DBTB INT Chester Chen r
summary
Demonstrates how to instrument ML algorithms to stream real-time ML visualizations that reveal cost function shape and convergence behavior during long runs, not just final metrics. Shows simple code changes and a lightweight data path used at Alpine Data to emit training traces, and how those traces drive dashboards and alerts. Proposes concrete hooks into Spark and lightweight adapters so models expose monitor cost function convergence and intermediate diagnostics to production systems.
AI-written summary · about
related talks
32mMaking Models Work: How an optimized ML lifecycle can drive better ROI
10mML-based Log Analysis for Faster Debugging
29mML in Production on Databricks
37mBig Data Analytics, Machine Learning, Real Time Processing
28mDataEngBytes 2023 - PER-T1-05 - Ellie Gholami
16mKeynote: DBRX: Training Modern LLMs on Databricks
16mOperationalizing AI in Finance: Real-Time Analytics
14mTony Baer, dbinsight & Doug Henschen, Constellation Research | Databricks Data + AI Summit 2023