CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Thu 19 Nov, 04:02 UTC
%
0.9160458 of 500 eval, 42 wrong
▲ +8.20 pts across 3 evals▲ +1.40 pts on the eval before
0%25%50%75%100%0%25%50%75%100%0%25%50%75%100%

Familiar decider: ModernBERT-large

finishedRadeon 8060S

One of the small models that decide whether Familiar, a desktop companion, reacts to something happening on the PC, and how. Every candidate trains on the same 6,000 decision points from synthetic timelines. The encoders also get one run on the full pool.

started
Thu 19 Nov, 02:14 UTC
ran for
96:03
config
batch 16, seed 2 of 2

run5/12step1125/1125epoch3/3config4/9seed2/2batch16modelModernBERT-large 395Mrev45bb465items6000+500

Loss train 0.0736 ▼ −1.6180 lowest 0.0075 at step 670 run before, run 4 ▲ +0.0616 step 1125
Timebase
Range
Smoothing
Reference
113 readings, 30 min/div30 min/div10 min/div
012300:0030:0060:0090:00phase starting, 0.0 s to 80.9 sphase training, 80.9 s to 96:03epoch 1, 33:05epoch 2, 64:36epoch 3, 95:43trainingepoch 1epoch 2−1.6180since first1.6916Lowest last train loss across this project's 12 runs, run 11lowest last 0.0007, run 110.0736012300:0030:0060:0090:00phase starting, 0.0 s to 80.9 sphase training, 80.9 s to 96:03epoch 1, 33:05epoch 2, 64:36epoch 3, 95:43trainingepoch 1epoch 21.6916Lowest last train loss across this project's 12 runs, run 11lowest last 0.0007, run 110.0736▼ −1.6180012300:0010:0020:0030:0040:0050:0060:0070:0080:0090:00phase starting, 0.0 s to 80.9 sphase training, 80.9 s to 96:03epoch 1, 33:05epoch 2, 64:36epoch 3, 95:43trainingepoch 1epoch 21.6916Lowest last train loss across this project's 12 runs, run 11lowest last 0.0007, run 110.0736▼ −1.6180

Eval measures

Elapsed

1878.9 s

epoch interval

1963.8 sto eval 11912.8 sto epoch 21866.7 sto epoch 3

1963.8 s to eval 1, 1912.8 s to epoch 2, 1866.7 s to epoch 3

1125 of 1125 steps, 3 of 3 epochs, 100%

5.07 s per step, 11.8 steps a minute

Eval accuracy by eval

+0.0140

▲ on the eval before

eval 1 0.8340, eval 3 0.9160

Against the other runs

4 of 11

by last eval accuracy

best so far 0.9600, run 7

▲ +0.0080 against run 4

Set-up

6000 train, 500 eval

synthetic, batch size 16, epochs 3, max length 512, precision bf16-mixed, loss ce, optimizer adamw, scheduler linear, seed 2

Seed spread, observed

0.0080
0.900.95

2 finished seeds

Learning rate

0.00e+0

▼ −2.99e-6 since first

On one axis

phase starting, 0.0 s to 80.9 sphase training, 80.9 s to 96:03E1E2E3traininglowest last 0.0007Train loss0.0736▼ −1.61803.000.001.69160.0736last ▼ −1.6180best 0.9600, run 7Last-value range, 11 runs, 0.8854 to 0.9600Eval accuracy0.9160▲ +0.08201.000.800.83400.9160last ▲ +0.0820Learning rate0.00e+0▼ −2.99e-62.0e-50.0e+02.99e-6last 0.00e+0Power, whole chip, CPU and GPU53 W▼ −1 W56 W50 W54 Wlast 53 WHottest, one board zone69 °C▲ +1 °C75 °C60 °C68 °Clast 69 °C00:0010:0020:0030:0040:0050:0060:0070:0080:0090:00
GPUpower: whole chip, CPU and GPUtemperature: one board zone

GPU 0

Wwhole chip, CPU and GPU

53 W whole chip, CPU and GPU

use42%
memory, whole GPU2.6 GB

peak 26.4 GB, whole GPU

69 °C one board zone

history 0 to 55 W

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
1.000.851.000.85finished, eval accuracy 0.9000, config 1 of 9 (learning rate 0.0001, batch 16), seed 0, Wed 18 Nov, 12:57 UTCfinished, eval accuracy 0.9200, config 2 of 9 (learning rate 0.00002, batch 16), seed 0, Wed 18 Nov, 14:41 UTCfinished, eval accuracy 0.9020, config 3 of 9 (learning rate 0.00005, batch 16), seed 0, Wed 18 Nov, 16:41 UTCfinished, eval accuracy 0.9080, config 4 of 9 (batch 16), seed 1 of 2, Thu 19 Nov, 00:21 UTCfinished, eval accuracy 0.9160, config 4 of 9 (batch 16), seed 2 of 2, Thu 19 Nov, 02:14 UTCfailed at step 470 of 7274, eval accuracy -, config 5 of 9 (batch 16), seed 0, Fri 20 Nov, 13:18 UTCfinished, eval accuracy 0.9600, config 5 of 9 (batch 16), seed 0, Sat 21 Nov, 01:57 UTCfinished, eval accuracy 0.8876, config 6 of 9 (learning rate 0.0001, batch 16), seed 0, Sat 21 Nov, 12:36 UTCfinished, eval accuracy 0.9191, config 7 of 9 (learning rate 0.00002, batch 16), seed 0, Sat 21 Nov, 13:15 UTCfinished, eval accuracy 0.8944, config 8 of 9 (learning rate 0.00005, batch 16), seed 0, Sat 21 Nov, 13:54 UTCfinished, eval accuracy 0.8854, config 9 of 9 (batch 16), seed 1 of 2, Sat 21 Nov, 19:01 UTCfinished, eval accuracy 0.8921, config 9 of 9 (batch 16), seed 2 of 2, Sat 21 Nov, 19:40 UTC1.000.85finished, eval accuracy 0.9000, config 1 of 9 (learning rate 0.0001, batch 16), seed 0, Wed 18 Nov, 12:57 UTCfinished, eval accuracy 0.9200, config 2 of 9 (learning rate 0.00002, batch 16), seed 0, Wed 18 Nov, 14:41 UTCfinished, eval accuracy 0.9020, config 3 of 9 (learning rate 0.00005, batch 16), seed 0, Wed 18 Nov, 16:41 UTCfinished, eval accuracy 0.9080, config 4 of 9 (batch 16), seed 1 of 2, Thu 19 Nov, 00:21 UTCfinished, eval accuracy 0.9160, config 4 of 9 (batch 16), seed 2 of 2, Thu 19 Nov, 02:14 UTCfailed at step 470 of 7274, eval accuracy -, config 5 of 9 (batch 16), seed 0, Fri 20 Nov, 13:18 UTCfinished, eval accuracy 0.9600, config 5 of 9 (batch 16), seed 0, Sat 21 Nov, 01:57 UTCfinished, eval accuracy 0.8876, config 6 of 9 (learning rate 0.0001, batch 16), seed 0, Sat 21 Nov, 12:36 UTCfinished, eval accuracy 0.9191, config 7 of 9 (learning rate 0.00002, batch 16), seed 0, Sat 21 Nov, 13:15 UTCfinished, eval accuracy 0.8944, config 8 of 9 (learning rate 0.00005, batch 16), seed 0, Sat 21 Nov, 13:54 UTCfinished, eval accuracy 0.8854, config 9 of 9 (batch 16), seed 1 of 2, Sat 21 Nov, 19:01 UTCfinished, eval accuracy 0.8921, config 9 of 9 (batch 16), seed 2 of 2, Sat 21 Nov, 19:40 UTC

5 of 12 finished, 0.9160, config 4 of 9 (batch 16), seed 2 of 2, Thu 19 Nov, 02:14 UTC

Runs as a table
StartedStatusEval accuracyStepsConfig
Sat 21 Nov, 19:40 UTC| finished0.89211032 of 1032batch 16, seed 2 of 2
Sat 21 Nov, 19:01 UTC| finished0.88541032 of 1032batch 16, seed 1 of 2
Sat 21 Nov, 13:54 UTC| finished0.89441032 of 1032learning rate 0.00005, batch 16, seed 0
Sat 21 Nov, 13:15 UTC| finished0.91911032 of 1032learning rate 0.00002, batch 16, seed 0
Sat 21 Nov, 12:36 UTC| finished0.88761032 of 1032learning rate 0.0001, batch 16, seed 0
Sat 21 Nov, 01:57 UTC| finished0.96007274 of 7274batch 16, seed 0
Fri 20 Nov, 13:18 UTC× failed-470 of 7274batch 16, seed 0
Thu 19 Nov, 02:14 UTC| finished0.91601125 of 1125batch 16, seed 2 of 2
Thu 19 Nov, 00:21 UTC| finished0.90801125 of 1125batch 16, seed 1 of 2
Wed 18 Nov, 16:41 UTC| finished0.90201125 of 1125learning rate 0.00005, batch 16, seed 0
Wed 18 Nov, 14:41 UTC| finished0.92001125 of 1125learning rate 0.00002, batch 16, seed 0
Wed 18 Nov, 12:57 UTC| finished0.90001125 of 1125learning rate 0.0001, batch 16, seed 0

Plate

Model
answerdotai/ModernBERT-large, 395M parameters revision 45bb465
Data
Familiar decisions, shared draw, synthetic6000 train500 eval
Config
batch size 16, epochs 3, max length 512, precision bf16-mixed, loss ce, optimizer adamw, scheduler linear, seed 2
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
2.2 swent public
80.9 sphase training
1963.8 seval 1
1985.6 sepoch 1
3876.6 sepoch 2
3897.1 seval 2
5743.4 sepoch 3
5763.9 seval 3

Seeds

SeedStatusEval accuracy
seed 1| finished0.9080
seed 2| finished0.9160

observed spread 0.0080, 2 finished seeds

Readings

114 rows as a table

rows 1 to 100 of 114

StepSecondsTrain lossEval accuracyLearning rate
1080.91.6916-2.99e-6
20115.01.5425-5.97e-6
30150.01.2463-8.96e-6
40202.80.8693-1.19e-5
50247.21.6809-1.49e-5
60300.91.7904-1.79e-5
70371.41.6976-1.99e-5
80442.11.4695-1.98e-5
90493.91.8254-1.96e-5
100546.41.6429-1.94e-5
110590.51.6155-1.92e-5
120650.32.3942-1.90e-5
130702.01.5254-1.88e-5
140771.91.0324-1.86e-5
150815.90.5876-1.84e-5
160876.10.7966-1.82e-5
170911.61.1216-1.81e-5
180955.10.6585-1.79e-5
1901007.60.3393-1.77e-5
2001059.10.9096-1.75e-5
2101104.50.9630-1.73e-5
2201157.11.0434-1.71e-5
2301208.40.6223-1.69e-5
2401253.30.7550-1.67e-5
2501314.50.3318-1.65e-5
2601365.90.3034-1.64e-5
2701418.60.1617-1.62e-5
2801463.40.9395-1.60e-5
2901524.80.4884-1.58e-5
3001586.60.7356-1.56e-5
3101629.90.2152-1.54e-5
3201656.30.2581-1.52e-5
3301725.30.5340-1.50e-5
3401779.30.9761-1.48e-5
3501822.40.9007-1.47e-5
3601865.80.4360-1.45e-5
3701917.00.3803-1.43e-5
3751963.8-0.8340-
3801985.60.1292-1.41e-5
3902020.30.2503-1.39e-5
4002055.30.3301-1.37e-5
4102124.50.1502-1.35e-5
4202150.40.9623-1.33e-5
4302209.11.3505-1.31e-5
4402271.30.4705-1.29e-5
4502330.50.3462-1.28e-5
4602391.30.1976-1.26e-5
4702417.30.2610-1.24e-5
4802486.20.5740-1.22e-5
4902529.00.5245-1.20e-5
5002581.10.8897-1.18e-5
5102631.80.1491-1.16e-5
5202684.70.3375-1.14e-5
5302727.70.5741-1.12e-5
5402772.70.5326-1.11e-5
5502807.20.2911-1.09e-5
5602858.50.1256-1.07e-5
5702918.80.6324-1.05e-5
5802961.50.1887-1.03e-5
5903021.00.6221-1.01e-5
6003081.60.2076-9.92e-6
6103135.30.6593-9.74e-6
6203178.40.1748-9.55e-6
6303238.10.3400-9.36e-6
6403298.40.9842-9.17e-6
6503367.60.0901-8.98e-6
6603410.40.1006-8.79e-6
6703471.30.0075-8.60e-6
6803522.10.5312-8.41e-6
6903565.10.7333-8.22e-6
7003633.90.1111-8.03e-6
7103685.30.2323-7.84e-6
7203728.10.1238-7.66e-6
7303755.30.0517-7.47e-6
7403815.20.2725-7.28e-6
7503897.10.30750.90207.09e-6
7603958.30.2279-6.90e-6
7704001.10.0886-6.71e-6
7804027.40.1918-6.52e-6
7904087.80.1262-6.33e-6
8004149.30.0603-6.14e-6
8104219.30.2732-5.95e-6
8204270.90.3840-5.77e-6
8304314.20.5417-5.58e-6
8404359.20.3075-5.39e-6
8504394.20.4665-5.20e-6
8604437.40.3216-5.01e-6
8704480.20.2419-4.82e-6
8804540.50.0158-4.63e-6
8904583.70.0419-4.44e-6
9004627.20.3964-4.25e-6
9104670.70.1132-4.06e-6
9204705.60.2945-3.88e-6
9304757.70.3069-3.69e-6
9404800.60.1325-3.50e-6
9504878.40.0967-3.31e-6
9604921.40.6398-3.12e-6
9704974.00.3114-2.93e-6
9805025.60.0081-2.74e-6
9905062.30.0666-2.55e-6

GPU readings

114 rows as a table

rows 1 to 100 of 112

StepGPUUse %Power W, whole chip, CPU and GPUMemory GB, whole GPUTemp C, one board zone
100053.719.067.8
200053.222.866.8
300051.922.868.8
400053.522.867.8
5002653.822.867.8
600054.122.867.8
7001252.622.869.8
800051.922.869.8
900053.222.866.8
1000053.522.870.8
1100053.426.469.8
1200053.926.467.8
1300054.926.468.8
1400053.426.467.8
15002152.926.468.8
1600053.126.468.8
1700052.926.469.8
1800251.626.468.8
19001951.926.469.8
2000053.126.467.8
2100052.526.469.8
2200054.026.467.8
23001152.126.467.8
2400052.926.468.8
2500052.726.467.8
2600053.026.468.8
2700053.426.472.8
2800053.626.471.8
2900053.826.472.8
3000053.626.468.8
3100053.726.464.8
3200051.626.469.8
3300051.726.467.8
3400054.326.468.8
3500952.226.470.8
3600053.326.469.8
3700051.926.469.8
3750054.126.465.8
38001653.426.468.8
3900053.226.465.8
4000053.126.468.8
4100152.626.468.8
4200153.026.470.8
4300052.926.468.8
4400052.626.467.8
4500252.726.469.8
4700052.826.472.8
4800054.226.470.8
4900051.826.468.8
5000053.526.467.8
5100252.726.470.8
5200053.626.466.8
5300051.626.467.8
5400055.026.472.8
5500054.026.468.8
5600053.326.468.8
5700252.726.468.8
5800053.026.468.8
5900052.726.468.8
6000053.026.469.8
6100053.226.471.8
6200252.526.470.8
6300052.826.467.8
6400053.326.467.8
6500053.126.469.8
6600352.626.468.8
6700053.326.467.8
68002452.626.470.8
6900051.126.469.8
7000053.326.468.8
7100055.326.469.8
7200053.426.466.8
7300053.326.467.8
7400053.726.468.8
7500053.126.467.8
7600052.726.467.8
7800051.726.469.8
7900054.426.468.8
8000053.326.468.8
8100952.526.469.8
8200054.326.469.8
8300052.326.467.8
8400053.426.470.8
8500053.326.469.8
8600053.026.467.8
8700054.026.468.8
8800053.926.469.8
8900053.126.467.8
90001751.326.468.8
9100052.126.468.8
9200054.126.471.8
9300053.026.466.8
9400054.526.469.8
9500053.426.466.8
9600052.426.467.8
9700452.426.468.8
9800053.326.468.8
9900053.026.468.8
10000053.226.466.8
10100052.126.467.8

Worked out

ReadoutValueWorked out from
Run complete100%1125 steps over 1125
Step time5.07 s5579.4 s over 1100 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute11.860 s over the step time of 5.07 s
Epoch interval1878.9 sMean of 2 gaps between epoch marks
Train loss since first reading−1.61800.0736 at step 1125 minus 1.6916 at step 10
Lowest train loss0.0075Lowest of 113 readings, at step 670
Eval accuracy, as shown91.6%0.9160 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval+0.08200.9160 minus 0.8340
Eval change+0.0140Eval 3 at 0.9160 minus eval 2 at 0.9020
Wrong answers42500 eval items minus 458 correct. 458 over 500 is the only count that gives 0.9160
Rank4 of 11By last eval accuracy, across this project's 11 runs with an eval
Best so far0.9600Highest last eval accuracy, run 7
Against the run before+0.00800.9160 minus 0.9080 on run 4
Lowest last train loss0.0007Lowest last train loss across this project's 12 runs, run 11
Seed spread, observed0.0080High 0.9160 minus low 0.9080 across 2 finished seeds
Items per second3.21 step over 5.07 s, times batch size 16
Train loss against run 4+0.06160.0736 minus 0.0120 on run 4, both at step 1125
Learning rate since first reading−2.99e-60.00e+0 minus 2.99e-6
Peak memory per GPU, whole GPU26.4 GBHighest sensor reading of any GPU, GPU 0, counting every process on it