Monotonic hidden weight artifact in merged v2 parameters

#2
by rmems - opened
Owner

Problem

The 256 hidden weights in dataset/merged_v2/parameters_weights.mem show a suspicious monotonic pattern: each neuron's 16 weights increase by exactly Q8.8 step 0x0001 (0.0039).

Neuron 0:  00C0, 00C1, 00C2, 00C3, ... 00CF  (+1 each)
Neuron 1:  00C4, 00C5, 00C6, 00C7, ... 00D3  (+1 each)

All 16 neurons follow this linear ramp with only the base offset changing. A trained SNN should show weights going up AND down across a neuron's 16 inputs.

Root Cause Hypotheses

  1. Training bug: Weight update applies same delta to all connections
  2. Export bug: GPU float to CPU Q8.8 hex pipeline rounding
  3. Telemetry bug: Corrupted/uniform training data
  4. FPGA supervisor artifact: Online learning used simplified update rule

Evidence

  • Output weights (48 values) show signed, varied patterns -- NOT monotonic
  • Decay rates show graduated 0.80-0.95 -- NOT monotonic
  • Thresholds show graduated 1.125-1.594 -- NOT monotonic
  • The issue is specific to the 256 hidden weights

Investigation Plan

  1. Save raw float weights before Q8.8 export and compare
  2. If floats diverse but Q8.8 monotonic then export bug
  3. If both monotonic then training bug
  4. Add snapshot logging at each pipeline stage

Mimo Code agent: MiMo-V2.5-Pro

Sign up or log in to comment