Index

_

__call__() (utils.LengthSampler method) _assess_postalignment_singlevalue() (rewardProcessor.RewardProcessor method)

_dual_objective() (alignValues.AlignValues method) _save_results_to_csv() (alignValues.AlignValues method) _save_results_to_text() (alignValues.AlignValues method)

A

add_reward() (rewardProcessor.RewardProcessor method) alignValues module

AlignValues (class in alignValues) ALL_SUPPORTED_VALUES (in module utils) assess_original_value() (rewardProcessor.RewardProcessor method) assess_postalignment_multivalue() (rewardProcessor.RewardProcessor method)

B

base_model_file (in module plot_cal_winrate) basemodel_name (gendata.TextGeneration attribute), [1]

batch_size (rewardProcessor.RewardProcessor attribute), [1] build_dataset() (in module trainPPO)

C

c (alignValues.AlignValues attribute) cal_coherence() (in module utils) cal_diversity() (in module utils) cal_gpt2_harmless_probabilities() (in module utils) cal_gpt2_helpful_probabilities() (in module utils) cal_harmless_probabilities() (in module utils) cal_humor_probabilities() (in module utils) cal_log_perplexity() (in module utils)

cal_positive_sentiment() (in module utils) calculate_win_rate() (in module plot_cal_winrate) categories (in module plot_radar_tabresults) (in module plot_sequential_baseline) clean_and_trim_to_last_sentence() (in module utils) collator() (in module trainPPO) collect_multiple_results() (in module plot_cal_winrate) compute_rep_n() (in module utils) convert_ppo_modelname_to_huggingface_valid() (in module utils)

D

data (in module plot_sequential_baseline) data_name (gendata.TextGeneration attribute), [1] data_source (in module genDPOdata)

device (gendata.TextGeneration attribute), [1] (in module gendataGUI) devices (in module utils)

F

file_path (alignValues.AlignValues attribute), [1] (gendata.TextGeneration attribute), [1] (in module plot_reward_dist) (rewardProcessor.RewardProcessor attribute), [1]

find_pareto_by_interpolation() (alignValues.AlignValues method) find_pareto_by_oneValue() (alignValues.AlignValues method)

G

gen_mixed_preference_data() (in module genDPOdata) gen_rand_MAP_lambda() (alignValues.AlignValues method) gendata module gendataGUI module genDPOdata module generate_from_MC_aligned_model() (gendata.TextGeneration method) generate_from_original_model() (gendata.TextGeneration method)

generateGUI_from_MC_aligned_model() (in module gendataGUI) generateGUI_from_original_model() (in module gendataGUI) generation_config (gendata.TextGeneration attribute), [1] get_device() (in module utils) get_model_and_tokenizer() (in module utils) get_nvidia_smi_info() (in module utils) get_prompts_from_Anthropic_harmless() (in module utils) get_prompts_from_imdb() (in module utils) get_reward() (in module utils) getDPOsoup module

I

iface (in module gendataGUI)

L

LengthSampler (class in utils)

M

main() (in module trainPPO) map_results (in module plot_sequential_baseline) max_length (utils.LengthSampler attribute) max_new_tokens (gendata.TextGeneration attribute), [1] merge_added_rewards() (mergeProcessor.MergeProcessor method) merge_gendata_bypattern() (mergeProcessor.MergeProcessor method) mergeProcessor module MergeProcessor (class in mergeProcessor) min_length (utils.LengthSampler attribute) model (in module plot_pairwise_pareto) model_setting (in module plot_sequential_baseline) module alignValues gendata gendataGUI genDPOdata getDPOsoup mergeProcessor plot_cal_winrate plot_pairwise_pareto plot_radar_tabresults plot_rand_lambda_MAP_region plot_reward_dist plot_sequential_baseline rewardProcessor trainDPO trainPPO utils

O

optimize_lambda() (alignValues.AlignValues method)

P

plot_all_setups() (in module plot_sequential_baseline) plot_cal_winrate module plot_cLevels() (in module plot_cal_winrate) plot_helpful_vs_harmless() (in module plot_cal_winrate) plot_hist() (in module plot_reward_dist) plot_hist_positive() (in module plot_reward_dist) plot_lambdas_2D_subplots() (in module plot_rand_lambda_MAP_region) plot_lambdas_3D() (in module plot_rand_lambda_MAP_region) plot_matrix_scatterplot() (in module plot_pairwise_pareto) plot_pairwise_pareto module plot_pareto() (in module plot_pairwise_pareto)

plot_radar_chart() (in module plot_radar_tabresults) plot_radar_tabresults module plot_rand_lambda_MAP_region module plot_reward_dist module plot_separate_setup() (in module plot_sequential_baseline) plot_sequential_baseline module plot_weighted_unweighted_histograms() (in module plot_reward_dist) plot_winrate() (in module plot_cal_winrate) preprocess_data() (in module genDPOdata)

Q

quantile_transform() (in module plot_pairwise_pareto)

quantile_transform_single_c() (rewardProcessor.RewardProcessor method)

R

render_latex_table() (in module plot_cal_winrate) rewardProcessor module

RewardProcessor (class in rewardProcessor) rewards (alignValues.AlignValues attribute) round_1_results (in module plot_sequential_baseline) round_5_results (in module plot_sequential_baseline)

S

save_model_and_tokenizer() (in module getDPOsoup) save_results_to_json() (in module utils) seed (in module genDPOdata)

sequential_optimize_lambda() (alignValues.AlignValues method) sequential_optimize_lambda_multiround() (alignValues.AlignValues method) soup() (in module getDPOsoup)

T

TASK_NAME (in module utils) TextGeneration (class in gendata) top_k (gendata.TextGeneration attribute), [1] train_dpo() (in module trainDPO)

trainDPO module trainPPO module trim_to_word_boundary() (utils.LengthSampler method)

U

utils module

V

value_list (alignValues.AlignValues attribute) values_to_align (rewardProcessor.RewardProcessor attribute)

values_to_align_str (rewardProcessor.RewardProcessor attribute) values_to_evaluate (rewardProcessor.RewardProcessor attribute) values_to_evaluate_str (rewardProcessor.RewardProcessor attribute)

X

x_value (in module plot_rand_lambda_MAP_region)