OLMo 2 1124 7B Instruct
Allen Institute for AI (Ai2) · 2025-01-03 · 7.3B parameters
OLMo-2-1124-7B-Instruct is Ai2's dense text-only OLMo 2 7B instruction checkpoint, post-trained from the OLMo 2 7B base model with supervised fine-tuning on a Tülu 3 SFT mixture, DPO, and RLVR. After the initial November 26, 2024 post-trained release used pre-tokenization logic that differed from the base model, Ai2 retrained the post-trained model and released the corrected weights under the same repository ID on January 3, 2025; the original weights are preserved as OLMo-2-1124-7B-Instruct-preview.