promotion/Llama-3.1-8B-RewardedSoups-baseline

Rewarded Soups on UltraFeedback

Rewarded Soups (Rame et al., NeurIPS 2023): one expert per objective from the shared initialisation, linearly interpolated; the interpolation weight was chosen on validation prompts only.

• —Panel: UltraFeedback
• —Objectives: instruction following, truthfulness, honesty, helpfulness
• —Backbone / reference policy: meta-llama/Llama-3.1-8B-Instruct
• —Training budget: 300 optimiser updates, global batch 16
• —Reported in: Nash Bargaining Preference Optimization (NBPO), Table 2 (primary cross-method evaluation)

Evaluation protocol: independent objective-wise win rate against the common reference, judged by Llama-3.3-70B-Instruct on prompt-disjoint held-out prompts, both presentation orders.