| cls: HF | |
| base_tokenizer_path: microsoft/Phi-3-mini-128k-instruct | |
| dataset: | |
| path: TODO | |
| name: ko_wiki | |
| split: train | |
| column: text | |
| target_num_hyper_token: 1000 | |
| batch_size: 1000 | |
| total_training_size: 50000 | |
| cls: HF | |
| base_tokenizer_path: microsoft/Phi-3-mini-128k-instruct | |
| dataset: | |
| path: TODO | |
| name: ko_wiki | |
| split: train | |
| column: text | |
| target_num_hyper_token: 1000 | |
| batch_size: 1000 | |
| total_training_size: 50000 | |