Bases: EmbeddingModelOperationalMetric
Prompt tokens served from cache during embedding inference.
Example:
program.compile(
metrics=[
synalinks.metrics.EmbeddingCachedTokens(),
],
)
Source code in synalinks/src/metrics/em_metrics.py
| @synalinks_export("synalinks.metrics.EmbeddingCachedTokens")
class EmbeddingCachedTokens(EmbeddingModelOperationalMetric):
"""Prompt tokens served from cache during embedding inference.
Example:
```python
program.compile(
metrics=[
synalinks.metrics.EmbeddingCachedTokens(),
],
)
```
"""
def __init__(self, name="embedding_cached_tokens"):
super().__init__(name=name)
def result(self):
return int(self._delta("cached_tokens"))
|