OPT-2.7B Standard SFT

facebook/opt-2.7b fine-tuned normally (dense): all neurons in the last layers' fc1 are updated, with no predictor-based masking.

This is the baseline to compare against the predictor-guided version: PoHao/opt2.7-predictor-guided-sft.

Each subfolder is one training run (different random seed):

opt2.7_SFT_Standard_numlayers32_lr2e-5__seed380/

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for PoHao/opt2.7-standard-sft

Base model

Finetuned

(6)

this model