---
pid: 3204107
cwd: "/home/ubuntu/hpml_nyu"
command: "cd /home/ubuntu/hpml_nyu && source venv/bin/activate && PYTHONUNBUFFERED=1 python lab2.py --cuda --task c4 2>&1"
started_at: 2026-03-05T09:50:48.204Z
running_for_seconds: 2221  
---
Device: cuda
ResNet-18 params: 11,173,962

============================================================
 C4: GPU vs CPU
============================================================

  [CPU] Training 5 epochs...
    Ep 1: loss=1.8457  acc=33.18%  time=453.39s
    Ep 2: loss=1.3745  acc=49.67%  time=445.87s
    Ep 3: loss=1.1178  acc=60.17%  time=435.31s
    Ep 4: loss=0.9305  acc=67.32%  time=430.32s
    Ep 5: loss=0.7908  acc=72.25%  time=421.75s
  [CPU] Avg epoch time: 437.33s

  [CUDA] Training 5 epochs...
    Ep 1: loss=1.9944  acc=29.43%  time=7.02s
    Ep 2: loss=1.4759  acc=45.76%  time=6.57s
    Ep 3: loss=1.1884  acc=57.23%  time=6.52s
    Ep 4: loss=0.9725  acc=65.46%  time=6.52s
    Ep 5: loss=0.8229  acc=71.01%  time=6.57s
  [CUDA] Avg epoch time: 6.64s

============================================================
 C4 Summary: GPU vs CPU (workers=4)
============================================================
  CPU avg: 437.33s/epoch
  GPU avg: 6.64s/epoch
  GPU speedup: 65.9x

---
exit_code: 0
elapsed_ms: 2225603
ended_at: 2026-03-05T10:27:53.807Z
---