CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Sun 22 Nov, 12:28 UTC
%
0.9060453 of 500 eval, 47 wrong
▲ +6.40 pts across 3 evals▲ +3.80 pts on the eval before
0%25%50%75%100%0%25%50%75%100%0%25%50%75%100%

Familiar decider: ettin-encoder-68m

finishedRadeon 8060S

One of the small models that decide whether Familiar, a desktop companion, reacts to something on the PC, and how. Trained on the same 6,000 synthetic decision points as every candidate, two ways: the reaction itself, and the plain-case intent that rules adjust for her traits.

started
Sun 22 Nov, 12:18 UTC
ran for
07:38
config
batch 16, seed 1 of 2

run7/10step1125/1125epoch3/3config7/8seed1/2batch16modelettin-encoder-68m 68Mrevac19ae4items6000+500

Loss train 0.0167 ▼ −1.8851 lowest 0.0167 at step 1125 run before, run 6
Timebase
Range
Smoothing
Reference
113 readings, 2 min/div2 min/div1 min/div
012300:0002:0004:0006:00phase starting, 0.0 s to 15.7 sphase training, 15.7 s to 07:38epoch 1, 02:42epoch 2, 05:04epoch 3, 07:33trainingepoch 1epoch 2−1.8851since first1.9018Lowest last train loss across this project's 10 runs, run 9lowest last 0.0119, run 90.0167012300:0002:0004:0006:00phase starting, 0.0 s to 15.7 sphase training, 15.7 s to 07:38epoch 1, 02:42epoch 2, 05:04epoch 3, 07:33trainingepoch 1epoch 21.9018Lowest last train loss across this project's 10 runs, run 9lowest last 0.0119, run 90.0167▼ −1.8851012300:0001:0002:0003:0004:0005:0006:0007:00phase starting, 0.0 s to 15.7 sphase training, 15.7 s to 07:38epoch 1, 02:42epoch 2, 05:04epoch 3, 07:33trainingepoch 1epoch 21.9018Lowest last train loss across this project's 10 runs, run 9lowest last 0.0119, run 90.0167▼ −1.8851

Eval measures

Elapsed

145.7 s

epoch interval

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

0.39 s per step, 152.8 steps a minute

Eval accuracy by eval

+0.0380

▲ on the eval before

eval 1 0.8420, eval 3 0.9060

Against the other runs

2 of 10

by last eval accuracy

best so far 0.9120, run 8

▲ +0.0251 against run 6

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 1

Seed spread, observed

0.0060
0.900.95

2 finished seeds

Learning rate

0.00e+0

▼ −7.46e-6 since first

On one axis

phase starting, 0.0 s to 15.7 sphase training, 15.7 s to 07:38E1E2E3traininglowest last 0.0119Train loss0.0167▼ −1.88513.000.001.90180.0167last ▼ −1.8851best 0.9120, run 8Last-value range, 10 runs, 0.8584 to 0.9120Eval accuracy0.9060▲ +0.06400.930.830.84200.9060last ▲ +0.0640Learning rate0.00e+0▼ −7.46e-66.0e-50.0e+07.46e-6last 0.00e+0Power, whole chip, CPU and GPU116 W▼ −2 W120 W114 W118 Wlast 116 WHottest, one board zone91 °C▼ −4 °C95 °C80 °C95 °Clast 91 °C00:0001:0002:0003:0004:0005:0006:0007:00
GPUpower: whole chip, CPU and GPUtemperature: one board zone

GPU 0

Wwhole chip, CPU and GPU

116 W whole chip, CPU and GPU

use85%
memory, whole GPU0.8 GB

peak 9.1 GB, whole GPU

91 °C one board zone

history 0 to 120 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
0.950.850.950.85finished, eval accuracy 0.8780, config 1 of 8 (learning rate 0.0001, batch 16), seed 0, Sun 22 Nov, 09:36 UTCfinished, eval accuracy 0.8800, config 2 of 8 (learning rate 0.00002, batch 16), seed 0, Sun 22 Nov, 09:46 UTCfinished, eval accuracy 0.8860, config 3 of 8 (learning rate 0.00005, batch 16), seed 0, Sun 22 Nov, 09:56 UTCfinished, eval accuracy 0.8831, config 4 of 8 (learning rate 0.0001, batch 16), seed 0, Sun 22 Nov, 10:05 UTCfinished, eval accuracy 0.9034, config 5 of 8 (learning rate 0.00002, batch 16), seed 0, Sun 22 Nov, 10:14 UTCfinished, eval accuracy 0.8809, config 6 of 8 (learning rate 0.00005, batch 16), seed 0, Sun 22 Nov, 10:24 UTCfinished, eval accuracy 0.9060, config 7 of 8 (batch 16), seed 1 of 2, Sun 22 Nov, 12:18 UTCfinished, eval accuracy 0.9120, config 7 of 8 (batch 16), seed 2 of 2, Sun 22 Nov, 12:28 UTCfinished, eval accuracy 0.8831, config 8 of 8 (batch 16), seed 1 of 2, Sun 22 Nov, 12:38 UTCfinished, eval accuracy 0.8584, config 8 of 8 (batch 16), seed 2 of 2, Sun 22 Nov, 12:47 UTC0.950.85finished, eval accuracy 0.8780, config 1 of 8 (learning rate 0.0001, batch 16), seed 0, Sun 22 Nov, 09:36 UTCfinished, eval accuracy 0.8800, config 2 of 8 (learning rate 0.00002, batch 16), seed 0, Sun 22 Nov, 09:46 UTCfinished, eval accuracy 0.8860, config 3 of 8 (learning rate 0.00005, batch 16), seed 0, Sun 22 Nov, 09:56 UTCfinished, eval accuracy 0.8831, config 4 of 8 (learning rate 0.0001, batch 16), seed 0, Sun 22 Nov, 10:05 UTCfinished, eval accuracy 0.9034, config 5 of 8 (learning rate 0.00002, batch 16), seed 0, Sun 22 Nov, 10:14 UTCfinished, eval accuracy 0.8809, config 6 of 8 (learning rate 0.00005, batch 16), seed 0, Sun 22 Nov, 10:24 UTCfinished, eval accuracy 0.9060, config 7 of 8 (batch 16), seed 1 of 2, Sun 22 Nov, 12:18 UTCfinished, eval accuracy 0.9120, config 7 of 8 (batch 16), seed 2 of 2, Sun 22 Nov, 12:28 UTCfinished, eval accuracy 0.8831, config 8 of 8 (batch 16), seed 1 of 2, Sun 22 Nov, 12:38 UTCfinished, eval accuracy 0.8584, config 8 of 8 (batch 16), seed 2 of 2, Sun 22 Nov, 12:47 UTC

7 of 10 finished, 0.9060, config 7 of 8 (batch 16), seed 1 of 2, Sun 22 Nov, 12:18 UTC

Runs as a table
StartedStatusEval accuracyStepsConfig
Sun 22 Nov, 12:47 UTC| finished0.85841032 of 1032batch 16, seed 2 of 2
Sun 22 Nov, 12:38 UTC| finished0.88311032 of 1032batch 16, seed 1 of 2
Sun 22 Nov, 12:28 UTC| finished0.91201125 of 1125batch 16, seed 2 of 2
Sun 22 Nov, 12:18 UTC| finished0.90601125 of 1125batch 16, seed 1 of 2
Sun 22 Nov, 10:24 UTC| finished0.88091032 of 1032learning rate 0.00005, batch 16, seed 0
Sun 22 Nov, 10:14 UTC| finished0.90341032 of 1032learning rate 0.00002, batch 16, seed 0
Sun 22 Nov, 10:05 UTC| finished0.88311032 of 1032learning rate 0.0001, batch 16, seed 0
Sun 22 Nov, 09:56 UTC| finished0.88601125 of 1125learning rate 0.00005, batch 16, seed 0
Sun 22 Nov, 09:46 UTC| finished0.88001125 of 1125learning rate 0.00002, batch 16, seed 0
Sun 22 Nov, 09:36 UTC| finished0.87801125 of 1125learning rate 0.0001, batch 16, seed 0

Plate

Model
jhu-clsp/ettin-encoder-68m, 68M parameters revision ac19ae4
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 1
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
10.6 swent public
15.7 sphase training
160.2 seval 1
162.5 sepoch 1
304.4 sepoch 2
309.5 seval 2
453.9 sepoch 3
459.0 seval 3

Seeds

SeedStatusEval accuracy
seed 1| finished0.9060
seed 2| finished0.9120

observed spread 0.0060, 2 finished seeds

Readings

114 rows as a table

rows 1 to 100 of 114

StepSecondsTrain lossEval accuracyLearning rate
1015.71.9018-7.46e-6
2019.40.9805-1.49e-5
3023.20.9286-2.24e-5
4027.01.6443-2.99e-5
5030.91.7241-3.73e-5
6034.72.7386-4.48e-5
7038.60.9631-4.99e-5
8042.41.3308-4.94e-5
9046.21.4064-4.89e-5
10050.00.8865-4.84e-5
11053.82.4850-4.80e-5
12057.71.2724-4.75e-5
13061.51.1321-4.70e-5
14065.40.9063-4.66e-5
15069.21.2409-4.61e-5
16073.00.9675-4.56e-5
17076.90.9017-4.51e-5
18080.80.5907-4.47e-5
19084.61.3731-4.42e-5
20088.40.4460-4.37e-5
21092.20.5752-4.32e-5
22096.00.5137-4.28e-5
23099.90.2731-4.23e-5
240103.70.4045-4.18e-5
250107.40.6535-4.14e-5
260111.30.8193-4.09e-5
270115.10.5597-4.04e-5
280118.90.8371-3.99e-5
290122.70.4153-3.95e-5
300126.50.2880-3.90e-5
310130.31.0782-3.85e-5
320134.10.3925-3.80e-5
330137.90.9814-3.76e-5
340141.80.4430-3.71e-5
350145.70.3516-3.66e-5
360149.40.2749-3.62e-5
370153.30.3151-3.57e-5
375160.2-0.8420-
380162.50.5741-3.52e-5
390166.30.3498-3.47e-5
400170.20.4951-3.43e-5
410174.00.5666-3.38e-5
420177.80.8649-3.33e-5
430181.60.7076-3.28e-5
440185.50.2933-3.24e-5
450189.31.0221-3.19e-5
460193.10.6721-3.14e-5
470196.90.2584-3.10e-5
480200.70.7521-3.05e-5
490204.60.2206-3.00e-5
500208.50.1764-2.95e-5
510212.40.3979-2.91e-5
520216.20.5820-2.86e-5
530220.10.1890-2.81e-5
540223.90.1737-2.76e-5
550227.80.3781-2.72e-5
560231.60.8770-2.67e-5
570235.50.5802-2.62e-5
580239.31.0061-2.58e-5
590243.00.0842-2.53e-5
600246.90.1288-2.48e-5
610250.70.1762-2.43e-5
620254.50.5541-2.39e-5
630258.30.2325-2.34e-5
640262.20.1207-2.29e-5
650266.00.5233-2.24e-5
660269.80.0261-2.20e-5
670273.60.2661-2.15e-5
680277.50.6367-2.10e-5
690281.30.6175-2.06e-5
700285.20.5371-2.01e-5
710289.10.4756-1.96e-5
720293.00.4720-1.91e-5
730296.80.4444-1.87e-5
740300.60.4170-1.82e-5
750309.50.12690.86801.77e-5
760313.70.4227-1.72e-5
770317.50.1106-1.68e-5
780321.40.4977-1.63e-5
790325.20.4897-1.58e-5
800329.00.4406-1.54e-5
810332.80.2295-1.49e-5
820336.70.2931-1.44e-5
830340.50.8919-1.39e-5
840344.30.2931-1.35e-5
850348.10.7564-1.30e-5
860352.00.7329-1.25e-5
870355.80.0235-1.21e-5
880359.60.3202-1.16e-5
890363.50.1026-1.11e-5
900367.30.0472-1.06e-5
910371.10.2372-1.02e-5
920374.90.0426-9.69e-6
930378.80.4871-9.22e-6
940382.60.1223-8.74e-6
950386.50.1025-8.27e-6
960390.30.3577-7.80e-6
970394.10.4802-7.33e-6
980398.00.2886-6.85e-6
990402.00.0635-6.38e-6

GPU readings

114 rows as a table
StepGPUUse %Power W, whole chip, CPU and GPUMemory GB, whole GPUTemp C, one board zone
20072118.06.494.8
1100100116.27.784.8
190070119.49.185.8
280069119.39.186.8
3600100117.59.187.8
4300100116.49.187.8
5200100116.19.187.8
600070119.69.188.8
6900100117.79.188.8
760069119.79.190.8
8400100116.49.188.8
930072118.59.188.8
10100100117.19.188.8
1090076118.89.188.8
1125085115.70.890.8

Worked out

ReadoutValueWorked out from
Run complete100%1125 steps over 1125
Step time0.39 s432.1 s over 1100 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute152.860 s over the step time of 0.39 s
Epoch interval145.7 sMean of 2 gaps between epoch marks
Train loss since first reading−1.88510.0167 at step 1125 minus 1.9018 at step 10
Lowest train loss0.0167Lowest of 113 readings, at step 1125
Eval accuracy, as shown90.6%0.9060 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval+0.06400.9060 minus 0.8420
Eval change+0.0380Eval 3 at 0.9060 minus eval 2 at 0.8680
Wrong answers47500 eval items minus 453 correct. 453 over 500 is the only count that gives 0.9060
Rank2 of 10By last eval accuracy, across this project's 10 runs with an eval
Best so far0.9120Highest last eval accuracy, run 8
Against the run before+0.02510.9060 minus 0.8809 on run 6
Lowest last train loss0.0119Lowest last train loss across this project's 10 runs, run 9
Seed spread, observed0.0060High 0.9120 minus low 0.9060 across 2 finished seeds
Items per second40.71 step over 0.39 s, times batch size 16
Learning rate since first reading−7.46e-60.00e+0 minus 7.46e-6
Peak memory per GPU, whole GPU9.1 GBHighest sensor reading of any GPU, GPU 0, counting every process on it