CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Sun 22 Nov, 00:55 UTC
%
0.8769114 of 130 eval, 16 wrong
▲ +36.92 pts across 2 evals▲ +36.92 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
Sun 22 Nov, 00:18 UTC
ran for
36:01
config
learning rate 0.00002, batch 8, seed 0

run5/7step339/339epoch3/3config3/4seed0lr0.00002batch8modelOneJev1.6-0.8B 0.8Brevc89933citems1800+130

Loss train 0.0009 ▼ −1.0510 eval 0.5878 lowest 0.0004 at step 255 run before, run 4 ▼ −0.9172 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 90.0 sphase training, 90.0 s to 36:01epoch 1, 90.0 sepoch 2, 13:59epoch 3, 25:06trainingepoch 2epoch 3−1.0510since first1.0519Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 20.000900.511.5200:0010:0020:0030:00phase starting, 0.0 s to 90.0 sphase training, 90.0 s to 36:01epoch 1, 90.0 sepoch 2, 13:59epoch 3, 25:06trainingepoch 2epoch 31.0519Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 2eval 0.58780.0009▼ −1.051000.511.5200:0005:0010:0015:0020:0025:0030:0035:00phase starting, 0.0 s to 90.0 sphase training, 90.0 s to 36:01epoch 1, 90.0 sepoch 2, 13:59epoch 3, 25:06trainingepoch 2epoch 31.0519Lowest last train loss across this project's 6 runs, run 2lowest last 0.0000, run 2eval 0.58780.0009▼ −1.0510

Eval measures

Eval losslower is better
0.5878▼ −0.4748 since first
02 →
ECEcalibration error, lower is better
0.1166▼ −0.0870 since first
01

Elapsed

708.3 s

epoch interval

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

6.19 s per step, 9.7 steps a minute

Eval accuracy by eval

+0.3692

▲ on the eval before

eval 1 0.5077, eval 2 0.8769

Against the other runs

4 of 7

by last eval accuracy

best so far 0.9385, run 7

▲ +0.5769 against run 4

Train minus eval

−0.5869

step 339

train 0.0009, eval 0.5878

Set-up

1800 train, 130 eval

synthetic, learning rate 0.00002, batch size 8, grad accum 2, epochs 3, optimizer adamw, scheduler cosine, seed 0

Learning rate

0.00e+0

▼ −1.00e-5 since first

On one axis

phase starting, 0.0 s to 90.0 sphase training, 90.0 s to 36:01E1E2E3traininglowest last 0.0000Train loss0.0009▼ −1.05102.000.001.05190.0009last ▼ −1.0510best 0.9385, run 7Last-value range, 7 runs, 0.3000 to 0.9385Eval accuracy0.8769▲ +0.36921.000.250.50770.8769last ▲ +0.3692Learning rate0.00e+0▼ −1.00e-52.0e-50.0e+01.00e-5last 0.00e+0Power, whole chip, CPU and GPU118 W▲ +28 W140 W80 W90 Wlast 118 WHottest, one board zone90 °C▼ −8 °C100 °C85 °C98 °Clast 90 °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

use100%
memory, whole GPU38.1 GB

peak 41.7 GB, whole GPU

90 °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

5 of 7 finished, 0.8769, config 3 of 4 (learning rate 0.00002, batch 8), seed 0, Sun 22 Nov, 00:18 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 0.00002, 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
13.4 swent public
40.1 seval 1
90.0 sphase training
90.0 sepoch 1
839.6 sepoch 2
1506.7 sepoch 3
2161.5 seval 2

Readings

69 rows as a table
StepSecondsTrain lossEval lossEval accuracyECELearning rate
040.1-1.06260.50770.2036-
590.01.0519---1.00e-5
10132.40.7411---2.00e-5
15170.80.5990---2.00e-5
20209.20.9493---2.00e-5
25250.00.5313---1.99e-5
30288.10.5987---1.98e-5
35328.00.8266---1.97e-5
40360.00.5588---1.96e-5
45393.51.0115---1.94e-5
50427.70.2170---1.93e-5
55462.20.5476---1.91e-5
60495.61.0619---1.89e-5
65532.60.3052---1.87e-5
70562.50.3065---1.84e-5
75592.60.4267---1.81e-5
80621.60.8246---1.78e-5
85658.20.3644---1.75e-5
90691.70.4404---1.72e-5
95721.70.5530---1.69e-5
100750.91.0610---1.65e-5
105782.50.4850---1.62e-5
110813.60.8582---1.58e-5
115839.60.4048---1.54e-5
120868.80.2309---1.50e-5
125896.30.4249---1.46e-5
130926.60.2451---1.41e-5
135955.90.2962---1.37e-5
140988.70.1813---1.32e-5
1451017.40.7052---1.28e-5
1501046.10.2051---1.23e-5
1551074.40.1963---1.19e-5
1601102.90.2471---1.14e-5
1651130.80.3661---1.09e-5
1701159.80.7624---1.04e-5
1751190.30.7924---9.95e-6
1801217.31.0178---9.48e-6
1851246.20.0532---9.00e-6
1901275.10.0223---8.53e-6
1951304.70.2551---8.05e-6
2001333.40.8939---7.59e-6
2051364.20.0178---7.13e-6
2101392.50.5809---6.67e-6
2151424.01.0609---6.23e-6
2201452.20.4756---5.79e-6
2251480.80.3427---5.36e-6
2301506.70.0008---4.94e-6
2351534.90.0007---4.54e-6
2401563.41.3988---4.15e-6
2451590.90.0030---3.77e-6
2501621.70.0064---3.40e-6
2551651.10.0004---3.05e-6
2601680.70.0008---2.71e-6
2651710.70.3655---2.39e-6
2701739.20.0016---2.09e-6
2751768.40.0011---1.81e-6
2801796.10.0087---1.55e-6
2851824.90.6866---1.30e-6
2901852.80.0008---1.07e-6
2951882.50.1875---8.70e-7
3001911.30.0008---6.85e-7
3051942.20.0006---5.22e-7
3101970.50.3868---3.81e-7
3152000.50.0005---2.61e-7
3202027.70.0008---1.64e-7
3252055.60.0015---8.92e-8
3302084.70.0006---3.69e-8
3352113.40.2756---7.29e-9
3392161.50.00090.58780.87690.11660.00e+0

GPU readings

69 rows as a table
StepGPUUse %Power W, whole chip, CPU and GPUMemory GB, whole GPUTemp C, one board zone
005390.026.697.8
500119.429.397.8
1000118.731.397.8
1506397.429.797.8
200100110.814.397.8
2500118.229.297.8
300100107.915.097.8
3500117.431.697.8
400100114.931.197.8
45085104.133.996.8
500100113.431.396.8
5500119.030.397.8
600100109.016.097.8
6500115.730.096.8
7000121.233.097.8
7505087.926.595.8
8000118.033.295.8
85069101.432.497.8
900100111.133.996.8
9500118.433.293.8
10000117.434.894.8
10500118.534.196.8
1100100112.933.096.8
11504484.320.992.8
1200100117.234.894.8
1250100112.934.496.8
130010096.219.595.8
1350100118.933.397.8
14000117.541.793.8
1450100117.241.795.8
1550100113.841.793.8
160057117.941.795.8
16500119.241.795.8
1700100118.333.195.8
17500119.834.193.8
18000117.134.195.8
1850100117.334.195.8
19000118.036.394.8
1950100117.236.394.8
200066118.037.495.8
21000120.337.493.8
21500123.329.796.8
2200100113.133.694.8
2250100111.337.895.8
2300100117.737.895.8
23500118.137.894.8
2400100111.937.893.8
245046112.725.095.8
25000118.334.295.8
260097116.434.096.8
2650100116.937.594.8
270091119.037.594.8
2750100116.234.295.8
280067118.934.295.8
2850100115.134.393.8
29000120.831.996.8
29500117.537.394.8
300071119.935.093.8
30500118.435.094.8
3150100112.734.596.8
3200100111.134.592.8
3250100116.035.594.8
330064119.738.094.8
3350100117.938.195.8
3390100117.938.189.8

Worked out

ReadoutValueWorked out from
Run complete100%339 steps over 339
Step time6.19 s2073.4 s over 335 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute9.760 s over the step time of 6.19 s
Epoch interval708.3 sMean of 2 gaps between epoch marks
Train loss since first reading−1.05100.0009 at step 339 minus 1.0519 at step 5
Lowest train loss0.0004Lowest of 68 readings, at step 255
Eval accuracy, as shown87.7%0.8769 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval+0.36920.8769 minus 0.5077
Eval loss since first eval−0.47480.5878 minus 1.0626
ECE since first eval−0.08700.1166 minus 0.2036
Eval change+0.3692Eval 2 at 0.8769 minus eval 1 at 0.5077
Train minus eval loss−0.58690.0009 minus 0.5878, both at step 339
Wrong answers16130 eval items minus 114 correct. 114 over 130 is the only count that gives 0.8769
Rank4 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+0.57690.8769 minus 0.3000 on run 4
Lowest last train loss0.0000Lowest last train loss across this project's 6 runs, run 2
Items per second1.31 step over 6.19 s, times batch size 8
Train loss against run 4−0.91720.0009 minus 0.9181 on run 4, both at step 339
Learning rate since first reading−1.00e-50.00e+0 minus 1.00e-5
Peak memory per GPU, whole GPU41.7 GBHighest sensor reading of any GPU, GPU 0, counting every process on it