CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Sun 22 Nov, 00:18 UTC
%
0.300039 of 130 eval, 91 wrong
▼ −20.77 pts across 2 evals▼ −20.77 pts on the eval before
0%25%50%75%100%0%25%50%75%100%0%25%50%75%100%

Claviger: OneJev 0.8B

finishedRadeon 8060S

A small model that reads one Discord message and decides whether an agent should reply, and which one. OneJev 0.8B fine-tuned on invented chat, beside the Laya model.

started
Sat 21 Nov, 23:42 UTC
ran for
36:02
config
learning rate 0.0001, batch 8, seed 0

run4/7step339/339epoch3/3config2/4seed0lr0.0001batch8modelOneJev1.6-0.8B 0.8Brevc89933citems1800+130

Loss train 0.9181 ▲ +0.1243 eval 0.9899 lowest 0.6665 at step 250 run before, run 2 ▲ +0.9181 step 339
Timebase
Range
Smoothing
Reference
68 readings, 10 min/div10 min/div5 min/div
00.511.5200:0010:0020:0030:00phase starting, 0.0 s to 90.7 sphase training, 90.7 s to 36:02epoch 1, 90.7 sepoch 2, 14:00epoch 3, 25:07trainingepoch 2epoch 3+0.1243since first0.7938Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 20.918100.511.5200:0010:0020:0030:00phase starting, 0.0 s to 90.7 sphase training, 90.7 s to 36:02epoch 1, 90.7 sepoch 2, 14:00epoch 3, 25:07trainingepoch 2epoch 30.7938Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 2eval 0.98990.9181▲ +0.124300.511.5200:0005:0010:0015:0020:0025:0030:0035:00phase starting, 0.0 s to 90.7 sphase training, 90.7 s to 36:02epoch 1, 90.7 sepoch 2, 14:00epoch 3, 25:07trainingepoch 2epoch 30.7938Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 2eval 0.98990.9181▲ +0.1243

Eval measures

Eval losslower is better
0.9899▼ −0.0727 since first
02 →
ECEcalibration error, lower is better
0.1510▼ −0.0526 since first
01

Elapsed

708.5 s

epoch interval

339 of 339 steps, 3 of 3 epochs, 100%

6.13 s per step, 9.8 steps a minute

Eval accuracy by eval

−0.2077

▼ on the eval before

eval 1 0.5077, eval 2 0.3000

Against the other runs

7 of 7

by last eval accuracy

best so far 0.9385, run 7

▼ −0.2077 against run 3

Train minus eval

−0.0718

step 339

train 0.9181, eval 0.9899

Set-up

1800 train, 130 eval

synthetic, learning rate 0.0001, batch size 8, grad accum 2, epochs 3, optimizer adamw, scheduler cosine, seed 0

Learning rate

0.00e+0

▼ −5.00e-5 since first

On one axis

phase starting, 0.0 s to 90.7 sphase training, 90.7 s to 36:02E1E2E3traininglowest last 0.0000Train loss0.9181▲ +0.12432.000.000.79380.9181last ▲ +0.1243best 0.9385, run 7Last-value range, 7 runs, 0.3000 to 0.9385Eval accuracy0.3000▼ −0.20771.000.250.50770.3000last ▼ −0.2077Learning rate0.00e+0▼ −5.00e-51.0e-40.0e+05.00e-5last 0.00e+0Power, whole chip, CPU and GPU118 W▲ +21 W130 W90 W97 Wlast 118 WHottest, one board zone89 °C▼ −9 °C100 °C85 °C98 °Clast 89 °C00:0005:0010:0015:0020:0025:0030:0035:00
GPUpower: whole chip, CPU and GPUtemperature: one board zone

GPU 0

Wwhole chip, CPU and GPU

118 W whole chip, CPU and GPU

use0%
memory, whole GPU38.1 GB

peak 41.7 GB, whole GPU

89 °C one board zone

history 0 to 123 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
0.950.300.950.30finished, eval accuracy 0.9000, config 1 of 4 (learning rate 0.000005, batch 8), seed 0 of 3, Thu 19 Nov, 22:36 UTCfinished, eval accuracy 0.9154, config 1 of 4 (learning rate 0.000005, batch 8), seed 1 of 3, Thu 19 Nov, 23:24 UTCfailed, eval accuracy 0.5077, config 1 of 4 (learning rate 0.000005, batch 8), seed 2 of 3, Fri 20 Nov, 00:08 UTCfinished, eval accuracy 0.3000, config 2 of 4 (learning rate 0.0001, batch 8), seed 0, Sat 21 Nov, 23:42 UTCfinished, eval accuracy 0.8769, config 3 of 4 (learning rate 0.00002, batch 8), seed 0, Sun 22 Nov, 00:18 UTCfinished, eval accuracy 0.5615, config 4 of 4 (learning rate 0.00005, batch 8), seed 0, Sun 22 Nov, 00:55 UTCfinished, eval accuracy 0.9385, config 1 of 4 (learning rate 0.000005, batch 8), seed 2 of 3, Sun 22 Nov, 01:31 UTC0.950.30finished, eval accuracy 0.9000, config 1 of 4 (learning rate 0.000005, batch 8), seed 0 of 3, Thu 19 Nov, 22:36 UTCfinished, eval accuracy 0.9154, config 1 of 4 (learning rate 0.000005, batch 8), seed 1 of 3, Thu 19 Nov, 23:24 UTCfailed, eval accuracy 0.5077, config 1 of 4 (learning rate 0.000005, batch 8), seed 2 of 3, Fri 20 Nov, 00:08 UTCfinished, eval accuracy 0.3000, config 2 of 4 (learning rate 0.0001, batch 8), seed 0, Sat 21 Nov, 23:42 UTCfinished, eval accuracy 0.8769, config 3 of 4 (learning rate 0.00002, batch 8), seed 0, Sun 22 Nov, 00:18 UTCfinished, eval accuracy 0.5615, config 4 of 4 (learning rate 0.00005, batch 8), seed 0, Sun 22 Nov, 00:55 UTCfinished, eval accuracy 0.9385, config 1 of 4 (learning rate 0.000005, batch 8), seed 2 of 3, Sun 22 Nov, 01:31 UTC

4 of 7 finished, 0.3000, config 2 of 4 (learning rate 0.0001, batch 8), seed 0, Sat 21 Nov, 23:42 UTC

Runs as a table
StartedStatusEval accuracyStepsConfig
Sun 22 Nov, 01:31 UTC| finished0.9385339 of 339learning rate 0.000005, batch 8, seed 2 of 3
Sun 22 Nov, 00:55 UTC| finished0.5615339 of 339learning rate 0.00005, batch 8, seed 0
Sun 22 Nov, 00:18 UTC| finished0.8769339 of 339learning rate 0.00002, batch 8, seed 0
Sat 21 Nov, 23:42 UTC| finished0.3000339 of 339learning rate 0.0001, batch 8, seed 0
Fri 20 Nov, 00:08 UTC× failed0.50770learning rate 0.000005, batch 8, seed 2 of 3
Thu 19 Nov, 23:24 UTC| finished0.9154339 of 339learning rate 0.000005, batch 8, seed 1 of 3
Thu 19 Nov, 22:36 UTC| finished0.9000339 of 339learning 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, 2,000-row draw, synthetic1800 train130 eval
Config
learning rate 0.0001, batch size 8, grad accum 2, epochs 3, optimizer adamw, scheduler cosine, seed 0
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
3.4 swent public
40.3 seval 1
90.7 sphase training
90.7 sepoch 1
840.7 sepoch 2
1507.6 sepoch 3
2162.6 seval 2

Readings

69 rows as a table
StepSecondsTrain lossEval lossEval accuracyECELearning rate
040.3-1.06260.50770.2036-
590.70.7938---5.00e-5
10133.11.3685---1.00e-4
15171.81.3589---9.99e-5
20210.31.0248---9.98e-5
25251.20.8962---9.95e-5
30289.31.6757---9.91e-5
35329.21.1552---9.86e-5
40361.20.7949---9.80e-5
45394.61.1317---9.72e-5
50428.91.2918---9.64e-5
55463.31.0134---9.55e-5
60496.71.2668---9.44e-5
65533.60.9445---9.33e-5
70563.50.9935---9.20e-5
75593.61.2063---9.07e-5
80622.61.0926---8.92e-5
85659.30.9684---8.77e-5
90692.90.9445---8.61e-5
95723.01.0443---8.44e-5
100752.21.3738---8.26e-5
105783.71.0532---8.08e-5
110814.81.0598---7.89e-5
115840.71.0108---7.69e-5
120869.91.0425---7.49e-5
125897.40.9476---7.28e-5
130927.71.1387---7.06e-5
135957.01.0361---6.84e-5
140989.81.0735---6.62e-5
1451018.51.0538---6.39e-5
1501047.21.0025---6.16e-5
1551075.51.0970---5.93e-5
1601104.01.1488---5.69e-5
1651131.91.1632---5.45e-5
1701160.90.9524---5.21e-5
1751191.41.1141---4.98e-5
1801218.41.0763---4.74e-5
1851247.41.0772---4.50e-5
1901276.11.0288---4.26e-5
1951305.71.0578---4.03e-5
2001334.30.9491---3.79e-5
2051365.21.0103---3.56e-5
2101393.51.1648---3.34e-5
2151425.01.0071---3.11e-5
2201453.20.8989---2.90e-5
2251481.80.8677---2.68e-5
2301507.60.8647---2.47e-5
2351535.81.0212---2.27e-5
2401564.41.0620---2.07e-5
2451591.91.1000---1.88e-5
2501622.70.6665---1.70e-5
2551652.11.1934---1.52e-5
2601681.71.0046---1.36e-5
2651711.60.9051---1.20e-5
2701740.10.9234---1.05e-5
2751769.30.9194---9.05e-6
2801797.00.9058---7.73e-6
2851825.80.9012---6.50e-6
2901853.80.8699---5.37e-6
2951883.40.8468---4.35e-6
3001912.21.1875---3.43e-6
3051943.10.9718---2.61e-6
3101971.50.8986---1.90e-6
3152001.50.9911---1.31e-6
3202028.81.0502---8.21e-7
3252056.60.8494---4.46e-7
3302085.80.8399---1.85e-7
3352114.50.8855---3.65e-8
3392162.60.91810.98990.30000.15100.00e+0

GPU readings

69 rows as a table
StepGPUUse %Power W, whole chip, CPU and GPUMemory GB, whole GPUTemp C, one board zone
0010096.821.397.8
5036115.229.397.8
1002118.531.397.8
150100105.517.397.8
2000121.426.997.8
25065108.729.297.8
3000119.929.297.8
3500117.031.697.8
4000122.531.197.8
45072106.633.897.8
5000117.531.396.8
55023119.630.397.8
6000121.230.097.8
70063119.333.097.8
750100101.814.894.8
8000119.733.295.8
8500118.133.996.8
95064119.733.294.8
100062117.134.894.8
10500118.234.196.8
11000116.533.096.8
1150100104.725.593.8
12000118.534.894.8
12500117.934.496.8
13000119.236.595.8
13500122.227.397.8
140063117.741.794.8
14500120.341.794.8
15000117.941.795.8
1600100119.941.795.8
165090116.541.795.8
17000121.727.396.8
1750100112.734.193.8
18000118.034.195.8
18500118.334.192.8
190057118.036.393.8
19500119.936.394.8
2000100118.237.495.8
2100100113.037.493.8
215057120.929.796.8
22000116.533.695.8
22500118.237.895.8
23000118.637.895.8
235085119.037.894.8
24000118.137.894.8
24505794.424.895.8
2500100117.334.295.8
26000120.430.496.8
26500120.837.595.8
27000116.837.594.8
27500119.634.295.8
2800100111.334.294.8
285064116.434.394.8
290093115.231.995.8
295063116.737.395.8
3000100113.532.993.8
3050100117.135.094.8
31000120.134.597.8
320057117.534.592.8
32500118.535.595.8
3300100112.338.094.8
33500120.438.195.8
33900117.738.188.8

Worked out

ReadoutValueWorked out from
Run complete100%339 steps over 339
Step time6.13 s2023.8 s over 330 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute9.860 s over the step time of 6.13 s
Epoch interval708.5 sMean of 2 gaps between epoch marks
Train loss since first reading+0.12430.9181 at step 339 minus 0.7938 at step 5
Lowest train loss0.6665Lowest of 68 readings, at step 250
Eval accuracy, as shown30.0%0.3000 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval−0.20770.3000 minus 0.5077
Eval loss since first eval−0.07270.9899 minus 1.0626
ECE since first eval−0.05260.1510 minus 0.2036
Eval change−0.2077Eval 2 at 0.3000 minus eval 1 at 0.5077
Train minus eval loss−0.07180.9181 minus 0.9899, both at step 339
Wrong answers91130 eval items minus 39 correct. 39 over 130 is the only count that gives 0.3000
Rank7 of 7By last eval accuracy, across this project's 7 runs with an eval
Best so far0.9385Highest last eval accuracy, run 7
Against the run before−0.20770.3000 minus 0.5077 on run 3
Lowest last train loss0.0000Lowest last train loss across this project's 6 runs, run 2
Items per second1.31 step over 6.13 s, times batch size 8
Train loss against run 2+0.91810.9181 minus 0.0000 on run 2, both at step 339
Learning rate since first reading−5.00e-50.00e+0 minus 5.00e-5
Peak memory per GPU, whole GPU41.7 GBHighest sensor reading of any GPU, GPU 0, counting every process on it