添加转换函数 Args: transform_func: 转换函数，接收一个字典，返回一个字典或字典列表 - 返回字典表示一对一转换

Convert arbitrary group labels to a contiguous 1-D torch.long tensor (0..G-1). Args: index: Any iterable of labels or tensor/ndarray

Build an optimizer based on the configuration. Dynamically imports and instantiates an optimizer class from the specified module. Args:

Check and disable incompatible configurations for older Megatron version. Args: original_config (dict): The original model configura

Compute off-policy diagnostic metrics (helper function). This helper function operates on raw tensors and is used internally by: - compute_ro

round: 0 - 初始状态 1 - 发玩家1的私牌 2 - 发玩家2的私牌 3 - 小盲 / 大盲 4 - 小盲出牌 5 - 大盲出牌 ......

Args: tensor: a tensor with shape (bsz,) nested_tensor: a nested tensor with shape (bsz, xxx) Returns: a tensor wit

Fallback implementation for older transformers versions. Returns False to disable features that require this function.

Loads the full state dict (could be only on rank 0) into the sharded model. This is done by broadcasting the parameters from rank 0 to all ot

Get the advantage estimator function with a given name. Args: name_or_enum: `(str)` or `(AdvantageEstimator)` The name or enu

Get all registered dispatch and collect dp_ranks. Returns: dict[str, int]: A dictionary mapping mesh names to the

获取外网可访问的IP地址

Build the model.

Get the number of gpus in this cluster.

Get a generator that yields per-tensor parameters and optional peft config. Returns: Generator[tuple[str, torch.Tensor]]

Convert huggingface PretrainedConfig to mcore TransformerConfig. Args: hf_config: The huggingface PretrainedConfig. dtype: The dt

Perform inference on a batch of data. Args: data: The input data for inference, typically containing tensors and metadat

Launch an SGLang HTTP server process and wait for it to be ready. This function starts a new process running an SGLang HTTP server, then waits

从字符串路径动态加载类 Args: class_path: 类的完整路径，如 'internbootcamp.bootcamps.example_bootcamp.example_evaluator.ExampleEvaluator'

Given a config, do we need critic

解析单个样本结果 - 支持多种格式

Process validation metrics into a structured format with statistical analysis. This function organizes validation metrics by data source and

读取JSONL文件（支持增量读取）

Reset the model engine to the initial state. If the engine is not initialized, we initialize it. Otherwise, reload ckpt and reset sta

Save checkpoint using FSDPCheckpointManager with improved tracking

Set the dispatch and collect dp_ranks for all registered meshes. Args: mesh_name (str): Mesh name to set dispatch and collect dp_

Stop profiling on all rollout replicas.

Load a simple tokenizer for testing.

Perform a training step on a batch of data. Args: data: The input data for training, typically containing tensors and me

Update the module config with the override_config_kwargs. Args: module_config: The module config from Huggingface Transformers. ov

Wait for the broadcast operation to complete. Returns: dict[str, TensorMeta]: The bucket meta after broadcast.

Block until remote agent read complete.

Wait for the broadcast operation to complete. Returns: dict[str, TensorMeta]: The bucket meta after broadcast.

broadcast tensor in tp shards across mp_group

broadcast tensor in tp shards across mp_group

broadcast tensor in tp shards across mp_group

Build multimodal embeddings and attention mask

Compute loss with optional sequence parallelism and remove padding features

Dump rollout/validation samples as JSONL.

Extract reward from already-computed rm_scores if available. This is used when use_reward_loop=True and rewards are already computed

Dynamically import attention functions based on available hardware.

Log validation data