HerrHruby/mr_midtrained_9b_v3_async_v2_cispo_step_150
032
MR 9B v3 async v2 CISPO — step 150
Hugging Face-compatible model weights exported from checkpoint step 150 of the wdblemve run. The model was initialized from HerrHruby/MR_midtrain_9B_v3. This checkpoint is intended for inference and evaluation.
