CITADEL
StandbyNothing is live right now.
Eval accuracy
%
no eval

Claviger router: Kev-4B (LoRA) (backfilled)

finishedRadeon 8060S

A model that routes a request to the one destination, of many, that should take it. 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, 22:34 UTC
ran for
113:54
config
learning rate 0.00002, batch 1, seed 2 of 3

run3/3step84/84epoch1/1seed2/3lr0.00002batch1modelkev-4b 4Brev6cfce5c

Loss train 0.0000 ▼ −0.0523 lowest 0.0000 at step 24 run before, run 2 ▼ −0.0007 step 84
Timebase
Range
Smoothing
Reference
84 readings, 30 min/div30 min/div30 min/div
00.511.500:0030:0060:0090:00phase starting, 0.0 s to 81.4 sphase training, 81.4 s to 113:54epoch 1, 81.4 straining0.0523Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.000000.511.500:0030:0060:0090:00phase starting, 0.0 s to 81.4 sphase training, 81.4 s to 113:54epoch 1, 81.4 straining0.0523Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.0000▼ −0.052300.511.500:0030:0060:0090:00phase starting, 0.0 s to 81.4 sphase training, 81.4 s to 113:54epoch 1, 81.4 straining0.0523Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.0000▼ −0.0523

Elapsed

81.4 s to epoch 1, 6753.4 s to last reading

84 of 84 steps, 1 of 1 epochs, 100%

81.4 s per step, 0.7 steps a minute

Set-up

666 train

synthetic, learning rate 0.00002, batch size 1, grad accum 8, epochs 1, optimizer adamw, seed 2

Learning rate

8.71e-9

▼ −1.64e-6 since first

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 newer ▸

3 of 3 finished, -, config 1 of 1 (learning rate 0.00002, batch 1), seed 2 of 3, Tue 10 Nov, 22:34 UTC

▸
Runs as a table
StartedStatusEval accuracyStepsConfig
Tue 10 Nov, 22:34 UTC| finished-84 of 84learning rate 0.00002, batch 1, seed 2 of 3
Tue 10 Nov, 20:40 UTC| finished-84 of 84learning rate 0.00002, batch 1, seed 1 of 3
Tue 10 Nov, 18:45 UTC| finished-84 of 84learning rate 0.00002, batch 1, seed 0 of 3

Plate

Model
jaredpalmer/kev-4b, 4B parameters revision 6cfce5c
Data
Synthetic routing worlds, synthetic
Config
learning rate 0.00002, batch size 1, grad accum 8, epochs 1, optimizer adamw, seed 2
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
81.4 sphase training
81.4 sepoch 1
182 h 18 min laterwent public

Seeds

SeedStatusEval accuracy
seed 0| finished-
seed 1| finished-
seed 2| finished-

Readings

84 rows as a table
StepSecondsTrain lossLearning rate
181.40.05231.65e-6
2162.70.00054.06e-6
3244.10.01117.59e-6
4325.50.01681.16e-5
5406.80.00321.54e-5
6488.20.08841.84e-5
7569.60.00531.99e-5
8650.90.00092.00e-5
9732.30.12132.00e-5
10813.70.00131.99e-5
11895.00.00091.99e-5
12976.40.01971.98e-5
131057.80.00011.97e-5
141139.10.00001.96e-5
151220.50.00031.95e-5
161301.90.00001.94e-5
171383.20.00001.92e-5
181464.60.00011.90e-5
191546.00.00021.89e-5
201627.30.00011.87e-5
211708.70.00161.84e-5
221790.10.00101.82e-5
231871.40.00001.80e-5
241952.80.00001.77e-5
252034.20.00001.74e-5
262115.50.00001.72e-5
272196.90.00021.69e-5
282278.30.00001.66e-5
292359.60.00001.62e-5
302441.00.00001.59e-5
312522.40.00001.56e-5
322603.70.00111.52e-5
332685.10.00001.49e-5
342766.50.00001.45e-5
352847.80.00021.41e-5
362929.20.00061.37e-5
373010.60.02091.33e-5
383091.90.00031.29e-5
393173.30.00001.25e-5
403254.70.00001.21e-5
413336.00.00001.17e-5
423417.40.00001.13e-5
433498.80.10791.09e-5
443580.10.00001.05e-5
453661.50.00001.01e-5
463742.90.03579.67e-6
473824.20.00009.25e-6
483905.60.00158.84e-6
493987.00.00008.43e-6
504068.30.00828.02e-6
514149.70.00007.61e-6
524231.10.00007.21e-6
534312.40.00006.82e-6
544393.80.00006.42e-6
554475.20.00016.04e-6
564556.50.00005.66e-6
574637.90.63805.29e-6
584719.30.00004.93e-6
594800.60.00014.57e-6
604882.00.00004.23e-6
614963.40.00003.90e-6
625044.70.00003.57e-6
635126.11.24923.26e-6
645207.50.00032.96e-6
655288.80.00002.67e-6
665370.20.00002.39e-6
675451.60.00002.13e-6
685532.90.00001.88e-6
695614.30.00001.65e-6
705695.70.00001.42e-6
715777.00.00011.22e-6
725858.40.00001.03e-6
735939.80.00008.51e-7
746021.10.00006.91e-7
756102.50.00005.48e-7
766183.90.00004.20e-7
776265.20.00003.09e-7
786346.60.00002.15e-7
796428.00.00001.38e-7
806509.30.00007.77e-8
816590.70.00003.46e-8
826672.10.00008.71e-9
836753.40.00008.00e-11
846834.80.00008.71e-9

Worked out

ReadoutValueWorked out from
Run complete100%84 steps over 84
Step time81.4 s6753.4 s over 83 steps, between readings. Startup and intervals that hold an eval are left out
Steps per minute0.760 s over the step time of 81.37 s
Train loss since first reading−0.05230.0000 at step 84 minus 0.0523 at step 1
Lowest train loss0.0000Lowest of 84 readings, at step 24
Lowest last train loss0.0000Lowest last train loss across this project's 3 runs, run 1
Items per second0.01 step over 81.37 s, times batch size 1
Train loss against run 2−0.00070.0000 minus 0.0007 on run 2, both at step 84
Learning rate since first reading−1.64e-68.71e-9 minus 1.65e-6