|
|
--- |
|
|
{} |
|
|
--- |
|
|
# allenai/open_instruct: Generation Dataset |
|
|
|
|
|
See https://github.com/allenai/open-instruct/blob/main/docs/algorithms/rejection_sampling.md for more detail |
|
|
|
|
|
## Configs |
|
|
|
|
|
``` |
|
|
args: |
|
|
{'add_timestamp': True, |
|
|
'dataset_end_idx': 143000, |
|
|
'dataset_mixer_list': ['VGraf/no_safety_tulu_pref', '170000'], |
|
|
'dataset_shuffle_seed': 42, |
|
|
'dataset_splits': ['train', 'train'], |
|
|
'dataset_start_idx': 142000, |
|
|
'hf_entity': 'VGraf', |
|
|
'hf_repo_id': 'generation_multi', |
|
|
'include_persona': False, |
|
|
'max_qs': 8, |
|
|
'mode': 'generation', |
|
|
'model2_name_or_path': 'gpt-3.5-turbo-0125', |
|
|
'model_name_or_path': 'gpt-3.5-turbo-0125', |
|
|
'mt_token_cutoff': 8000, |
|
|
'num_turns': 4, |
|
|
'push_to_hub': True, |
|
|
'revision': 'main', |
|
|
'revision2': 'main', |
|
|
'save_filename': '/weka/oe-adapt-default/victoriag/synth_data/self-talk/clarify_nopersona_1000samples_4turns_1completions_gpt3.5_gpt3.5_tulupref.jsonl', |
|
|
'skill': 'chat', |
|
|
'user_type': 'clarify'} |
|
|
|
|
|
dataset_config: |
|
|
{'batched': False, |
|
|
'binary_messages_key': 'messages', |
|
|
'chat_template': 'simple_chat', |
|
|
'convert_preference_to_binary_dataset': False, |
|
|
'label': 'binary_labels', |
|
|
'load_from_cache_file': False, |
|
|
'max_prompt_token_lenth': None, |
|
|
'max_token_length': None, |
|
|
'ncols': 2, |
|
|
'num_proc': 1, |
|
|
'preference_chosen_key': 'chosen', |
|
|
'preference_rejected_key': 'rejected', |
|
|
'sanity_check': True, |
|
|
'sanity_check_max_samples': 100, |
|
|
'sft_messages_key': 'messages', |
|
|
'train_only_on_prompt': False} |
|
|
|
|
|
gen_args: |
|
|
{'num_completions': 1, |
|
|
'response_length': 2048, |
|
|
'temperature': 1.0, |
|
|
'tensor_parallel_size': 1, |
|
|
'top_p': 0.9} |
|
|
``` |
|
|
|
|
|
## Reproduce this dataset |
|
|
|
|
|
1. Download the `['generation_multi.py', 'clarify_nopersona_1000samples_4turns_1completions_gpt3.5_gpt3.5_tulupref.jsonl']` from the https://huggingface.co/datasets/VGraf/generation_multi_1749247351. |
|
|
2. Run `python open_instruct/rejection_sampling/generation_multi.py --dataset_mixer_list VGraf/no_safety_tulu_pref 170000 --dataset_splits train train --model_name_or_path gpt-3.5-turbo-0125 --model2_name_or_path gpt-3.5-turbo-0125 --dataset_shuffle_seed 42 --dataset_start_idx 142000 --dataset_end_idx 143000 --num_turns 4 --num_completions 1 --user_type clarify --save_filename /weka/oe-adapt-default/victoriag/synth_data/self-talk/clarify_nopersona_1000samples_4turns_1completions_gpt3.5_gpt3.5_tulupref.jsonl --sanity_check --push_to_hub --temperature 1.0` |
|
|
|
