DF

Governed RAG + PEFT

DomainForge

Live APIGitHub

DomainForge · taxonomy slice

Train plane — S0→S4 + adaptation methods

DomainForge owns the solution ladder and QLoRA/DPO training path. ModelForge owns posture, CUDA receipts, and the full cross-product taxonomy.

← Back to workbench · ModelForge taxonomy tab ↗ · venkat-ai.com/model-plane ↗

Adaptation methods (DomainForge scope)

MethodWhereStatus
Regular / base LMS0 baselineLive workbench
RAG (S1/S2)Facts planeLive — Chroma + hybrid
QLoRA SFT (S3)PEFT train CLILive — GPU path documented
DPO (S4)Preference pairsLive — win-rate compare
Multi-LoRA servevLLM Lab Path BEducational — ADR-022

Task types trained here

  • Intent classification (~27 intents)LLM JSON triage. Not sklearn tabular
  • Structured JSON / schemaS3/S4 + golden-eval. format_validate span
  • RAG-grounded factsS1/S2 retrieve. Separate from PEFT weights

Classical ML stack (org lane)

Tabular classification/regression, registry, skew, and ML CI/CD are interview-playbook depth — not a live sklearn service in this repo. See the full taxonomy classical ML tab for mlops-llmops study links and honesty labels.