Scalytics 3.5.0 Is Here: Enterprise-Grade vLLM Inference, Torch Resilience, and Deep Search Refined with Private-Data Training
Scalytics 3.5.0 combines enterprise-grade streaming inference powered by vLLM with resilient PyTorch model support, a fully configurable graph-based Deep Search engine, and end-to-end private-data fine-tuning workflows. Dynamic context-window detection and GPU-aware allocation optimize performance, while simplified tokenizer checks and robust JSON handling ensure Torch model reliability. Deep Search 3.5.0’s LangGraph architecture offers multi-hop retrieval, vector pre-filtering, and legal data integration. Secure token management, JWT-backed model activation, and a storage insights API deliver transparent governance and continuous private-data refinement behind your firewall.