r""" Handles the early stopping logic. If the policy KL is greater than the target KL, then the gradient is zeroed and the optimizatio

Fast generation with embeddings using dLLM cache optimization. This method incorporates all fast dLLM related optimizations.

Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token first unpad the input, th

Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token first unpad the input, th

Generate responses for a list of query tensors. args: query_tensors (list[torch.Tensor]): A list of query tensors to gen

Generate samples from the model Args: iterations (int): Number of iterations to generate samples for batch_s

Get configuration parameters.

Get configuration parameters.

Extracts the model name from the model arguments.

Precompute the number of tokens to transition at each step.

Creates a dictionary of tasks index with the following metadata, - `type`, that can be either `task`, `python_task`, `group` or `tags`.

Handle non-serializable objects by converting them to serializable types. Args: o (Any): The object to be handled. Returns:

r""" Initializes the weights of the value head. The default initialization strategy is random. Users can pass a different initializati

r""" We initialize the weights of the value head.

Log results as JSON artifact to W&B.

Log results as JSON artifact to W&B.

Generate and log evaluation results as a table to W&B.

Generate and log evaluation results as a table to W&B.

r""" Some pre-processing hacks to make the model `accelerate` compatible. Check https://github.com/huggingface/transformers/pull/21707

Validate the `rope_scaling` configuration.

Validate the `rope_scaling` configuration.

Sanitize the results dictionary.

Sanitize the results dictionary.

Check if the input data is valid for training. Args: input_ids (List[`torch.LongTensor`]): List of tenso

Check if the input data is valid for training. Args: batch_size (int): Batch size from the config file.

Given a set of prompts and number of tokens to generate: - tokenize prompts - set the sequence length to be the max of length of prompts

Train on a batch of samples. Main training segment Args: inner_epoch (int): The current inner epoch epoch (i

r""" Activates the neftune as presented in this code: https://github.com/neelsjain/NEFTune and paper: https://arxiv.org/abs/2310.05914

r""" Add and load a reward modeling adapter. This method can only be used if the model is a `PeftModel` and if you have initialized th

https://github.com/QwenLM/Qwen-VL/blob/master/eval_mm/infographicsvqa_eval.py

https://github.com/QwenLM/Qwen-VL/blob/master/eval_mm/infographicsvqa_eval.py

Build a set of Instances for a task, and store them in task.instances

Calculate the loss for a batch of an unpacked sample Args: latents (torch.Tensor): The latents sampled f

Check if the given string a number. https://github.com/MMMU-Benchmark/MMMU/blob/51ce7f3e829c16bb44bc5445782686b4c3508794/eval/eval_utils.py#L

Check if the given string a number. https://github.com/MMMU-Benchmark/MMMU/blob/51ce7f3e829c16bb44bc5445782686b4c3508794/eval/eval_utils.py#L

Check if the given string a number. https://github.com/MMMU-Benchmark/MMMU/blob/51ce7f3e829c16bb44bc5445782686b4c3508794/eval/eval_utils.py#L

Tensor extension to torch.clamp https://github.com/pytorch/pytorch/issues/2793#issuecomment-428784713

Compute the reward for a list of histories.