hku_diffusion_dllm / reference /code /flm /configs /data /openwebtext-streaming.yaml
Ouzhang's picture
Add files using upload-large-folder tool
31dc8dc verified
Raw
History Blame Contribute Delete
160 Bytes
train: openwebtext
valid: wikitext103
tokenizer_name_or_path: gpt2
cache_dir: /tmp/data
wrap: True
streaming: True
insert_train_eos: True
insert_valid_eos: True