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, 18:45 UTC
ran for
114:14
config
learning rate 0.00002, batch 1, seed 0 of 3

run1/3step84/84epoch1/1seed0/3lr0.00002batch1modelkev-4b 4Brev6cfce5c

Loss train 0.0000 ▼ −0.0065 lowest 0.0000 at step 16
Timebase
Range
Smoothing
84 readings, 30 min/div30 min/div30 min/div
012300:0030:0060:0090:00phase starting, 0.0 s to 81.6 sphase training, 81.6 s to 114:14epoch 1, 81.6 straining−0.0065since first0.0065Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.0000012300:0030:0060:0090:00phase starting, 0.0 s to 81.6 sphase training, 81.6 s to 114:14epoch 1, 81.6 straining0.0065Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.0000▼ −0.0065012300:0030:0060:0090:00phase starting, 0.0 s to 81.6 sphase training, 81.6 s to 114:14epoch 1, 81.6 straining0.0065Lowest last train loss across this project's 3 runs, run 1lowest last 0.0000, run 10.0000▼ −0.0065

Elapsed

81.6 s to epoch 1, 6773.1 s to last reading

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

81.6 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 0

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 ◂ older

1 of 3 finished, -, config 1 of 1 (learning rate 0.00002, batch 1), seed 0 of 3, Tue 10 Nov, 18:45 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 0
Hardware
Radeon 8060S, ROCm

Log

Since startMark
0.0 sphase starting
81.6 sphase training
81.6 sepoch 1
186 h 6 min laterwent public

Seeds

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

Readings

84 rows as a table
StepSecondsTrain lossLearning rate
181.60.00651.65e-6
2163.20.00784.06e-6
3244.80.00427.59e-6
4326.40.00101.16e-5
5408.00.00031.54e-5
6489.60.00031.84e-5
7571.20.00941.99e-5
8652.80.16312.00e-5
9734.40.00082.00e-5
10816.00.01221.99e-5
11897.60.00091.99e-5
12979.20.00011.98e-5
131060.80.00011.97e-5
141142.40.00171.96e-5
151224.10.00061.95e-5
161305.70.00001.94e-5
171387.30.00001.92e-5
181468.90.00741.90e-5
191550.50.00001.89e-5
201632.10.00001.87e-5
211713.70.00001.84e-5
221795.30.00001.82e-5
231876.90.00001.80e-5
241958.50.00001.77e-5
252040.10.00041.74e-5
262121.70.00001.72e-5
272203.30.00001.69e-5
282284.90.00001.66e-5
292366.50.00001.62e-5
302448.10.08811.59e-5
312529.70.00011.56e-5
322611.30.01201.52e-5
332692.90.00001.49e-5
342774.50.00001.45e-5
352856.10.00001.41e-5
362937.70.00001.37e-5
373019.30.00001.33e-5
383100.90.01991.29e-5
393182.50.00001.25e-5
403264.10.00001.21e-5
413345.70.00011.17e-5
423427.30.00021.13e-5
433509.00.00001.09e-5
443590.60.00471.05e-5
453672.20.00001.01e-5
463753.80.00009.67e-6
473835.40.00009.25e-6
483917.00.00008.84e-6
493998.60.00008.43e-6
504080.20.00008.02e-6
514161.80.00007.61e-6
524243.40.00007.21e-6
534325.00.00006.82e-6
544406.60.00006.42e-6
554488.20.00006.04e-6
564569.80.00005.66e-6
574651.40.00005.29e-6
584733.00.00004.93e-6
594814.60.00004.57e-6
604896.20.00004.23e-6
614977.80.00003.90e-6
625059.40.00003.57e-6
635141.00.00003.26e-6
645222.60.00002.96e-6
655304.22.64262.67e-6
665385.80.00002.39e-6
675467.40.00002.13e-6
685549.00.00531.88e-6
695630.60.00001.65e-6
705712.20.00001.42e-6
715793.90.00001.22e-6
725875.50.00001.03e-6
735957.10.00008.51e-7
746038.70.00006.91e-7
756120.30.00005.48e-7
766201.90.00004.20e-7
776283.50.08383.09e-7
786365.10.00002.15e-7
796446.70.00001.38e-7
806528.30.00007.77e-8
816609.90.00003.46e-8
826691.50.00008.71e-9
836773.10.00008.00e-11
846854.70.00008.71e-9

Worked out

ReadoutValueWorked out from
Run complete100%84 steps over 84
Step time81.6 s6773.1 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 81.60 s
Train loss since first reading−0.00650.0000 at step 84 minus 0.0065 at step 1
Lowest train loss0.0000Lowest of 84 readings, at step 16
Lowest last train loss0.0000Lowest last train loss across this project's 3 runs, run 1
Items per second0.01 step over 81.60 s, times batch size 1
Learning rate since first reading−1.64e-68.71e-9 minus 1.65e-6