Configuration Guide#

This section provides a detailed description of the configuration files used in Trinity-RFT.

Overview#

The configuration for Trinity-RFT is defined in a YAML file and organized into multiple sections based on different modules. Here’s an example of a basic configuration file:

project: Trinity-RFT
name: example
mode: both
checkpoint_root_dir: /PATH/TO/CHECKPOINT
continue_from_checkpoint: true

algorithm:
  # Algorithm-related parameters
  ...
model:
  # Model-specific configurations
  ...
cluster:
  # Cluster node and GPU settings
  ...
buffer:
  # Data buffer configurations
  ...
explorer:
  # Explorer-related settings (rollout models, workflow runners)
  ...
trainer:
  # Trainer-specific parameters
  ...
synchronizer:
  # Model weight synchronization settings
  ...
monitor:
  # Monitoring configurations (e.g., WandB, TensorBoard or MLFlow)
  ...
service:
  # Services to use
  ...
data_processor:
  # Preprocessing data settings
  ...
log:
  # Ray actor logging
  ...

Each of these sections will be explained in detail below. For additional details about specific parameters not covered here, please refer to the source code.

Tip

Trinity-RFT uses OmegaConf to load YAML configuration files. It supports some advanced features like variable interpolation and environment variable substitution. Users can use these features to simplify configuration.

Global Configuration#

These are general settings that apply to the entire experiment.

project: Trinity-RFT
name: example
mode: both
checkpoint_root_dir: ${oc.env:CHECKPOINT_ROOT_DIR}   # CHECKPOINT_ROOT_DIR is an environment variable set in advance

project: The name of the project.
name: The name of the current experiment.
mode: Running mode of Trinity-RFT. Options include:
- both: Launches both the trainer and explorer (default).
- train: Only launches the trainer.
- explore: Only launches the explorer.
- bench: Used for benchmarking.
checkpoint_root_dir: Root directory where all checkpoints and logs will be saved. Checkpoints for this experiment will be stored in <checkpoint_root_dir>/<project>/<name>/.
continue_from_checkpoint: If set to true, the experiment will continue from the latest checkpoint in the checkpoint path (if any); otherwise, it will rename the current experiment to <name>_<timestamp> and start a new experiment.
ray_namespace: Namespace for the modules launched in the current experiment. If not specified, it will be set to <project>/<name>.

Algorithm Configuration#

Specifies the algorithm type and its related hyperparameters.

algorithm:
  algorithm_type: grpo
  repeat_times: 8

  # The following parameters are optional
  # If not specified, they will automatically be set based on the `algorithm_type`
  sample_strategy: "default"
  advantage_fn: "ppo"
  kl_penalty_fn: "none"
  kl_loss_fn: "k2"
  entropy_loss_fn: "default"

algorithm_type: Type of reinforcement learning algorithm. Supported types: ppo, grpo, opmd, dpo, sft, mix.
repeat_times: Number of times each task is repeated. Default is 1. In dpo, this is automatically set to 2. Some algorithms such as GRPO and OPMD require repeat_times > 1.
sample_strategy: The sampling strategy used for loading experiences from experience buffer.
advantage_fn: The advantage function used for computing advantages.
kl_penalty_fn: The KL penalty function used for computing KL penalty applied in reward.
kl_loss_fn: The KL loss function used for computing KL loss.
entropy_loss_fn: The entropy loss function used for computing entropy loss.

Monitor Configuration#

Used to log training metrics during execution.

monitor:
  monitor_type: wandb
  monitor_args:
    base_url: http://localhost:8080
    api_key: your_api_key
  enable_ray_timeline: False

monitor_type: Type of monitoring system. Options:
- wandb: Logs to Weights & Biases. Requires logging in and setting WANDB_API_KEY. Project and run names match the project and name fields in global configs.
- tensorboard: Logs to TensorBoard. Files are saved under <checkpoint_root_dir>/<project>/<name>/monitor/tensorboard.
- mlflow: Logs to MLFlow. If MLFlow authentication is setup, set MLFLOW_TRACKING_USERNAME and MLFLOW_TRACKING_PASSWORD as environment variables before running.
monitor_args: Dictionary of arguments for monitor initialization.
- For wandb:
  - base_url: Overrides WANDB_BASE_URL if set.
  - api_key: Overrides WANDB_API_KEY if set.
- For mlflow:
  - uri: The URI of your MLFlow instance. Strongly recommended to set; defaults to http://localhost:5000.
  - username: Overrides MLFLOW_TRACKING_USERNAME if set.
  - password: Overrides MLFLOW_TRACKING_PASSWORD if set.
enable_ray_timeline: If True, exports a timeline.json file to <checkpoint_root_dir>/<project>/<name>/monitor. Viewable in Chrome at chrome://tracing.

Model Configuration#

Defines the model paths and token limits.

model:
  model_path: ${oc.env:MODEL_PATH}  # MODEL_PATH is an environment variable set in advance
  critic_model_path: ${model.model_path}  # use the value of model.model_path
  max_response_tokens: 16384
  max_model_len: 20480

model_path: Path to the model being trained.
critic_model_path: Optional path to a separate critic model. If empty, defaults to model_path.
max_response_tokens: Maximum number of tokens allowed in generated responses.
max_model_len: Maximum number of tokens in a sequence.

Cluster Configuration#

Defines how many nodes and GPUs per node are used.

cluster:
  node_num: 1
  gpu_per_node: 8

node_num: Total number of compute nodes.
gpu_per_node: Number of GPUs available per node.

Buffer Configuration#

Configures the data buffers used by the explorer and trainer.

buffer:
  batch_size: 32
  train_batch_size: 256
  total_epochs: 100

  explorer_input:
    taskset:
      ...
    eval_tasksets:
      ...
  trainer_input:
    experience_buffer:
      ...
    sft_warmup_dataset:
      ...

  default_workflow_type: 'math_workflow'
  default_reward_fn_type: 'countdown_reward'

batch_size: Number of tasks used per training step. Please do not multiply this value by the algorithm.repeat_times manually.
train_batch_size: Number of experiences used per training step. Defaults to batch_size * algorithm.repeat_times.
total_epochs: Total number of training epochs.
total_steps: Optional. The total number of training steps. If specified, total_epochs will be ignored.

Explorer Input#

Defines the dataset(s) used by the explorer for training and evaluation.

buffer:
  explorer_input:
    taskset:
      name: countdown_train
      storage_type: file
      path: /PATH/TO/DATA
      split: train
      format:
        prompt_key: 'question'
        response_key: 'answer'
      rollout_args:
        temperature: 1.0
      default_workflow_type: 'math_workflow'
      default_reward_fn_type: 'countdown_reward'

    eval_tasksets:
    - name: countdown_eval
      storage_type: file
      path: /PATH/TO/DATA
      split: test
      repeat_times: 1
      format:
        prompt_type: `plaintext`
        prompt_key: 'question'
        response_key: 'answer'
      rollout_args:
        temperature: 0.1
      default_workflow_type: 'math_workflow'
      default_reward_fn_type: 'countdown_reward'
    ...

buffer.explorer_input.taskset: Task dataset used for training exploration policies.
buffer.explorer_input.eval_taskset: List of task datasets used for evaluation.

The configuration for each task dataset is defined as follows:

name: Name of the dataset. This name will be used as the Ray actor’s name, so it must be unique.
storage_type: How the dataset is stored. Options: file, queue, sql.
- file: The dataset is stored in jsonl/parquet files. The data file organization is required to meet the huggingface standard. We recommand using this storage type for most cases.
- queue: The dataset is stored in a queue. The queue is a simple FIFO queue that stores the task dataset. Do not use this storage type for task dataset unless you know what you are doing.
- sql: The dataset is stored in a SQL database. This type is unstable and will be optimized in the future versions.
path: The path to the task dataset.
- For file storage type, the path points to the directory that contains the task dataset files.
- For queue storage type, the path is optional. You can back up the data in the queue by specifying a sqlite database path here.
- For sql storage type, the path points to the sqlite database file.
subset_name: The subset name of the task dataset. Default is None.
split: The split of the task dataset. Default is train.
repeat_times: The number of rollouts generated for a task. If not set, it will be automatically set to algorithm.repeat_times for taskset, and 1 for eval_tasksets.
rollout_args: The parameters for rollout.
- temperature: The temperature for sampling.
default_workflow_type: Type of workflow logic applied to this dataset. If not specified, the buffer.default_workflow_type is used.
default_reward_fn_type: Reward function used during exploration. If not specified, the buffer.default_reward_fn_type is used.
workflow_args: A dictionary of arguments used to supplement dataset-level parameters.

Trainer Input#

Defines the experience buffer and optional SFT warm-up dataset.

buffer:
  ...
  trainer_input:
    experience_buffer:
      name: countdown_buffer
      storage_type: queue
      path: sqlite:///countdown_buffer.db
      max_read_timeout: 1800

    sft_warmup_dataset:
      name: warmup_data
      storage_type: file
      path: /PATH/TO/WARMUP_DATA
      format:
        prompt_key: 'question'
        response_key: 'answer'

    sft_warmup_steps: 0

experience_buffer: It is the input of Trainer and also the output of Explorer. This field is required even in explore mode.
- name: The name of the experience buffer. This name will be used as the Ray actor’s name, so it must be unique.
- storage_type: The storage type for the experience buffer.
  - queue: Experience data is stored in a queue. This storage type is recommended for most use cases.
  - sql: Experience data is stored in a SQL database.
  - file: Experience data is stored in a JSON file. This storage type should be used only for debugging purposes in explore mode.
- path: The path to the experience buffer.
  - For queue storage type, this field is optional. You can specify a SQLite database or JSON file path here to back up the queue data.
  - For file storage type, the path points to the directory containing the dataset files.
  - For sql storage type, the path points to the SQLite database file.
- format: Defines keys for prompts and responses in the dataset.
  - prompt_type: Specifies the type of prompts in the dataset. We support plaintext, messages for now.
    - plaintext: The prompt is in string format.
    - messages: The prompt is organized as a message list.
  - prompt_key: Specifies which column in the dataset contains the user prompt data. Only for plaintext.
  - response_key: Specifies which column in the dataset contains the response data. Only for plaintext.
  - system_prompt_key: Specifies which column in the dataset contains the system prompt data. Only for plaintext.
  - system_prompt: Specifies the system prompt in string format. It has lower priority than system_prompt_key. Only for plaintext.
  - messages_key: Specifies which column in the dataset contains the messages data. Only for messages.
  - tools_key: Specifies which column in the dataset contains the tools data. Support both plaintext and messages, but the tool data should be organized as a list of dict.
  - chosen_key: Specifies which column in the dataset contains the DPO chosen data. Support both plaintext and messages, and the data type should be consistent with the prompt type.
  - rejected_key: Similar to chosen_key, but it specifies which column in the dataset contains the DPO rejected data.
  - enable_concatenated_multi_turn: Enable concatenated multi-turn SFT data preprocess. Only for messages and only take effect with SFT algorithm.
  - chat_template: Specifies the chat template in string format. If not provided, use model.custom_chat_template.
- max_read_timeout: The maximum waiting time (in seconds) to read new experience data. If exceeded, an incomplete batch will be returned directly. Only take effect when storage_type is queue. Default is 1800 seconds (30 minutes).
- use_priority_queue: Only take effect when storage_type is queue. If set to True, the queue will be a priority queue, which allows for prioritizing certain experiences over others. Default is False.
- reuse_cooldown_time: Only take effect when storage_type is queue and use_priority_queue is True. If set, it specifies the cooldown time (in seconds) for reusing experiences. If not specified, the default value is None, meaning experiences can not be reused.
sft_warmup_dataset: Optional dataset used for pre-training (SFT warmup). Its configuration is similar to the experience_buffer, but only for SFT usage.
sft_warmup_steps: Number of steps to use SFT warm-up before RL begins.

Explorer Configuration#

Controls the rollout models and workflow execution.

explorer:
  name: explorer
  runner_per_model: 8
  max_timeout: 900
  max_retry_times: 2
  env_vars: {}
  rollout_model:
    engine_type: vllm
    engine_num: 1
    tensor_parallel_size: 1
    enable_history: False
  auxiliary_models:
  - model_path: /PATH/TO/MODEL
    tensor_parallel_size: 1
  eval_interval: 100
  eval_on_startup: True

name: Name of the explorer. This name will be used as the Ray actor’s name, so it must be unique.
runner_per_model: Number of parallel workflow runners per each rollout model.
max_timeout: Maximum time (in seconds) for a workflow to complete.
max_retry_times: Maximum number of retries for a workflow.
env_vars: Environment variables to be set for every workflow runners.
rollout_model.engine_type: Type of inference engine. For now, only vllm_async and vllm is supported, they have the same meaning and both use the asynchronous engine. In subsequent versions, only vllm may be retained for simplicity.
rollout_model.engine_num: Number of inference engines.
rollout_model.tensor_parallel_size: Degree of tensor parallelism.
rollout_model.enable_history: Whether to enable model call history recording. If set to True, the model wrapper automatically records the return experiences of model calls. Please periodically extract the history via extract_experience_from_history to avoid out-of-memory issues. Default is False.
auxiliary_models: Additional models used for custom workflows.
eval_interval: Interval (in steps) for evaluating the model.
eval_on_startup: Whether to evaluate the model on startup. More precisely, at step 0 with the original model, so it will not be triggered when restarting.
runner_num: (Deprecated) Number of parallel workflow runners.

Synchronizer Configuration#

Controls how model weights are synchronized between trainer and explorer.

synchronizer:
  sync_method: 'nccl'
  sync_interval: 10
  sync_offset: 0
  sync_timeout: 1200

sync_method: Method of synchronization. Options:
- nccl: Uses NCCL for fast synchronization. Supported for both mode.
- checkpoint: Loads latest model from disk. Supported for train, explore, or bench mode.
sync_interval: Interval (in steps) of model weight synchronization between trainer and explorer.
sync_offset: Offset (in steps) of model weight synchronization between trainer and explorer. The explorer can run sync_offset steps before the trainer starts training.
sync_timeout: Timeout duration for synchronization.

Trainer Configuration#

Specifies the backend and behavior of the trainer.

trainer:
  name: trainer
  trainer_type: 'verl'
  save_interval: 100
  trainer_config_path: ''
  trainer_config: null

name: Name of the trainer. This name will be used as the Ray actor’s name, so it must be unique.
trainer_type: Trainer backend implementation. Currently only supports verl.
save_interval: Frequency (in steps) at which to save model checkpoints.
trainer_config_path: The path to the trainer configuration file.
trainer_config: The trainer configuration provided inline. Only one of trainer_config_path and trainer_config should be specified.

Service Configuration#

Configures services used by Trinity-RFT. Only support Data Juicer service for now.

service:
  data_juicer:
    server_url: 'http://127.0.0.1:5005'
    auto_start: true
    port: 5005

server_url: The url of data juicer server.
auto_start: Whether to automatically start the data juicer service.
port: The port for Data Juicer service when auto_start is true.

DataProcessor Configuration#

Configures the task / experience pipeline, please refer to Data Processing section for details.

data_processor:
  task_pipeline:
  # task pipeline related
  task_pipeline:
    num_process: 32
    operators:
      - name: "llm_difficulty_score_filter"
        args:
          api_or_hf_model: "qwen2.5-7b-instruct"
          min_score: 0.0
          input_keys: ["question", "answer"]
          field_names: ["Question", "Answer"]
    inputs:  # the output will be set to the explorer input automatically
      - /PATH/TO/GSM8K/DATA/FILE
    target_fields: ["question", "answer"]
  experience_pipeline:
    operators:
      - name: data_juicer
        args:
          config_path: 'examples/grpo_gsm8k_experience_pipeline/dj_scoring_exp.yaml'
      - name: reward_shaping_mapper
        args:
          reward_shaping_configs:
            - stats_key: 'llm_quality_score'
              op_type: ADD
              weight: 1.0

–

Log Configuration#

Ray actor logging configuration.

log:
  level: INFO
  group_by_node: False

level: The logging level (supports DEBUG, INFO, WARNING, ERROR).
group_by_node: Whether to group logs by node IP. If set to True, an actor’s logs will be save to <checkpoint_root_dir>/<project>/<name>/log/<node_ip>/<actor_name>.log, otherwise it will be saved to <checkpoint_root_dir>/<project>/<name>/log/<actor_name>.log.

veRL Trainer Configuration (Advanced)#

For advanced users working with the verl trainer backend. This includes fine-grained settings for actor/critic models, optimizer parameters, and training loops.

For full parameter meanings, refer to the veRL documentation.

actor_rollout_ref:
  hybrid_engine: True
  model:
    external_lib: null
    override_config: { }
    enable_gradient_checkpointing: True
    use_remove_padding: True
    use_fused_kernels: False
    fused_kernel_options:
      impl_backend: None
  actor:
    strategy: fsdp  # This is for backward-compatibility
    # ppo_micro_batch_size: 8 # will be deprecated, use ppo_micro_batch_size_per_gpu
    ppo_micro_batch_size_per_gpu: 4
    use_dynamic_bsz: True
    ppo_max_token_len_per_gpu: 16384 # n * ${data.max_model_len}
    grad_clip: 1.0
    ppo_epochs: 1
    shuffle: False
    ulysses_sequence_parallel_size: 1 # sp size
    entropy_from_logits_with_chunking: false
    entropy_checkpointing: false
    checkpoint:
      load_contents: ['model', 'optimizer', 'extra']
      save_contents: ['model', 'optimizer', 'extra']
    optim:
      lr: 1e-6
      lr_warmup_steps_ratio: 0.  # the total steps will be injected during runtime
      # min_lr_ratio: null   # only useful for warmup with cosine
      warmup_style: constant  # select from constant/cosine
      total_training_steps: -1  # must be override by program
    fsdp_config:
      wrap_policy:
        # transformer_layer_cls_to_wrap: None
        min_num_params: 0
      param_offload: False
      optimizer_offload: False
      fsdp_size: -1
      forward_prefetch: False
  ref:
    fsdp_config:
      param_offload: False
      wrap_policy:
        # transformer_layer_cls_to_wrap: None
        min_num_params: 0
      fsdp_size: -1
      forward_prefetch: False
    # log_prob_micro_batch_size: 4 # will be deprecated, use log_prob_micro_batch_size_per_gpu
    log_prob_micro_batch_size_per_gpu: 4
    log_prob_use_dynamic_bsz: ${actor_rollout_ref.actor.use_dynamic_bsz}
    log_prob_max_token_len_per_gpu: ${actor_rollout_ref.actor.ppo_max_token_len_per_gpu}
    ulysses_sequence_parallel_size: ${actor_rollout_ref.actor.ulysses_sequence_parallel_size} # sp size
    entropy_from_logits_with_chunking: ${actor_rollout_ref.actor.entropy_from_logits_with_chunking}
    entropy_checkpointing: ${actor_rollout_ref.actor.entropy_checkpointing}

critic:
  strategy: fsdp
  optim:
    lr: 1e-5
    lr_warmup_steps_ratio: 0.  # the total steps will be injected during runtime
    # min_lr_ratio: null   # only useful for warmup with cosine
    warmup_style: constant  # select from constant/cosine
    total_training_steps: -1  # must be override by program
  model:
    override_config: { }
    external_lib: ${actor_rollout_ref.model.external_lib}
    enable_gradient_checkpointing: True
    use_remove_padding: False
    fsdp_config:
      param_offload: False
      optimizer_offload: False
      wrap_policy:
        # transformer_layer_cls_to_wrap: None
        min_num_params: 0
      fsdp_size: -1
      forward_prefetch: False
  ppo_micro_batch_size_per_gpu: 8
  forward_micro_batch_size_per_gpu: ${critic.ppo_micro_batch_size_per_gpu}
  use_dynamic_bsz: ${actor_rollout_ref.actor.use_dynamic_bsz}
  ppo_max_token_len_per_gpu: 32768 # (${actor_rollout_ref.actor.ppo_max_token_len_per_gpu}) * 2
  forward_max_token_len_per_gpu: ${critic.ppo_max_token_len_per_gpu}
  ulysses_sequence_parallel_size: 1 # sp size
  ppo_epochs: ${actor_rollout_ref.actor.ppo_epochs}
  shuffle: ${actor_rollout_ref.actor.shuffle}
  grad_clip: 1.0
  cliprange_value: 0.5

trainer:
  balance_batch: True
  # total_training_steps: null
  # auto: find the last ckpt to resume. If can't find, start from scratch
  resume_mode: auto # or auto or resume_path if
  resume_from_path: ""
  critic_warmup: 0
  default_hdfs_dir: null
  remove_previous_ckpt_in_save: False
  del_local_ckpt_after_load: False
  val_before_train: False
  max_actor_ckpt_to_keep: 5
  max_critic_ckpt_to_keep: 5

actor_rollout_ref.model.enable_gradient_checkpointing: Whether to enable gradient checkpointing, which will reduce GPU memory usage.
actor_rollout_ref.model.use_remove_padding: Whether to remove pad tokens, which will reduce training time.
actor_rollout_ref.model.use_fused_kernels: Whether to use custom fused kernels (e.g., FlashAttention, fused MLP).
actor_rollout_ref.model.fused_kernel_options.impl_backend: Implementation backend for fused kernels. If use_fused_kernels is true, this will be used. Options: “triton” or “torch”.
actor_rollout_ref.actor.use_dynamic_bsz: Whether to reorganize the batch data, specifically to splice the shorter data to reduce the batch size in the actual training process.
actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu: Batch size for one GPU in one forward pass.
actor_rollout_ref.actor.ulysses_sequence_parallel_size: Ulysses sequence parallel size.
actor_rollout_ref.actor.entropy_from_logits_with_chunking: Calculate entropy with chunking to reduce memory peak.
actor_rollout_ref.actor.entropy_checkpointing: Recompute entropy.
actor_rollout_ref.actor.checkpoint: Contents to be loaded and saved. With ‘hf_model’ you can save whole model as hf format; now only use sharded model checkpoint to save space.
actor_rollout_ref.actor.optim.lr: Learning rate for actor model.
actor_rollout_ref.actor.optim.lr_warmup_steps_ratio: Ratio of warmup steps for learning rate.
actor_rollout_ref.actor.optim.warmup_style: Warmup style for learning rate.
actor_rollout_ref.actor.optim.total_training_steps: Total training steps for actor model.
actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu: Batch size for one GPU in one reference model forward pass.
critic.model.enable_gradient_checkpointing: Whether to enable gradient checkpointing, which will reduce GPU memory usage.
critic.model.use_remove_padding: Whether to remove pad tokens, which will reduce training time.
critic.optim.lr: Learning rate for critic model.
critic.optim.lr_warmup_steps_ratio: Ratio of warmup steps for learning rate.
critic.optim.warmup_style: Warmup style for learning rate.
critic.optim.total_training_steps: Total training steps for critic model.
critic.ppo_micro_batch_size_per_gpu: Batch size for one GPU in one critic model forward pass.
critic.ulysses_sequence_parallel_size: Ulysses sequence parallel size.
critic.grad_clip: Gradient clip for critic model training.
critic.cliprange_value: Used for compute value loss.
trainer.balance_batch: Whether to balance batch size between GPUs during training.
trainer.resume_mode: Resume mode for training. Support disable, auto and resume_path.
trainer.resume_from_path: Path to resume from.
trainer.critic_warmup: The number of steps to train the critic model before actual policy learning.
trainer.default_hdfs_dir: Default HDFS directory for saving checkpoints.
trainer.remove_previous_ckpt_in_save: Whether to remove previous checkpoints in save.
trainer.del_local_ckpt_after_load: Whether to delete local checkpoints after loading.
trainer.max_actor_ckpt_to_keep: Maximum number of actor checkpoints to keep.
trainer.max_critic_ckpt_to_keep: Maximum number of critic checkpoints to keep.