Training runs
Every experiment with its metrics and hardware
Runs this week
38
+9
Best val loss (ranker)
0.398
-0.021
GPU hours used
1,140
72% of quota
Queue wait
18m
-6m
Runs
Click a run for its curve and settings
- ranker-v7-lr3e4
- State
- running
- Epoch
- 14/20
- Val loss
- 0.412
- Accuracy
- 87.9%
- Hardware
- 8× A100
- GPU hours
- 6.2
- ranker-v7-lr1e4
- State
- finished
- Epoch
- 20/20
- Val loss
- 0.398
- Accuracy
- 88.6%
- Hardware
- 8× A100
- GPU hours
- 9.1
- intent-small-distil
- State
- finished
- Epoch
- 8/8
- Val loss
- 0.231
- Accuracy
- 94.2%
- Hardware
- 2× L4
- GPU hours
- 1.4
- ranker-v7-warmup
- State
- failed
- Epoch
- 3/20
- Val loss
- 1.120
- Accuracy
- 61.0%
- Hardware
- 8× A100
- GPU hours
- 0.8
- embed-multi-v3
- State
- queued
- Epoch
- 0/12
- Val loss
- —
- Accuracy
- —
- Hardware
- 16× H100
- GPU hours
- 0.0
- intent-base
- State
- finished
- Epoch
- 10/10
- Val loss
- 0.219
- Accuracy
- 94.9%
- Hardware
- 4× L4
- GPU hours
- 2.7