danielharkin21 commited on
Commit
5106469
·
verified ·
1 Parent(s): a04572a

Upload training_status.txt with huggingface_hub

Browse files
Files changed (1) hide show
  1. training_status.txt +12 -12
training_status.txt CHANGED
@@ -1,23 +1,23 @@
1
  =======================================================
2
  EDDIE TRAINING STATUS — 2x T4 DDP (SPEED EXP)
3
  =======================================================
4
- Updated : 2026-09-16 06:40:12 UTC
5
- Epoch : 3290
6
- Train CE : 0.818512
7
  Best CE : 0.809309
8
  Cortex base : nan
9
- U-loss (Huber): 0.109876
10
- Total loss : 0.928389
11
  -------------------------------------------------------
12
  u target : excess (signed excess of p; ~0 = trust p)
13
- excess mean : +0.0027 (p calibrated on avg if ~0)
14
- u mean : -0.0417 (should track excess mean)
15
- u corr : 0.0072 (Pearson u vs realized excess)
16
- u R^2 : -0.0331 (>0 = beats constant)
17
- u spread : +0.0010 (excess top-q u minus bottom-q u; want >0)
18
- const Huber : 0.1138 (u useful if U-loss < this: YES)
19
  -------------------------------------------------------
20
- Speed : 343 samples/s | 1.6 min/epoch
21
  Batch/GPU : 32 | GPUs: 2
22
  Grad Accum : 8 | Eff BS: 512
23
  Samples/ep : 32,000
 
1
  =======================================================
2
  EDDIE TRAINING STATUS — 2x T4 DDP (SPEED EXP)
3
  =======================================================
4
+ Updated : 2026-09-16 06:48:03 UTC
5
+ Epoch : 3295
6
+ Train CE : 0.818778
7
  Best CE : 0.809309
8
  Cortex base : nan
9
+ U-loss (Huber): 0.109683
10
+ Total loss : 0.928461
11
  -------------------------------------------------------
12
  u target : excess (signed excess of p; ~0 = trust p)
13
+ excess mean : +0.0007 (p calibrated on avg if ~0)
14
+ u mean : -0.0419 (should track excess mean)
15
+ u corr : 0.0126 (Pearson u vs realized excess)
16
+ u R^2 : -0.0334 (>0 = beats constant)
17
+ u spread : +0.0061 (excess top-q u minus bottom-q u; want >0)
18
+ const Huber : 0.1120 (u useful if U-loss < this: YES)
19
  -------------------------------------------------------
20
+ Speed : 341 samples/s | 1.6 min/epoch
21
  Batch/GPU : 32 | GPUs: 2
22
  Grad Accum : 8 | Eff BS: 512
23
  Samples/ep : 32,000