Why should you monitor your usage of Anyscale?
Monitor your application powered by Anyscale language models to ensure, get visibility to what you send to Anyscale, responses received from Anyscale, latency, usage and errors. By monitoring the usage, you can infer the cost.
Track the LLM's performance:
Monitor the input & output, latency and errors of your LLM provider. Track performance changes with the providers and versions of your LLM. Monitor usage to understand the cost, rate limits, and general performance.
Track your app:
By tracking key metrics like latency, throughput, error rates, and input & output, you can gain insights into your LangChain app's performance and identify areas of improvement.
Early issue detection:
Detect and address issues early to prevent them from affecting model performance.
Comprehensive Anyscale monitoring quickstart
Our Anyscale quickstart provides metrics including error rate, input & output, latency, queries, and lets you integrate with different language models.
What’s included in the Anyscale quickstart?
New Relic Anyscale monitoring quickstart provides a variety of pre-built dashboards, which will help you gain insights into the health and performance of your Anyscale usage. These reports include:
- Dashboards (average tokens, LLM completion’s details, chain’s details, tool details, top tool names and many more)
- Alerts (errors, request per model and response time)