Output lines
Print these from your code, one per line.
| Line |
Fields |
KVERITAS_METRIC |
name=<id> value=<float> [step=<label>] |
KVERITAS_PHASE |
name=<phase> |
KVERITAS_CLAIM |
metric=<id> value=<float> [phase=<phase>] |
KVERITAS_INPUT |
src=seed:<value> |
KVERITAS_MODEL |
params=<int> arch=<name> precision=<fp16\|bf16\|fp32> |
KVERITAS_WORKLOAD |
dataset_size=<int> epochs=<float> batch_size=<int> [seq_len=<int>] |
KVERITAS_ARTIFACT |
role=<model\|dataset> [name=<ref>] path=<file> visibility=<public\|private> |
| Item |
Rule |
<id> |
Letter first, then letters, digits, underscores. |
<float> |
Integer, decimal, or scientific. |
step |
Any label without spaces. |
| Repeats |
A metric name repeated over steps is a series; the last value is final. |
| Line number |
Every claim and seed records the output line it came from. |
Examples
KVERITAS_PHASE name=train
KVERITAS_METRIC name=train_loss value=0.4182 step=3
KVERITAS_METRIC name=val_accuracy value=0.9142 step=3
KVERITAS_INPUT src=seed:42
KVERITAS_MODEL params=11689512 arch=resnet18 precision=fp32
KVERITAS_WORKLOAD dataset_size=50000 epochs=30 batch_size=128
KVERITAS_ARTIFACT role=dataset name=cifar10 path=data/cifar-10-batches-py visibility=public
KVERITAS_CLAIM metric=test_accuracy value=0.9387