Source code for verl.utils.tracking

# Copyright 2024 Bytedance Ltd. and/or its affiliates
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
A unified tracking interface that supports logging data to different backend
"""

import dataclasses
import json
import logging
import os
from contextlib import contextmanager
from enum import Enum
from functools import partial
from pathlib import Path
from typing import Any

import orjson
from packaging.version import Version

logger = logging.getLogger(__name__)

MLFLOW_MAX_ATTEMPTS = 3
MLFLOW_SLEEP_SECONDS = 5


[docs] class Tracking: """A unified tracking interface for logging experiment data to multiple backends. This class provides a centralized way to log experiment metrics, parameters, and artifacts to various tracking backends including WandB, MLflow, SwanLab, TensorBoard, and console. Attributes: supported_backend: List of supported tracking backends. logger: Dictionary of initialized logger instances for each backend. """ supported_backend = [ "wandb", "mlflow", "swanlab", "vemlp_wandb", "tensorboard", "console", "clearml", "trackio", "file", "rl_insight", ] def __init__(self, project_name, experiment_name, default_backend: str | list[str] = "console", config=None): if isinstance(default_backend, str): default_backend = [default_backend] for backend in default_backend: if backend == "tracking": import warnings warnings.warn("`tracking` logger is deprecated. use `wandb` instead.", DeprecationWarning, stacklevel=2) else: assert backend in self.supported_backend, f"{backend} is not supported" self.logger = {} self._finished = False if "tracking" in default_backend or "wandb" in default_backend: import os import wandb settings = None if config and config["trainer"].get("wandb_proxy", None): settings = wandb.Settings(https_proxy=config["trainer"]["wandb_proxy"]) entity = os.environ.get("WANDB_ENTITY", None) wandb.init(project=project_name, name=experiment_name, entity=entity, config=config, settings=settings) self.logger["wandb"] = wandb if "trackio" in default_backend: import trackio from trackio import context_vars if context_vars.current_run.get() is None: trackio.init(project=project_name, name=experiment_name, config=config) self.logger["trackio"] = _TrackioLoggingAdapter(trackio) if "mlflow" in default_backend: import os import time import mlflow for _mlflow_attempt in range(1, MLFLOW_MAX_ATTEMPTS + 1): try: MLFLOW_TRACKING_URI = os.environ.get("MLFLOW_TRACKING_URI", "sqlite:////tmp/mlruns.db") logger.info("Using MLFlow tracking URI: %s", MLFLOW_TRACKING_URI) mlflow.set_tracking_uri(MLFLOW_TRACKING_URI) # Some cloud providers like Azure ML or Databricks automatically set MLFLOW_RUN_ID # If set, attach to the existing run instead of creating a new one run_id = os.environ.get("MLFLOW_RUN_ID") if run_id: mlflow.start_run(run_id=run_id) else: # Project_name is actually experiment_name in MLFlow # If experiment does not exist, will create a new experiment experiment = mlflow.set_experiment(project_name) mlflow.start_run(experiment_id=experiment.experiment_id, run_name=experiment_name) mlflow.log_params(_compute_mlflow_params_from_objects(config)) self.logger["mlflow"] = _MlflowLoggingAdapter() break # Success except Exception as e: logger.warning( "MLflow initialization attempt %d/%d failed: %s", _mlflow_attempt, MLFLOW_MAX_ATTEMPTS, e ) if _mlflow_attempt < MLFLOW_MAX_ATTEMPTS: time.sleep(MLFLOW_SLEEP_SECONDS) else: logger.warning("All MLflow initialization attempts failed. Proceeding without MLflow tracking.") if "swanlab" in default_backend: import os import swanlab SWANLAB_API_KEY = os.environ.get("SWANLAB_API_KEY", None) SWANLAB_LOG_DIR = os.environ.get("SWANLAB_LOG_DIR", "swanlog") SWANLAB_MODE = os.environ.get("SWANLAB_MODE", "cloud") if SWANLAB_API_KEY: swanlab.login(SWANLAB_API_KEY) # NOTE: previous login information will be overwritten if config is None: config = {} # make sure config is not None, otherwise **config will raise error swanlab.init( project=project_name, experiment_name=experiment_name, config={"FRAMEWORK": "verl", **config}, logdir=SWANLAB_LOG_DIR, mode=SWANLAB_MODE, ) self.logger["swanlab"] = swanlab if "vemlp_wandb" in default_backend: import os import volcengine_ml_platform from volcengine_ml_platform import wandb as vemlp_wandb volcengine_ml_platform.init( ak=os.environ["VOLC_ACCESS_KEY_ID"], sk=os.environ["VOLC_SECRET_ACCESS_KEY"], region=os.environ["MLP_TRACKING_REGION"], ) vemlp_wandb.init( project=project_name, name=experiment_name, config=config, sync_tensorboard=True, ) self.logger["vemlp_wandb"] = vemlp_wandb if "tensorboard" in default_backend: self.logger["tensorboard"] = _TensorboardAdapter(project_name, experiment_name) if "console" in default_backend: from verl.utils.logger import LocalLogger self.console_logger = LocalLogger(print_to_console=True) self.logger["console"] = self.console_logger if "clearml" in default_backend: self.logger["clearml"] = ClearMLLogger(project_name, experiment_name, config) if "file" in default_backend: self.logger["file"] = FileLogger(project_name, experiment_name) if "rl_insight" in default_backend: self.logger["rl_insight"] = RLInsightLogger(project_name, experiment_name, config) def log(self, data, step, backend=None): for default_backend, logger_instance in self.logger.items(): if backend is None or default_backend in backend: logger_instance.log(data=data, step=step)
[docs] def finish(self, exit_code: int = 0): """Flush and finalize every configured backend exactly once.""" if getattr(self, "_finished", False): return self._finished = True loggers = getattr(self, "logger", {}) if "wandb" in loggers: loggers["wandb"].finish(exit_code=exit_code) if "swanlab" in loggers: loggers["swanlab"].finish() if "vemlp_wandb" in loggers: loggers["vemlp_wandb"].finish(exit_code=exit_code) if "tensorboard" in loggers: loggers["tensorboard"].finish() if "clearml" in loggers: loggers["clearml"].finish() if "trackio" in loggers: loggers["trackio"].finish() if "file" in loggers: loggers["file"].finish() if "rl_insight" in loggers: loggers["rl_insight"].finish()
def __del__(self): self.finish()
class RLInsightLogger: """Logger backend that exports scalar metrics and rl-insight runtime signals.""" ENABLE_ENV = "VERL_RL_INSIGHT_ENABLE" _init_done = False _rl_insight_module = None _registered_metrics: set[tuple[str | None, tuple[str, ...], str | None]] = set() def __init__(self, project_name, experiment_name, config=None): self.init(project_name=project_name, experiment_name=experiment_name, config=config) @classmethod def _get_rl_insight(cls): if cls._rl_insight_module is None: import rl_insight cls._rl_insight_module = rl_insight return cls._rl_insight_module @classmethod def enabled(cls) -> bool: """Return whether rl-insight is globally enabled in this process.""" return os.getenv(cls.ENABLE_ENV) == "1" @classmethod def init(cls, project_name=None, experiment_name=None, config=None): if not cls.enabled() or cls._init_done: return rl_insight_config = {} if config is not None: try: rl_insight_config = config.get("trainer", {}).get("rl_insight", {}) or {} except (AttributeError, KeyError, TypeError): pass cls._get_rl_insight().init(project=project_name, experiment_name=experiment_name, config=rl_insight_config) cls._init_done = True if config is not None: cls.register_transfer_queue_metrics(config) @classmethod def log(cls, data, step): if not cls.enabled(): return if not cls._init_done: cls._get_rl_insight().init() cls._init_done = True metric_gauge = cls._get_rl_insight().metric_gauge for key, value in data.items(): try: scalar = float(value) except (TypeError, ValueError): continue metric_gauge(str(key).replace("/", "_"), scalar) @classmethod def finish(cls): if not cls._init_done: return cls._get_rl_insight().finish() cls._init_done = False cls._registered_metrics.clear() @classmethod @contextmanager def trace_state( cls, state_name: str, *, state_lane_id: str | int | None = None, **labels: Any, ): if not cls.enabled(): yield return if not cls._init_done: cls._get_rl_insight().init() cls._init_done = True with cls._get_rl_insight().trace_state(state_name, state_lane_id=state_lane_id, **labels): yield @classmethod def register_rollout_metrics( cls, server_addresses: list[str], rollout_name: str | None, labels: list[dict[str, Any] | None] | None = None, ) -> None: cls.register_metrics(server_addresses, rollout_name, labels) @classmethod def register_transfer_queue_metrics(cls, config) -> None: if not (config or {}).get("transfer_queue", {}).get("metrics", {}).get("enabled", False): return try: import transfer_queue as tq endpoint = tq.get_metrics_endpoint() except Exception: logger.exception("[rl-insight] Failed to get transfer_queue metrics endpoint") return if endpoint: cls.register_metrics([endpoint], "transfer_queue", None) @classmethod def register_metrics( cls, server_addresses: list[str], job_name: str | None = None, labels: list[dict[str, Any] | None] | None = None, ) -> None: if not cls.enabled(): return metric_key = (job_name, tuple(server_addresses), repr(labels)) if metric_key in cls._registered_metrics: return try: cls._get_rl_insight().update_prometheus_config(server_addresses, job_name, labels) cls._registered_metrics.add(metric_key) except Exception: logger.exception("[rl-insight] Failed to register metrics endpoint") class ClearMLLogger: def __init__(self, project_name: str, experiment_name: str, config): self.project_name = project_name self.experiment_name = experiment_name import clearml self._task: clearml.Task = clearml.Task.init( task_name=experiment_name, project_name=project_name, continue_last_task=True, output_uri=False, ) self._task.connect_configuration(config, name="Hyperparameters") def _get_logger(self): return self._task.get_logger() def log(self, data, step): import numpy as np import pandas as pd # logs = self._rewrite_logs(data) logger = self._get_logger() for k, v in data.items(): title, series = k.split("/", 1) if isinstance(v, int | float | np.floating | np.integer): logger.report_scalar( title=title, series=series, value=v, iteration=step, ) elif isinstance(v, pd.DataFrame): logger.report_table( title=title, series=series, table_plot=v, iteration=step, ) else: logger.warning( f'Trainer is attempting to log a value of "{v}" of type {type(v)} for key "{k}". This ' f"invocation of ClearML logger's function is incorrect so this attribute was dropped. " ) def finish(self): self._task.close() class _TrackioLoggingAdapter: def __init__(self, trackio): self.trackio = trackio def log(self, data, step): self.trackio.log(data, step=step) def finish(self): from trackio import context_vars if context_vars.current_run.get() is not None: self.trackio.finish() class FileLogger: def __init__(self, project_name: str, experiment_name: str): self.project_name = project_name self.experiment_name = experiment_name self.filepath = os.getenv("VERL_FILE_LOGGER_PATH", None) if self.filepath is None: root_path = os.path.expanduser(os.getenv("VERL_FILE_LOGGER_ROOT", ".")) directory = os.path.join(root_path, self.project_name) os.makedirs(directory, exist_ok=True) self.filepath = os.path.join(directory, f"{self.experiment_name}.jsonl") print(f"Creating file logger at {os.path.abspath(self.filepath)}") self.fp = open(self.filepath, "wb", buffering=0) def log(self, data, step): data = {"step": step, "data": data} self.fp.write(orjson.dumps(data, option=orjson.OPT_SERIALIZE_NUMPY) + b"\n") def finish(self): self.fp.close() class _TensorboardAdapter: def __init__(self, project_name, experiment_name): import os from torch.utils.tensorboard import SummaryWriter tensorboard_dir = os.environ.get("TENSORBOARD_DIR", f"tensorboard_log/{project_name}/{experiment_name}") os.makedirs(tensorboard_dir, exist_ok=True) print(f"Saving tensorboard log to {tensorboard_dir}.") self.writer = SummaryWriter(tensorboard_dir) def log(self, data, step): for key in data: self.writer.add_scalar(key, data[key], step) def finish(self): self.writer.close() class _MlflowLoggingAdapter: def __init__(self): import logging import re self.logger = logging.getLogger(__name__) # Suppress noisy "Found credentials from IAM Role" on every MLflow request logging.getLogger("botocore.credentials").setLevel(logging.WARNING) # MLflow metric key validation logic: # https://github.com/mlflow/mlflow/blob/master/mlflow/utils/validation.py#L157C12-L157C44 # Only characters allowed: slashes, alphanumerics, underscores, periods, dashes, colons, # and spaces. self._invalid_chars_pattern = re.compile( r"[^/\w.\- :]" ) # Allowed: slashes, alphanumerics, underscores, periods, dashes, colons, and spaces. self._consecutive_slashes_pattern = re.compile(r"/+") self._sanitized_key_cache = {} def _sanitize_key(self, key): if key in self._sanitized_key_cache: return self._sanitized_key_cache[key] or key # First replace @ with _at_ for backward compatibility sanitized = key.replace("@", "_at_") # Replace consecutive slashes with a single slash (MLflow treats them as file paths) sanitized = self._consecutive_slashes_pattern.sub("/", sanitized) # Then replace any other invalid characters with _ sanitized = self._invalid_chars_pattern.sub("_", sanitized) if sanitized == key: self._sanitized_key_cache[key] = None else: self.logger.warning("[MLflow] Metric key '%s' sanitized to '%s' due to invalid characters.", key, sanitized) self._sanitized_key_cache[key] = sanitized return sanitized def log(self, data, step): import mlflow results = {self._sanitize_key(k): v for k, v in data.items()} for _attempt in range(MLFLOW_MAX_ATTEMPTS): try: mlflow.log_metrics(metrics=results, step=step) return except Exception as error: # No sleep between retries — this runs per training step, so we avoid blocking. msg = "mlflow.log_metrics failed (attempt %d/%d): %s" args = (_attempt + 1, MLFLOW_MAX_ATTEMPTS, error) if _attempt < MLFLOW_MAX_ATTEMPTS - 1: self.logger.info(msg, *args) else: self.logger.warning(msg, *args) def _compute_mlflow_params_from_objects(params) -> dict[str, Any]: if params is None: return {} return _flatten_dict(_transform_params_to_json_serializable(params, convert_list_to_dict=True), sep="/") def _transform_params_to_json_serializable(x, convert_list_to_dict: bool): _transform = partial(_transform_params_to_json_serializable, convert_list_to_dict=convert_list_to_dict) if dataclasses.is_dataclass(x): return _transform(dataclasses.asdict(x)) if isinstance(x, dict): return {k: _transform(v) for k, v in x.items()} if isinstance(x, list): if convert_list_to_dict: return {"list_len": len(x)} | {f"{i}": _transform(v) for i, v in enumerate(x)} else: return [_transform(v) for v in x] if isinstance(x, Path): return str(x) if isinstance(x, Enum): return x.value return x def _flatten_dict(raw: dict[str, Any], *, sep: str) -> dict[str, Any]: import pandas as pd ans = pd.json_normalize(raw, sep=sep).to_dict(orient="records")[0] assert isinstance(ans, dict) return ans @dataclasses.dataclass class ValidationGenerationsLogger: project_name: str = None experiment_name: str = None def log(self, loggers, samples, step): if "wandb" in loggers: self.log_generations_to_wandb(samples, step) if "swanlab" in loggers: self.log_generations_to_swanlab(samples, step) if "mlflow" in loggers: self.log_generations_to_mlflow(samples, step) if "trackio" in loggers: self.log_generations_to_trackio(samples, step) if "clearml" in loggers: self.log_generations_to_clearml(samples, step) if "tensorboard" in loggers: self.log_generations_to_tensorboard(samples, step) if "vemlp_wandb" in loggers: self.log_generations_to_vemlp_wandb(samples, step) def log_generations_to_vemlp_wandb(self, samples, step): from volcengine_ml_platform import wandb as vemlp_wandb self._log_generations_to_wandb(samples, step, vemlp_wandb) def log_generations_to_wandb(self, samples, step): import wandb self._log_generations_to_wandb(samples, step, wandb) def _log_generations_to_wandb(self, samples, step, wandb): """Log samples to wandb as a table""" # Create column names for all samples columns = ["step"] + sum( [[f"input_{i + 1}", f"output_{i + 1}", f"score_{i + 1}"] for i in range(len(samples))], [] ) if not hasattr(self, "validation_table"): # Initialize the table on first call self.validation_table = wandb.Table(columns=columns) # Create a new table with same columns and existing data # Workaround for https://github.com/wandb/wandb/issues/2981#issuecomment-1997445737 new_table = wandb.Table(columns=columns, data=self.validation_table.data) # Add new row with all data row_data = [] row_data.append(step) for sample in samples: row_data.extend(sample) new_table.add_data(*row_data) # Update reference and log if wandb.run is not None: wandb.log({"val/generations": new_table}, step=step) self.validation_table = new_table def log_generations_to_swanlab(self, samples, step): """Log samples to swanlab as text""" import swanlab swanlab_table = swanlab.echarts.Table() # Create column names headers = ["step", "input", "output", "score"] swanlab_row_list = [[step, *sample] for sample in samples] swanlab_table.add(headers=headers, rows=swanlab_row_list) # Log to swanlab swanlab.log({"val/generations": swanlab_table}, step=step) def log_generations_to_mlflow(self, samples, step): """Log validation generation to mlflow as artifacts""" # https://mlflow.org/docs/latest/api_reference/python_api/mlflow.html?highlight=log_artifact#mlflow.log_artifact import tempfile import mlflow try: with tempfile.TemporaryDirectory() as tmp_dir: validation_gen_step_file = Path(tmp_dir, f"val_step{step}.json") row_data = [] for sample in samples: data = {"input": sample[0], "output": sample[1], "score": sample[2]} row_data.append(data) with open(validation_gen_step_file, "w") as file: json.dump(row_data, file) mlflow.log_artifact(validation_gen_step_file) except Exception as e: print(f"WARNING: save validation generation file to mlflow failed with error {e}") def log_generations_to_trackio(self, samples, step): """Log validation generations to trackio as traces.""" import trackio traces = [] for sample_index, sample in enumerate(samples): if len(sample) >= 3: input_text, output_text, score = sample[0], sample[1], sample[2] else: input_text, output_text, score = sample, "", None traces.append( trackio.Trace( messages=[ {"role": "user", "content": str(input_text)}, {"role": "assistant", "content": str(output_text)}, ], metadata={ "source": "validation_generations", "sample_index": sample_index, "score": score, }, ) ) if traces: trackio.log({"val/generations": traces}, step=step) def log_generations_to_clearml(self, samples, step): """Log validation generation to clearml as table""" import clearml import pandas as pd task: clearml.Task | None = clearml.Task.current_task() if task is None: return table = [ { "step": step, "input": sample[0], "output": sample[1], "score": sample[2], } for sample in samples ] logger = task.get_logger() logger.report_table( series="Validation generations", title="Validation", table_plot=pd.DataFrame.from_records(table), iteration=step, ) def log_generations_to_tensorboard(self, samples, step): """Log samples to tensorboard as text""" # Initialize tensorboard writer if not exists if not hasattr(self, "writer"): from torch.utils.tensorboard import SummaryWriter # Use the same directory structure as _TensorboardAdapter if self.project_name and self.experiment_name: default_dir = os.path.join("tensorboard_log", self.project_name, self.experiment_name) else: default_dir = "tensorboard_log" tensorboard_dir = os.environ.get("TENSORBOARD_DIR", default_dir) os.makedirs(tensorboard_dir, exist_ok=True) self.writer = SummaryWriter(log_dir=tensorboard_dir) # Format the samples data into readable text text_content = f"**Generation Results - Step {step}**\n\n" for i, sample in enumerate(samples): text_content += f"### Sample {i + 1}\n" # Assuming sample contains [input, output, score] if len(sample) >= 3: input_text, output_text, score = sample[0], sample[1], sample[2] text_content += f"**Input:** {input_text}\n\n" text_content += f"**Output:** {output_text}\n\n" text_content += f"**Score:** {score}\n\n" else: # Handle cases where sample format might be different text_content += f"**Data:** {sample}\n\n" text_content += "---\n\n" # Log to tensorboard as text self.writer.add_text("val/generations", text_content, step) # Flush to ensure data is written self.writer.flush() @dataclasses.dataclass class DapoFilteredRewardTableLogger: """Wandb table of DAPO-filtered (no-signal) group counts per reward value. Each training step adds one row containing compact ``reward:count`` pairs. Wandb 0.20+ uploads rows incrementally; older versions rebuild the full table for compatibility. Intentionally wandb-only: this "value distribution over time" view is a table, which other tracking backends do not render usefully. Non-wandb backends are silently skipped. """ project_name: str = None experiment_name: str = None def log(self, loggers, reward_counts: dict, step: int): """reward_counts maps metric value -> count for this step (already merged across mini-batches).""" if "wandb" in loggers: self._log_to_wandb(reward_counts, step) def _log_to_wandb(self, reward_counts: dict, step: int): import wandb if wandb.run is None: return row = {float(value): int(count) for value, count in reward_counts.items()} counts_text = ", ".join(f"{value:g}:{row[value]}" for value in sorted(row)) columns = ["step", "reward_counts"] if not hasattr(self, "_use_incremental_table"): self._use_incremental_table = Version(wandb.__version__) >= Version("0.20.0") if self._use_incremental_table: self._table = wandb.Table(columns=columns, log_mode="INCREMENTAL") else: self._rows = [] logger.warning( "wandb<0.20.0 does not support incremental tables; " "the DAPO filtered-reward table will re-upload its full history each step." ) if self._use_incremental_table: self._table.add_data(step, counts_text) table = self._table else: self._rows.append([step, counts_text]) table = wandb.Table(columns=columns, data=list(self._rows)) wandb.log({"training/filter_groups/filtered_reward_counts": table}, step=step)