metadata
base_model: >-
RefalMachine/llama3_extended_darulm_20_05_24_part1-2_64000_bpe_full_lr2e4_bs256
datasets:
- IlyaGusev/saiga_scored
language:
- ru
- en
Model description
LoRa tuned version of RefalMachine/llama3_extended_darulm_20_05_24_part1-2_64000_bpe_full_lr2e4_bs256 after LEP (Learned Embedding Propagation, paper will be soon) procedure on saiga_scored d7 dataset.