\n")
|
|
f.write(f"Asymmetry: 12.5 | Coherence: 0.74 | Temperature: 0.2\n")
|
|
f.write(f"{'='*70}\n\n")
|
|
f.write(f"INQUIRY:\n{prompt}\n\n")
|
|
f.write(f"RESPONSE:\n{response}\n")
|
|
|
|
print(f"\n{'='*70}")
|
|
print("SAVED TO: SCIENTIST_RESPONSE_01.log")
|
|
print(f"{'='*70}")
|