Doug Blank

I've been working on agent optimization for real-world prompts (prompt is ~10k tokens) and our new algorithm is already up 17% ! Seeing some interesting differences between benchmarks and real-world performance, more to come soon

1/ Look at how well AI handles mirror surfaces. Collab with @jperldev #ad made for @nvidia using instant-ngp. #neuralrendering #artificialintelligence #instantNeRF

@urig @Cometml BTW, here are the docs on the native Vertex integration: comet.com/docs/v2/integr…

Hi @Cometml. Do you offer your a service where my ML training code runs on your infra or do I run my training payloads in Vertex AI / SageMaker etc and integrate with Comet via API calls?

@urig @DataMovesHer @Cometml Yes! You can run experiments in any fashion, including programmatically (like the CLI tool --- feel free to us that code), or even use another hyperparameter search tool. See: #comet-optimize" target="_blank" rel="nofollow noopener">comet.com/docs/v2/api-an… and: comet.com/docs/v2/guides…

@DataMovesHer @Cometml Thanks Kristen 🌷 I have another question if I may - Is it possible to make the Comet optimizer run experiments in parallel, programmatically? That is, without using the comet CLI tool?