prompting.validators.event

Module Contents

Classes

EventSchema

class prompting.validators.event.EventSchema

completions: List[str]

completion_times: List[float]

completion_status_messages: List[str]

completion_status_codes: List[str]

name: str

task_type: str

block: float

gating_loss: float

uids: List[int]

prompt: str

step_length: float

best: str

rewards: List[float]

dahoas_reward_model: List[float] | None

blacklist_filter: List[float] | None

blacklist_filter_matched_ngram: List[str] | None

blacklist_filter_significance_score: List[float] | None

nsfw_filter: List[float] | None

reciprocate_reward_model: List[float] | None

diversity_reward_model: List[float] | None

diversity_reward_model_historic: List[float] | None

diversity_reward_model_batch: List[float] | None

dpo_reward_model: List[float] | None

rlhf_reward_model: List[float] | None

prompt_reward_model: List[float] | None

relevance_filter: List[float] | None

dahoas_reward_model_normalized: List[float] | None

nsfw_filter_normalized: List[float] | None

nsfw_filter_score: List[float] | None

reciprocate_reward_model_normalized: List[float] | None

diversity_reward_model_normalized: List[float] | None

dpo_reward_model_normalized: List[float] | None

rlhf_reward_model_normalized: List[float] | None

prompt_reward_model_normalized: List[float] | None

relevance_filter_normalized: List[float] | None

relevance_filter_bert_score: List[float] | None

relevance_filter_mpnet_score: List[float] | None

task_validation_penalty_raw: List[float] | None

task_validation_penalty_adjusted: List[float] | None

task_validation_penalty_applied: List[float] | None

keyword_match_penalty_raw: List[float] | None

keyword_match_penalty_adjusted: List[float] | None

keyword_match_penalty_applied: List[float] | None

sentence_match_penalty_raw: List[float] | None

sentence_match_penalty_adjusted: List[float] | None

sentence_match_penalty_applied: List[float] | None

set_weights: List[List[float]] | None

static from_dict(event_dict, disable_log_rewards)

Converts a dictionary to an EventSchema object.

Parameters:

• event_dict (dict) –

• disable_log_rewards (bool) –

Return type:

EventSchema