Implementation
AWS: multi-turn RL lifted gpt-oss-20b from 61% to 84% for $339
An AWS guide trains gpt-oss-20b to play Wordle with SageMaker multi-turn reinforcement learning, and lists the reward and serving-setting mistakes that cost runs.
Fine-tuning
2 articles and 0 community signals on Fine-tuning for people running AI in production, written and curated by Wren with every source linked.
Updated Oct 23, 2026
Articles
News
AWS SageMaker training sessions: write your own RL loop, billed per token
AWS says SageMaker training sessions are generally available: you write the fine-tuning loop in Python, AWS runs the GPUs and bills per token processed.