danielharkin21 commited on
Commit
907eb6e
·
verified ·
1 Parent(s): 1bbe8da

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 : 2026-09-15 04:12:06 UTC
5
- Epoch : 2580
6
- Train CE : 0.816797
7
  Best CE : 0.811918
8
  Cortex base : nan
9
- U-loss (Huber): 0.109565
10
- Total loss : 0.926362
11
  -------------------------------------------------------
12
  u target : excess (signed excess of p; ~0 = trust p)
13
- excess mean : +0.0038 (p calibrated on avg if ~0)
14
- u mean : -0.0424 (should track excess mean)
15
- u corr : 0.0010 (Pearson u vs realized excess)
16
- u R^2 : -0.0364 (>0 = beats constant)
17
- u spread : -0.0060 (excess top-q u minus bottom-q u; want >0)
18
- const Huber : 0.1151 (u useful if U-loss < this: YES)
19
  -------------------------------------------------------
20
- Speed : 362 samples/s | 1.5 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 : 2026-09-15 04:19:34 UTC
5
+ Epoch : 2585
6
+ Train CE : 0.819402
7
  Best CE : 0.811918
8
  Cortex base : nan
9
+ U-loss (Huber): 0.109999
10
+ Total loss : 0.929402
11
  -------------------------------------------------------
12
  u target : excess (signed excess of p; ~0 = trust p)
13
+ excess mean : -0.0029 (p calibrated on avg if ~0)
14
+ u mean : -0.0417 (should track excess mean)
15
+ u corr : 0.0172 (Pearson u vs realized excess)
16
+ u R^2 : -0.0277 (>0 = beats constant)
17
+ u spread : +0.0101 (excess top-q u minus bottom-q u; want >0)
18
+ const Huber : 0.1109 (u useful if U-loss < this: YES)
19
  -------------------------------------------------------
20
+ Speed : 358 samples/s | 1.5 min/epoch
21
  Batch/GPU : 32 | GPUs: 2
22
  Grad Accum : 8 | Eff BS: 512
23
  Samples/ep : 32,000