CITADEL
StandbyNothing is live right now.
Eval accuracylast measured Wed 18 Nov, 03:42 UTC
%
0.8760438 of 500 eval, 62 wrong
▲ +86.40 pts across 4 evals▲ +1.60 pts on the eval before
0%25%50%75%100%0%25%50%75%100%0%25%50%75%100%

Familiar decider: Laya's encoder (backfilled)

finishedRadeon 8060S

One of the small models that decide whether Familiar, a desktop companion, reacts to something happening on the PC, and how. 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
Wed 18 Nov, 03:00 UTC
ran for
31:22
config
learning rate 0.00002, batch 16, seed 1 of 3

run4/5step1125/1125epoch3/3config2/3seed1/3lr0.00002batch16modellaya 395Mrev7b928d8items6000+500

Eval measures

Elapsed

627.4 s

epoch interval

627.4 sto eval 2627.4 sto eval 3627.4 sto eval 4

627.4 s to eval 2, 627.4 s to eval 3, 627.4 s to eval 4

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

1.67 s per step, 35.9 steps a minute

Eval accuracy by eval

+0.0160

▲ on the eval before

eval 1 0.0120, eval 4 0.8760

Against the other runs

3 of 5

by last eval accuracy

best so far 0.9100, run 2

▲ +0.0280 against run 3

Set-up

6000 train, 500 eval

synthetic, learning rate 0.00002, batch size 16, epochs 3, max length 512, precision bf16-mixed, loss ce, optimizer adamw, scheduler linear, seed 1

Seed spread, observed

0.0340
0.850.95

3 finished seeds

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

4 of 5 finished, 0.8760, config 2 of 3 (learning rate 0.00002, batch 16), seed 1 of 3, Wed 18 Nov, 03:00 UTC

Runs as a table
StartedStatusEval accuracyStepsConfig
Wed 18 Nov, 03:42 UTC| finished0.89401125 of 1125learning rate 0.00002, batch 16, seed 2 of 3
Wed 18 Nov, 03:00 UTC| finished0.87601125 of 1125learning rate 0.00002, batch 16, seed 1 of 3
Wed 18 Nov, 01:40 UTC| finished0.84801125 of 1125learning rate 0.00005, batch 16, seed 0
Wed 18 Nov, 00:58 UTC| finished0.91001125 of 1125learning rate 0.00002, batch 16, seed 0 of 3
Wed 18 Nov, 00:16 UTC| finished0.85601125 of 1125learning rate 0.0001, batch 16, seed 0

Plate

Model
convaiinnovations/laya, 395M parameters revision 7b928d8
Data
Familiar decisions, shared draw, synthetic6000 train500 eval
Config
learning rate 0.00002, 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
0.0 seval 1
627.4 seval 2
627.4 sepoch 1
1254.8 seval 3
1254.8 sepoch 2
1882.2 seval 4
1882.2 sepoch 3
9 h 52 min laterwent public

Seeds

SeedStatusEval accuracy
seed 0| finished0.9100
seed 1| finished0.8760
seed 2| finished0.8940

observed spread 0.0340, 3 finished seeds

Readings

StepSecondsEval accuracy
00.00.0120
375627.40.7920
7501254.80.8600
11251882.20.8760

Worked out

ReadoutValueWorked out from
Run complete100%1125 steps over 1125
Step time1.67 s627.4 s between epoch marks, over 375 steps an epoch. The reading interval holds the eval, so it is left out
Steps per minute35.960 s over the step time of 1.67 s
Epoch interval627.4 sMean of 2 gaps between epoch marks
Eval accuracy, as shown87.6%0.8760 times 100, to one decimal. Changes on that plate are percentage points
Eval accuracy since first eval+0.86400.8760 minus 0.0120
Eval change+0.0160Eval 4 at 0.8760 minus eval 3 at 0.8600
Wrong answers62500 eval items minus 438 correct. 438 over 500 is the only count that gives 0.8760
Rank3 of 5By last eval accuracy, across this project's 5 runs with an eval
Best so far0.9100Highest last eval accuracy, run 2
Against the run before+0.02800.8760 minus 0.8480 on run 3
Seed spread, observed0.0340High 0.9100 minus low 0.8760 across 3 finished seeds
Items per second9.61 step over 1.67 s, times batch size 16