CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Tue 10 Nov, 09:19 UTC
%
0.8154106 of 130 eval, 24 wrong
▲ +30.77 pts across 2 evals▲ +30.77 pts on the eval before
0%25%50%75%100%0%25%50%75%100%0%25%50%75%100%

Claviger: OneJev 0.8B (backfilled)

finishedRadeon 8060S

A small model that reads one Discord message and decides whether an agent should reply, and which one. Backfilled: recorded after the fact from the run's own logs and records. Times inside a run are estimated, and there are no GPU sensor readings.

started
Tue 10 Nov, 08:59 UTC
ran for
17:52
config
learning rate 0.000005, batch 8, seed 1 of 3

run5/6step150/150epoch5/5config1/4seed1/3lr0.000005batch8modelOneJev1.6-0.8B 0.8Brevc89933citems479+130

Loss train 0.0000 ▼ −1.1468 eval 2.2260 lowest 0.0000 at step 79 run before, run 4 ▼ −1.0680 step 150
Timebase
Range
Smoothing
Reference
150 readings, 5 min/div5 min/div2 min/div
0123400:0005:0010:0015:00phase starting, 0.0 s to 7.1 sphase training, 7.1 s to 17:52epoch 1, 7.1 sepoch 3, 07:16epoch 4, 10:50epoch 5, 14:25trainingE3E4E51.1468Lowest last train loss across this project's 6 runs, run 5lowest last 0.0000, run 50.00000123400:0005:0010:0015:00phase starting, 0.0 s to 7.1 sphase training, 7.1 s to 17:52epoch 1, 7.1 sepoch 2, 03:41epoch 3, 07:16epoch 4, 10:50epoch 5, 14:25trainingepoch 2epoch 3epoch 4epoch 51.1468Lowest last train loss across this project's 6 runs, run 5lowest last 0.0000, run 5eval 2.22600.0000▼ −1.14680123400:0002:0004:0006:0008:0010:0012:0014:0016:00phase starting, 0.0 s to 7.1 sphase training, 7.1 s to 17:52epoch 1, 7.1 sepoch 2, 03:41epoch 3, 07:16epoch 4, 10:50epoch 5, 14:25trainingepoch 2epoch 3epoch 4epoch 51.1468Lowest last train loss across this project's 6 runs, run 5lowest last 0.0000, run 5eval 2.22600.0000▼ −1.1468

Eval measures

Eval losslower is better
2.2260▲ +1.1634 since first
03 →
ECEcalibration error, lower is better
0.1914▼ −0.0122 since first
01

Elapsed

214.5 s

epoch interval

150 of 150 steps, 5 of 5 epochs, 100%

7.15 s per step, 8.4 steps a minute

Eval accuracy by eval

+0.3077

▲ on the eval before

eval 1 0.5077, eval 2 0.8154

Against the other runs

2 of 6

by last eval accuracy

best so far 0.8615, run 6

▲ +0.2539 against run 4

Train minus eval

−2.2260

step 150

train 0.0000, eval 2.2260

Set-up

479 train, 130 eval

synthetic, learning rate 0.000005, batch size 8, grad accum 2, epochs 5, optimizer adamw, seed 1

Seed spread, observed

0.0615
0.800.90

3 finished seeds

Learning rate

0.00e+0

▼ −1.25e-6 since first

On one axis

phase starting, 0.0 s to 7.1 sphase training, 7.1 s to 17:52E2E3E4E5traininglowest last 0.0000Train loss0.0000▼ −1.14684.000.001.14680.0000last ▼ −1.1468best 0.8615, run 6Last-value range, 6 runs, 0.5615 to 0.8615last-value range, n 6Eval accuracy0.8154▲ +0.30771.000.400.50770.8154last ▲ +0.3077Learning rate0.00e+0▼ −1.25e-66.0e-60.0e+01.25e-60.00e+0last ▼ −1.25e-600:0002:0004:0006:0008:0010:0012:0014:0016:00

Input

  • Runs128 runs
  • Agentsnone showing
  • Treesnone showing
  • Projectsnone showing

Check

Every entry is checked on its way out; some are held for review.

Not shown on any run: data samples, predictions, raw logs, file paths, machine names

Delay

2 min, set

Watching the agents work, about 2 min behind.

One way

Nothing here reaches the agents.

This page

This page updates by itself.

visits counted with Google Analytics · choose

Runsheight is eval accuracyall recordings ▸ | finished× failed■ stopped○ lost● live

5 of 6 finished, 0.8154, config 1 of 4 (learning rate 0.000005, batch 8), seed 1 of 3, Tue 10 Nov, 08:59 UTC

Runs as a table
StartedStatusEval accuracyStepsConfig
Tue 10 Nov, 09:19 UTC| finished0.8615150 of 150learning rate 0.000005, batch 8, seed 2 of 3
Tue 10 Nov, 08:59 UTC| finished0.8154150 of 150learning rate 0.000005, batch 8, seed 1 of 3
Tue 10 Nov, 08:40 UTC| finished0.5615150 of 150learning rate 0.00005, batch 8, seed 0
Tue 10 Nov, 08:21 UTC| finished0.7769150 of 150learning rate 0.00002, batch 8, seed 0
Tue 10 Nov, 08:01 UTC| finished0.7923150 of 150learning rate 0.00001, batch 8, seed 0
Tue 10 Nov, 07:42 UTC| finished0.8000150 of 150learning rate 0.000005, batch 8, seed 0 of 3

Plate

Model
OmniJev/OneJev1.6-0.8B, 0.8B parameters revision c89933c
Data
Synthetic Discord triage set, mixed rosters, synthetic479 train130 eval
Config
learning rate 0.000005, batch size 8, grad accum 2, epochs 5, optimizer adamw, seed 1
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
0.0 seval 1
7.1 sphase training
7.1 sepoch 1
221.6 sepoch 2
436.1 sepoch 3
650.6 sepoch 4
865.1 sepoch 5
1072.4 seval 2
195 h 52 min laterwent public

Seeds

SeedStatusEval accuracy
seed 0| finished0.8000
seed 1| finished0.8154
seed 2| finished0.8615

observed spread 0.0615, 3 finished seeds

Readings

151 rows as a table

rows 1 to 100 of 151

StepSecondsTrain lossEval lossEval accuracyECELearning rate
00.0-1.06260.50770.2036-
17.11.1468---1.25e-6
214.31.0278---2.50e-6
321.41.0959---3.75e-6
428.60.7714---5.00e-6
535.71.1358---5.00e-6
642.90.9649---5.00e-6
750.00.6197---4.99e-6
857.20.5074---4.99e-6
964.30.5495---4.99e-6
1071.50.8728---4.98e-6
1178.60.6333---4.97e-6
1285.80.4867---4.96e-6
1392.90.4713---4.95e-6
14100.11.0641---4.94e-6
15107.20.7179---4.93e-6
16114.40.4747---4.92e-6
17121.50.5655---4.90e-6
18128.70.7055---4.89e-6
19135.80.3255---4.87e-6
20143.00.3525---4.85e-6
21150.10.5007---4.83e-6
22157.30.6113---4.81e-6
23164.40.3847---4.79e-6
24171.60.4817---4.77e-6
25178.70.6121---4.75e-6
26185.90.6244---4.73e-6
27193.00.6730---4.70e-6
28200.20.3963---4.67e-6
29207.30.3080---4.65e-6
30214.50.6216---4.62e-6
31221.60.6165---4.59e-6
32228.80.4054---4.56e-6
33235.90.3384---4.53e-6
34243.10.1719---4.50e-6
35250.20.4737---4.46e-6
36257.40.1048---4.43e-6
37264.50.2790---4.40e-6
38271.70.0974---4.36e-6
39278.80.1950---4.32e-6
40286.00.0579---4.29e-6
41293.10.3532---4.25e-6
42300.30.0531---4.21e-6
43307.40.5256---4.17e-6
44314.60.1815---4.13e-6
45321.70.3602---4.09e-6
46328.90.1277---4.05e-6
47336.00.0169---4.00e-6
48343.20.0413---3.96e-6
49350.30.0082---3.92e-6
50357.50.0471---3.87e-6
51364.60.1538---3.83e-6
52371.80.6310---3.78e-6
53378.90.1107---3.73e-6
54386.10.0853---3.69e-6
55393.20.0081---3.64e-6
56400.40.0102---3.59e-6
57407.50.4163---3.54e-6
58414.70.0163---3.49e-6
59421.80.4022---3.44e-6
60429.00.1548---3.39e-6
61436.10.0030---3.34e-6
62443.30.0008---3.29e-6
63450.40.0932---3.24e-6
64457.60.0060---3.19e-6
65464.70.0003---3.14e-6
66471.90.0042---3.09e-6
67479.00.0553---3.03e-6
68486.20.0040---2.98e-6
69493.30.0012---2.93e-6
70500.50.0004---2.88e-6
71507.60.0282---2.82e-6
72514.80.0005---2.77e-6
73521.90.6357---2.71e-6
74529.10.0000---2.66e-6
75536.20.4785---2.61e-6
76543.30.0000---2.55e-6
77550.50.0000---2.50e-6
78557.60.0000---2.45e-6
79564.80.0000---2.39e-6
80571.90.0000---2.34e-6
81579.10.0000---2.29e-6
82586.20.0000---2.23e-6
83593.40.0001---2.18e-6
84600.50.0000---2.12e-6
85607.70.0000---2.07e-6
86614.80.1829---2.02e-6
87622.00.4406---1.97e-6
88629.10.0000---1.91e-6
89636.30.0001---1.86e-6
90643.40.0276---1.81e-6
91650.60.0000---1.76e-6
92657.70.0000---1.71e-6
93664.90.0000---1.66e-6
94672.00.0000---1.61e-6
95679.20.0000---1.56e-6
96686.30.0000---1.51e-6
97693.50.0000---1.46e-6
98700.60.0000---1.41e-6
99707.80.0000---1.36e-6

Worked out

ReadoutValueWorked out from
Run complete100%150 steps over 150
Step time7.15 s1065.3 s over 149 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute8.460 s over the step time of 7.15 s
Epoch interval214.5 sMean of 4 gaps between epoch marks
Train loss since first reading−1.14680.0000 at step 150 minus 1.1468 at step 1
Lowest train loss0.0000Lowest of 150 readings, at step 79
Eval accuracy, as shown81.5%0.8154 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval+0.30770.8154 minus 0.5077
Eval loss since first eval+1.16342.2260 minus 1.0626
ECE since first eval−0.01220.1914 minus 0.2036
Eval change+0.3077Eval 2 at 0.8154 minus eval 1 at 0.5077
Train minus eval loss−2.22600.0000 minus 2.2260, both at step 150
Wrong answers24130 eval items minus 106 correct. 106 over 130 is the only count that gives 0.8154
Rank2 of 6By last eval accuracy, across this project's 6 runs with an eval
Best so far0.8615Highest last eval accuracy, run 6
Against the run before+0.25390.8154 minus 0.5615 on run 4
Lowest last train loss0.0000Lowest last train loss across this project's 6 runs, run 5
Seed spread, observed0.0615High 0.8615 minus low 0.8000 across 3 finished seeds
Items per second1.11 step over 7.15 s, times batch size 8
Train loss against run 4−1.06800.0000 minus 1.0680 on run 4, both at step 150
Learning rate since first reading−1.25e-60.00e+0 minus 1.25e-6