Experiment Combines Continued Pretraining with RAG for Subway Navigation

A developer combined continued pretraining with RAG, training Qwen 3.5 4B via Unsloth on a fictional subway system so it can plan multi-transfer routes, while RAG handles dynamic information updates.

2026-09-16 ~ 2026-09-16 · 2 related posts

• Experiment: continued pretraining for stable domain knowledge, RAG for dynamic data — funJS · 2026-09-16
• Combining Continued Pretraining with RAG: Teaching Qwen 3.5 4B a Fictional Subway Map — funJS · 2026-09-16