FYRELINKZ — LOCAL AI MEMORY TROUBLESHOOTING LOG
Original blank worksheet. Not benchmark results.

Date / operating system:
GPU / VRAM / available memory:
System RAM / disk pressure:
Runtime and version:
Model repository / exact tag / revision:
Weight format and quantization:
Context OR image width, height and batch:
Concurrent requests or GPU jobs:
Failed node / execution stage:
Complete error (remove private details before sharing):

BASELINE
Cold load time:
Warm completion times (repeat):
PROCESSOR / CONTEXT from ollama ps, if applicable:
Output acceptance check:

ONE CHANGE PER RETRY
Setting changed / old value / new value:
Did the exact error recur?
Did output quality change?
Was the task accepted?
Next isolated check: