PATTERN Cited by 1 source
Two-phase foundation-then-post-train¶
Pattern¶
Separate model training into two phases with different update cadences: a Phase 1 (Foundation) that adapts a pre-trained LLM on domain-specific corpora for broad understanding (updated infrequently), and a Phase 2 (Task-Specific) that post-trains this base for the target objective with task-specific data and constraints (refreshed frequently).
Problem¶
A single monolithic training pipeline that retrains everything from scratch on every refresh is: - Expensive — domain adaptation over large corpora is slow. - Unstable — catastrophic forgetting of general capabilities when optimizing for a narrow task. - Inflexible — the same model can't serve multiple downstream tasks without duplicating all compute.
Solution¶
- Phase 1: Adapt an open-source LLM on proprietary domain corpora (member behavior patterns, content understanding, domain-specific language). Updated infrequently. Serves as a shared backbone for multiple applications.
- Phase 2: Post-train on task-specific data (ranking labels, reward signals, business constraints). Updated frequently to track content and behavioral drift. Optimized under serving cost constraints.
The phases naturally decompose the training investment: Phase 1 captures stable domain knowledge; Phase 2 captures fast-moving task-specific signals.
Consequences¶
- Phase-1 → Phase-2 lift: +35–50% offline metrics (fresh), growing to ~80% after 2 weeks as Phase 1 stales — demonstrates clear separation of concerns.
- Phase-2 data efficiency: 10–40× fewer labeled examples needed to match a production baseline — Phase 1 provides implicit prior.
- Multiple Phase 2 models can share one Phase 1 backbone (amortized foundation cost).
- Phase 2 refresh cadence can be independent of Phase 1 (operational simplicity).
Seen in¶
- sources/2026-07-30-netflix-genrec-towards-llm-native-recommendation — "Phase 1 is updated relatively infrequently and serves as a shared, Netflix-aware backbone for many applications." Phase 2 "is refreshed more frequently to track new content and evolving tastes."
- sources/2026-02-13-netflix-scaling-llm-post-training-at-netflix — Netflix's Post-Training Framework implements the same two-phase separation with SFT/DPO/RL/KD recipes on top of a foundation model.