Determines whether purchased products shares same attributes as goal

Returns list of available actions at the current step

Compute the log probabilities of the given labels under the given logits. Args: logits: Logits of the model (unnormalized). Shape

Generate samples from the model and reference model for the given batch of inputs.

Returns the evaluation [`~torch.utils.data.DataLoader`]. Subclass of transformers.src.transformers.trainer.get_eval_dataloader to pr

Get available memory for each GPU.

Get corresponding instruction text for environment current step

Get corresponding instruction text for current environment session

Calculate reward for purchased product's options w.r.t. goal options

Get reward value at current step of the environment

Determines the type reward - captures whether chosen product is in the same category

Check whether the output is a complete sentence.

Render and return the HTML for a product's sub page (i.e. description, features)

Load judge prompts. The return value is a python dict of type: Dict[judge_name: str -> dict]

Load model judgments. The return value is a dict of type: Dict[judge: Tuple -> Dict[game_key: tuple -> game_result: dict]

Loads products from the `items.json` file and combine them with reviews through `asin`. Return: dict[asin, product]

Load model judgments. The return value is a dict of type: Dict[judge: Tuple -> Dict[game_key: tuple -> game_result: dict]

A function that logs all the training stats. Call it at the end of each epoch. Args: stats (dict[str, Any]):

Calculate policy and value losses. Args: old_logprobs (`torch.FloatTensor`): Log probabilities of the mo

can be overrided to costomize the conversation template for different model workers.

Make dataset and collator for supervised fine-tuning.

Make dataset and collator for supervised fine-tuning.

Make the model names sorted in the game keys.

Make the model names sorted in a game key.

Check whether the text violates OpenAI moderation API.