OLMo/Tulu (Ai2)
1 model · Allen Institute for AI
The Allen Institute's Tulu — a fully open post-training recipe (SFT+DPO+RLVR) applied to Llama 3.1 405B, matching closed-lab finetunes with published data and code.
1 model · Allen Institute for AI
The Allen Institute's Tulu — a fully open post-training recipe (SFT+DPO+RLVR) applied to Llama 3.1 405B, matching closed-lab finetunes with published data and code.