Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+trackio/assets/trackio_logo_old.png filter=lfs diff=lfs merge=lfs -text

trackio/CHANGELOG.md

@@ -0,0 +1,135 @@
+# trackio
+
+## 0.15.0
+
+### Features
+
+- [#397](https://github.com/gradio-app/trackio/pull/397) [`6b38ad0`](https://github.com/gradio-app/trackio/commit/6b38ad02e5d73a0df49c4eede7e91331282ece04) - Adds `--host` cli option support.  Thanks @abidlabs!
+- [#394](https://github.com/gradio-app/trackio/pull/394) [`c47a3a3`](https://github.com/gradio-app/trackio/commit/c47a3a31f8c4b83bce1aa7fc22eeba3d9021ad3d) - Add wandb-compatible API for trackio.  Thanks @abidlabs!
+- [#378](https://github.com/gradio-app/trackio/pull/378) [`b02046a`](https://github.com/gradio-app/trackio/commit/b02046a5b0dad7c9854e099a87f884afba4aecb2) - Add JSON export button for line plots and upgrade gradio dependency.  Thanks @JamshedAli18!
+
+## 0.14.2
+
+### Features
+
+- [#386](https://github.com/gradio-app/trackio/pull/386) [`f9452cd`](https://github.com/gradio-app/trackio/commit/f9452cdb8f0819368f3610f7ac0ed08957305275) - Fixing some issues related to deployed Trackio Spaces.  Thanks @abidlabs!
+
+## 0.14.1
+
+### Features
+
+- [#382](https://github.com/gradio-app/trackio/pull/382) [`44fe9bb`](https://github.com/gradio-app/trackio/commit/44fe9bb264fb2aafb0ec302ff15227c045819a2c) - Fix app file path when Trackio is not installed from source.  Thanks @abidlabs!
+- [#380](https://github.com/gradio-app/trackio/pull/380) [`c3f4cff`](https://github.com/gradio-app/trackio/commit/c3f4cff74bc5676e812773d8571454894fcdc7cc) - Add CLI commands for querying projects, runs, and metrics.  Thanks @abidlabs!
+
+## 0.14.0
+
+### Features
+
+- [#377](https://github.com/gradio-app/trackio/pull/377) [`5c5015b`](https://github.com/gradio-app/trackio/commit/5c5015b68c85c5de51111dad983f735c27b9a05f) - fixed wrapping issue in Runs table.  Thanks @gaganchapa!
+- [#374](https://github.com/gradio-app/trackio/pull/374) [`388e26b`](https://github.com/gradio-app/trackio/commit/388e26b9e9f24cd7ad203affe9b709be885b3d24) - Save Optimized Parquet files.  Thanks @lhoestq!
+- [#371](https://github.com/gradio-app/trackio/pull/371) [`fbace9c`](https://github.com/gradio-app/trackio/commit/fbace9cd7732c166f34d268f54b05bb06846cc5d) - Add GPU metrics logging.  Thanks @kashif!
+- [#367](https://github.com/gradio-app/trackio/pull/367) [`862840c`](https://github.com/gradio-app/trackio/commit/862840c13e30fc960cbee5b9eac4d3c25beba9de) - Add option to only show latest run, and fix the double logo issue.  Thanks @abidlabs!
+
+## 0.13.1
+
+### Features
+
+- [#369](https://github.com/gradio-app/trackio/pull/369) [`767e9fe`](https://github.com/gradio-app/trackio/commit/767e9fe095d7c6ed102016caf927c1517fb8618c) - tiny pr removing unnecessary code.  Thanks @abidlabs!
+
+## 0.13.0
+
+### Features
+
+- [#358](https://github.com/gradio-app/trackio/pull/358) [`073715d`](https://github.com/gradio-app/trackio/commit/073715d1caf8282f68890117f09c3ac301205312) - Improvements to `trackio.sync()`.  Thanks @abidlabs!
+
+## 0.12.0
+
+### Features
+
+- [#357](https://github.com/gradio-app/trackio/pull/357) [`02ba815`](https://github.com/gradio-app/trackio/commit/02ba815358060f1966052de051a5bdb09702920e) - Redesign media and tables to show up on separate page.  Thanks @abidlabs!
+- [#359](https://github.com/gradio-app/trackio/pull/359) [`08fe9c9`](https://github.com/gradio-app/trackio/commit/08fe9c9ddd7fe99ee811555fdfb62df9ab88e939) - docs: Improve docstrings.  Thanks @qgallouedec!
+
+## 0.11.0
+
+### Features
+
+- [#355](https://github.com/gradio-app/trackio/pull/355) [`ea51f49`](https://github.com/gradio-app/trackio/commit/ea51f4954922f21be76ef828700420fe9a912c4b) - Color code run checkboxes and match with plot lines.  Thanks @abidlabs!
+- [#353](https://github.com/gradio-app/trackio/pull/353) [`8abe691`](https://github.com/gradio-app/trackio/commit/8abe6919aeefe21fc7a23af814883efbb037c21f) - Remove show_api from demo.launch.  Thanks @sergiopaniego!
+- [#351](https://github.com/gradio-app/trackio/pull/351) [`8a8957e`](https://github.com/gradio-app/trackio/commit/8a8957e530dd7908d1fef7f2df030303f808101f) - Add `trackio.save()`.  Thanks @abidlabs!
+
+## 0.10.0
+
+### Features
+
+- [#305](https://github.com/gradio-app/trackio/pull/305) [`e64883a`](https://github.com/gradio-app/trackio/commit/e64883a51f7b8b93f7d48b8afe55acdb62238b71) - bump to gradio 6.0, make `trackio` compatible, and fix related issues.  Thanks @abidlabs!
+
+## 0.9.1
+
+### Features
+
+- [#344](https://github.com/gradio-app/trackio/pull/344) [`7e01024`](https://github.com/gradio-app/trackio/commit/7e010241d9a34794e0ce0dc19c1a6f0cf94ba856) - Avoid redundant calls to /whoami-v2.  Thanks @Wauplin!
+
+## 0.9.0
+
+### Features
+
+- [#343](https://github.com/gradio-app/trackio/pull/343) [`51bea30`](https://github.com/gradio-app/trackio/commit/51bea30f2877adff8e6497466d3a799400a0a049) - Sync offline projects to Hugging Face spaces.  Thanks @candemircan!
+- [#341](https://github.com/gradio-app/trackio/pull/341) [`4fd841f`](https://github.com/gradio-app/trackio/commit/4fd841fa190e15071b02f6fba7683ef4f393a654) - Adds a basic UI test to `trackio`.  Thanks @abidlabs!
+- [#339](https://github.com/gradio-app/trackio/pull/339) [`011d91b`](https://github.com/gradio-app/trackio/commit/011d91bb6ae266516fd250a349285670a8049d05) - Allow customzing the trackio color palette.  Thanks @abidlabs!
+
+## 0.8.1
+
+### Features
+
+- [#336](https://github.com/gradio-app/trackio/pull/336) [`5f9f51d`](https://github.com/gradio-app/trackio/commit/5f9f51dac8677f240d7c42c3e3b2660a22aee138) - Support a list of `Trackio.Image` in a `trackio.Table` cell.  Thanks @abidlabs!
+
+## 0.8.0
+
+### Features
+
+- [#331](https://github.com/gradio-app/trackio/pull/331) [`2c02d0f`](https://github.com/gradio-app/trackio/commit/2c02d0fd0a5824160528782402bb0dd4083396d5) - Truncate table string values that are greater than 250 characters (configuirable via env variable).  Thanks @abidlabs!
+- [#324](https://github.com/gradio-app/trackio/pull/324) [`50b2122`](https://github.com/gradio-app/trackio/commit/50b2122e7965ac82a72e6cb3b7d048bc10a2a6b1) - Add log y-axis functionality to UI.  Thanks @abidlabs!
+- [#326](https://github.com/gradio-app/trackio/pull/326) [`61dc1f4`](https://github.com/gradio-app/trackio/commit/61dc1f40af2f545f8e70395ddf0dbb8aee6b60d5) - Fix: improve table rendering for metrics in Trackio Dashboard.  Thanks @vigneshwaran!
+- [#328](https://github.com/gradio-app/trackio/pull/328) [`6857cbb`](https://github.com/gradio-app/trackio/commit/6857cbbe557a59a4642f210ec42566d108294e63) - Support trackio.Table with trackio.Image columns.  Thanks @abidlabs!
+- [#323](https://github.com/gradio-app/trackio/pull/323) [`6857cbb`](https://github.com/gradio-app/trackio/commit/6857cbbe557a59a4642f210ec42566d108294e63) - add Trackio client implementations in Go, Rust, and JS.  Thanks @vaibhav-research!
+
+## 0.7.0
+
+### Features
+
+- [#277](https://github.com/gradio-app/trackio/pull/277) [`db35601`](https://github.com/gradio-app/trackio/commit/db35601b9c023423c4654c9909b8ab73e58737de) - fix: make grouped runs view reflect live updates.  Thanks @Saba9!
+- [#320](https://github.com/gradio-app/trackio/pull/320) [`24ae739`](https://github.com/gradio-app/trackio/commit/24ae73969b09fb3126acd2f91647cdfbf8cf72a1) - Add additional query parms for xmin, xmax, and smoothing.  Thanks @abidlabs!
+- [#270](https://github.com/gradio-app/trackio/pull/270) [`cd1dfc3`](https://github.com/gradio-app/trackio/commit/cd1dfc3dc641b4499ac6d4a1b066fa8e2b52c57b) - feature: add support for logging audio.  Thanks @Saba9!
+
+## 0.6.0
+
+### Features
+
+- [#309](https://github.com/gradio-app/trackio/pull/309) [`1df2353`](https://github.com/gradio-app/trackio/commit/1df23534d6c01938c8db9c0f584ffa23e8d6021d) - Add histogram support with wandb-compatible API.  Thanks @abidlabs!
+- [#315](https://github.com/gradio-app/trackio/pull/315) [`76ba060`](https://github.com/gradio-app/trackio/commit/76ba06055dc43ca8f03b79f3e72d761949bd19a8) - Add guards to avoid silent fails.  Thanks @Xmaster6y!
+- [#313](https://github.com/gradio-app/trackio/pull/313) [`a606b3e`](https://github.com/gradio-app/trackio/commit/a606b3e1c5edf3d4cf9f31bd50605226a5a1c5d0) - No longer prevent certain keys from being used. Instead, dunderify them to prevent collisions with internal usage.  Thanks @abidlabs!
+- [#317](https://github.com/gradio-app/trackio/pull/317) [`27370a5`](https://github.com/gradio-app/trackio/commit/27370a595d0dbdf7eebbe7159d2ba778f039da44) - quick fixes for trackio.histogram.  Thanks @abidlabs!
+- [#312](https://github.com/gradio-app/trackio/pull/312) [`aa0f3bf`](https://github.com/gradio-app/trackio/commit/aa0f3bf372e7a0dd592a38af699c998363830eeb) - Fix video logging by adding TRACKIO_DIR to allowed_paths.  Thanks @abidlabs!
+
+## 0.5.3
+
+### Features
+
+- [#300](https://github.com/gradio-app/trackio/pull/300) [`5e4cacf`](https://github.com/gradio-app/trackio/commit/5e4cacf2e7ce527b4ce60de3a5bc05d2c02c77fb) - Adds more environment variables to allow customization of Trackio dashboard.  Thanks @abidlabs!
+
+## 0.5.2
+
+### Features
+
+- [#293](https://github.com/gradio-app/trackio/pull/293) [`64afc28`](https://github.com/gradio-app/trackio/commit/64afc28d3ea1dfd821472dc6bf0b8ed35a9b74be) - Ensures that the TRACKIO_DIR environment variable is respected.  Thanks @abidlabs!
+- [#287](https://github.com/gradio-app/trackio/pull/287) [`cd3e929`](https://github.com/gradio-app/trackio/commit/cd3e9294320949e6b8b829239069a43d5d7ff4c1) - fix(sqlite): unify .sqlite extension, allow export when DBs exist, clean WAL sidecars on import.  Thanks @vaibhav-research!
+
+### Fixes
+
+- [#291](https://github.com/gradio-app/trackio/pull/291) [`3b5adc3`](https://github.com/gradio-app/trackio/commit/3b5adc3d1f452dbab7a714d235f4974782f93730) - Fix the wheel build.  Thanks @pngwn!
+
+## 0.5.1
+
+### Fixes
+
+- [#278](https://github.com/gradio-app/trackio/pull/278) [`314c054`](https://github.com/gradio-app/trackio/commit/314c05438007ddfea3383e06fd19143e27468e2d) - Fix row orientation of metrics plots.  Thanks @abidlabs!

trackio/__init__.py

@@ -0,0 +1,567 @@
+import glob
+import json
+import logging
+import os
+import warnings
+import webbrowser
+from pathlib import Path
+from typing import Any
+
+import huggingface_hub
+from gradio.themes import ThemeClass
+from gradio.utils import TupleNoPrint
+from gradio_client import Client, handle_file
+from huggingface_hub import SpaceStorage
+from huggingface_hub.errors import LocalTokenNotFoundError
+
+from trackio import context_vars, deploy, utils
+from trackio.api import Api
+from trackio.deploy import sync
+from trackio.gpu import gpu_available, log_gpu
+from trackio.histogram import Histogram
+from trackio.imports import import_csv, import_tf_events
+from trackio.media import TrackioAudio, TrackioImage, TrackioVideo
+from trackio.run import Run
+from trackio.sqlite_storage import SQLiteStorage
+from trackio.table import Table
+from trackio.typehints import UploadEntry
+from trackio.ui.main import CSS, HEAD, demo
+from trackio.utils import TRACKIO_DIR, TRACKIO_LOGO_DIR
+
+logging.getLogger("httpx").setLevel(logging.WARNING)
+
+warnings.filterwarnings(
+    "ignore",
+    message="Empty session being created. Install gradio\\[oauth\\]",
+    category=UserWarning,
+    module="gradio.helpers",
+)
+
+__version__ = json.loads(Path(__file__).parent.joinpath("package.json").read_text())[
+    "version"
+]
+
+__all__ = [
+    "init",
+    "log",
+    "log_system",
+    "log_gpu",
+    "finish",
+    "show",
+    "sync",
+    "delete_project",
+    "import_csv",
+    "import_tf_events",
+    "save",
+    "Image",
+    "Video",
+    "Audio",
+    "Table",
+    "Histogram",
+    "Api",
+]
+
+Image = TrackioImage
+Video = TrackioVideo
+Audio = TrackioAudio
+
+
+config = {}
+
+
+def init(
+    project: str,
+    name: str | None = None,
+    group: str | None = None,
+    space_id: str | None = None,
+    space_storage: SpaceStorage | None = None,
+    dataset_id: str | None = None,
+    config: dict | None = None,
+    resume: str = "never",
+    settings: Any = None,
+    private: bool | None = None,
+    embed: bool = True,
+    auto_log_gpu: bool | None = None,
+    gpu_log_interval: float = 10.0,
+) -> Run:
+    """
+    Creates a new Trackio project and returns a [`Run`] object.
+
+    Args:
+        project (`str`):
+            The name of the project (can be an existing project to continue tracking or
+            a new project to start tracking from scratch).
+        name (`str`, *optional*):
+            The name of the run (if not provided, a default name will be generated).
+        group (`str`, *optional*):
+            The name of the group which this run belongs to in order to help organize
+            related runs together. You can toggle the entire group's visibilitiy in the
+            dashboard.
+        space_id (`str`, *optional*):
+            If provided, the project will be logged to a Hugging Face Space instead of
+            a local directory. Should be a complete Space name like
+            `"username/reponame"` or `"orgname/reponame"`, or just `"reponame"` in which
+            case the Space will be created in the currently-logged-in Hugging Face
+            user's namespace. If the Space does not exist, it will be created. If the
+            Space already exists, the project will be logged to it.
+        space_storage ([`~huggingface_hub.SpaceStorage`], *optional*):
+            Choice of persistent storage tier.
+        dataset_id (`str`, *optional*):
+            If a `space_id` is provided, a persistent Hugging Face Dataset will be
+            created and the metrics will be synced to it every 5 minutes. Specify a
+            Dataset with name like `"username/datasetname"` or `"orgname/datasetname"`,
+            or `"datasetname"` (uses currently-logged-in Hugging Face user's namespace),
+            or `None` (uses the same name as the Space but with the `"_dataset"`
+            suffix). If the Dataset does not exist, it will be created. If the Dataset
+            already exists, the project will be appended to it.
+        config (`dict`, *optional*):
+            A dictionary of configuration options. Provided for compatibility with
+            `wandb.init()`.
+        resume (`str`, *optional*, defaults to `"never"`):
+            Controls how to handle resuming a run. Can be one of:
+
+            - `"must"`: Must resume the run with the given name, raises error if run
+              doesn't exist
+            - `"allow"`: Resume the run if it exists, otherwise create a new run
+            - `"never"`: Never resume a run, always create a new one
+        private (`bool`, *optional*):
+            Whether to make the Space private. If None (default), the repo will be
+            public unless the organization's default is private. This value is ignored
+            if the repo already exists.
+        settings (`Any`, *optional*):
+            Not used. Provided for compatibility with `wandb.init()`.
+        embed (`bool`, *optional*, defaults to `True`):
+            If running inside a jupyter/Colab notebook, whether the dashboard should
+            automatically be embedded in the cell when trackio.init() is called.
+        auto_log_gpu (`bool` or `None`, *optional*, defaults to `None`):
+            Controls automatic GPU metrics logging. If `None` (default), GPU logging
+            is automatically enabled when `nvidia-ml-py` is installed and an NVIDIA
+            GPU is detected. Set to `True` to force enable or `False` to disable.
+        gpu_log_interval (`float`, *optional*, defaults to `10.0`):
+            The interval in seconds between automatic GPU metric logs.
+            Only used when `auto_log_gpu=True`.
+
+    Returns:
+        `Run`: A [`Run`] object that can be used to log metrics and finish the run.
+    """
+    if settings is not None:
+        warnings.warn(
+            "* Warning: settings is not used. Provided for compatibility with wandb.init(). Please create an issue at: https://github.com/gradio-app/trackio/issues if you need a specific feature implemented."
+        )
+
+    if space_id is None and dataset_id is not None:
+        raise ValueError("Must provide a `space_id` when `dataset_id` is provided.")
+    try:
+        space_id, dataset_id = utils.preprocess_space_and_dataset_ids(
+            space_id, dataset_id
+        )
+    except LocalTokenNotFoundError as e:
+        raise LocalTokenNotFoundError(
+            f"You must be logged in to Hugging Face locally when `space_id` is provided to deploy to a Space. {e}"
+        ) from e
+    url = context_vars.current_server.get()
+    share_url = context_vars.current_share_server.get()
+
+    if url is None:
+        if space_id is None:
+            _, url, share_url = demo.launch(
+                css=CSS,
+                head=HEAD,
+                footer_links=["gradio", "settings"],
+                inline=False,
+                quiet=True,
+                prevent_thread_lock=True,
+                show_error=True,
+                favicon_path=TRACKIO_LOGO_DIR / "trackio_logo_light.png",
+                allowed_paths=[TRACKIO_LOGO_DIR, TRACKIO_DIR],
+                ssr_mode=False,
+            )
+            context_vars.current_space_id.set(None)
+        else:
+            url = space_id
+            share_url = None
+            context_vars.current_space_id.set(space_id)
+
+        context_vars.current_server.set(url)
+        context_vars.current_share_server.set(share_url)
+    if (
+        context_vars.current_project.get() is None
+        or context_vars.current_project.get() != project
+    ):
+        print(f"* Trackio project initialized: {project}")
+
+        if dataset_id is not None:
+            os.environ["TRACKIO_DATASET_ID"] = dataset_id
+            print(
+                f"* Trackio metrics will be synced to Hugging Face Dataset: {dataset_id}"
+            )
+        if space_id is None:
+            print(f"* Trackio metrics logged to: {TRACKIO_DIR}")
+            if utils.is_in_notebook() and embed:
+                base_url = share_url + "/" if share_url else url
+                full_url = utils.get_full_url(
+                    base_url, project=project, write_token=demo.write_token, footer=True
+                )
+                utils.embed_url_in_notebook(full_url)
+            else:
+                utils.print_dashboard_instructions(project)
+        else:
+            deploy.create_space_if_not_exists(
+                space_id, space_storage, dataset_id, private
+            )
+            user_name, space_name = space_id.split("/")
+            space_url = deploy.SPACE_HOST_URL.format(
+                user_name=user_name, space_name=space_name
+            )
+            print(f"* View dashboard by going to: {space_url}")
+            if utils.is_in_notebook() and embed:
+                utils.embed_url_in_notebook(space_url)
+    context_vars.current_project.set(project)
+
+    client = None
+    if not space_id:
+        client = Client(url, verbose=False)
+
+    if resume == "must":
+        if name is None:
+            raise ValueError("Must provide a run name when resume='must'")
+        if name not in SQLiteStorage.get_runs(project):
+            raise ValueError(f"Run '{name}' does not exist in project '{project}'")
+        resumed = True
+    elif resume == "allow":
+        resumed = name is not None and name in SQLiteStorage.get_runs(project)
+    elif resume == "never":
+        if name is not None and name in SQLiteStorage.get_runs(project):
+            warnings.warn(
+                f"* Warning: resume='never' but a run '{name}' already exists in "
+                f"project '{project}'. Generating a new name and instead. If you want "
+                "to resume this run, call init() with resume='must' or resume='allow'."
+            )
+            name = None
+        resumed = False
+    else:
+        raise ValueError("resume must be one of: 'must', 'allow', or 'never'")
+
+    if auto_log_gpu is None:
+        auto_log_gpu = gpu_available()
+        if auto_log_gpu:
+            print("* GPU detected, enabling automatic GPU metrics logging")
+
+    run = Run(
+        url=url,
+        project=project,
+        client=client,
+        name=name,
+        group=group,
+        config=config,
+        space_id=space_id,
+        auto_log_gpu=auto_log_gpu,
+        gpu_log_interval=gpu_log_interval,
+    )
+
+    if resumed:
+        print(f"* Resumed existing run: {run.name}")
+    else:
+        print(f"* Created new run: {run.name}")
+
+    context_vars.current_run.set(run)
+    globals()["config"] = run.config
+    return run
+
+
+def log(metrics: dict, step: int | None = None) -> None:
+    """
+    Logs metrics to the current run.
+
+    Args:
+        metrics (`dict`):
+            A dictionary of metrics to log.
+        step (`int`, *optional*):
+            The step number. If not provided, the step will be incremented
+            automatically.
+    """
+    run = context_vars.current_run.get()
+    if run is None:
+        raise RuntimeError("Call trackio.init() before trackio.log().")
+    run.log(
+        metrics=metrics,
+        step=step,
+    )
+
+
+def log_system(metrics: dict) -> None:
+    """
+    Logs system metrics (GPU, etc.) to the current run using timestamps instead of steps.
+
+    Args:
+        metrics (`dict`):
+            A dictionary of system metrics to log.
+    """
+    run = context_vars.current_run.get()
+    if run is None:
+        raise RuntimeError("Call trackio.init() before trackio.log_system().")
+    run.log_system(metrics=metrics)
+
+
+def finish():
+    """
+    Finishes the current run.
+    """
+    run = context_vars.current_run.get()
+    if run is None:
+        raise RuntimeError("Call trackio.init() before trackio.finish().")
+    run.finish()
+
+
+def delete_project(project: str, force: bool = False) -> bool:
+    """
+    Deletes a project by removing its local SQLite database.
+
+    Args:
+        project (`str`):
+            The name of the project to delete.
+        force (`bool`, *optional*, defaults to `False`):
+            If `True`, deletes the project without prompting for confirmation.
+            If `False`, prompts the user to confirm before deleting.
+
+    Returns:
+        `bool`: `True` if the project was deleted, `False` otherwise.
+    """
+    db_path = SQLiteStorage.get_project_db_path(project)
+
+    if not db_path.exists():
+        print(f"* Project '{project}' does not exist.")
+        return False
+
+    if not force:
+        response = input(
+            f"Are you sure you want to delete project '{project}'? "
+            f"This will permanently delete all runs and metrics. (y/N): "
+        )
+        if response.lower() not in ["y", "yes"]:
+            print("* Deletion cancelled.")
+            return False
+
+    try:
+        db_path.unlink()
+
+        for suffix in ("-wal", "-shm"):
+            sidecar = Path(str(db_path) + suffix)
+            if sidecar.exists():
+                sidecar.unlink()
+
+        print(f"* Project '{project}' has been deleted.")
+        return True
+    except Exception as e:
+        print(f"* Error deleting project '{project}': {e}")
+        return False
+
+
+def save(
+    glob_str: str | Path,
+    project: str | None = None,
+) -> str:
+    """
+    Saves files to a project (not linked to a specific run). If Trackio is running
+    locally, the file(s) will be moved to the project's files directory. If Trackio is
+    running in a Space, the file(s) will be uploaded to the Space's files directory.
+
+    Args:
+        glob_str (`str` or `Path`):
+            The file path or glob pattern to save. Can be a single file or a pattern
+            matching multiple files (e.g., `"*.py"`, `"models/**/*.pth"`).
+        project (`str`, *optional*):
+            The name of the project to save files to. If not provided, uses the current
+            project from `trackio.init()`. If no project is initialized, raises an
+            error.
+
+    Returns:
+        `str`: The path where the file(s) were saved (project's files directory).
+
+    Example:
+        ```python
+        import trackio
+
+        trackio.init(project="my-project")
+        trackio.save("config.yaml")
+        trackio.save("models/*.pth")
+        ```
+    """
+    if project is None:
+        project = context_vars.current_project.get()
+        if project is None:
+            raise RuntimeError(
+                "No project specified. Either call trackio.init() first or provide a "
+                "project parameter to trackio.save()."
+            )
+
+    glob_str = Path(glob_str)
+    base_path = Path.cwd().resolve()
+
+    matched_files = []
+    if glob_str.is_file():
+        matched_files = [glob_str.resolve()]
+    else:
+        pattern = str(glob_str)
+        if not glob_str.is_absolute():
+            pattern = str((Path.cwd() / glob_str).resolve())
+        matched_files = [
+            Path(f).resolve()
+            for f in glob.glob(pattern, recursive=True)
+            if Path(f).is_file()
+        ]
+
+    if not matched_files:
+        raise ValueError(f"No files found matching pattern: {glob_str}")
+
+    url = context_vars.current_server.get()
+    current_run = context_vars.current_run.get()
+
+    upload_entries = []
+
+    for file_path in matched_files:
+        try:
+            relative_to_base = file_path.relative_to(base_path)
+        except ValueError:
+            relative_to_base = Path(file_path.name)
+
+        if current_run is not None:
+            # If a run is active, use its queue to upload the file to the project's files directory
+            # as it's more efficent than uploading files one by one. But we should not use the run name
+            # as the files should be stored in the project's files directory, not the run's, hence
+            # the use_run_name flag is set to False.
+            current_run._queue_upload(
+                file_path,
+                step=None,
+                relative_path=str(relative_to_base.parent),
+                use_run_name=False,
+            )
+        else:
+            upload_entry: UploadEntry = {
+                "project": project,
+                "run": None,
+                "step": None,
+                "relative_path": str(relative_to_base),
+                "uploaded_file": handle_file(file_path),
+            }
+            upload_entries.append(upload_entry)
+
+    if upload_entries:
+        if url is None:
+            raise RuntimeError(
+                "No server available. Call trackio.init() before trackio.save() to start the server."
+            )
+
+        try:
+            client = Client(url, verbose=False, httpx_kwargs={"timeout": 90})
+            client.predict(
+                api_name="/bulk_upload_media",
+                uploads=upload_entries,
+                hf_token=huggingface_hub.utils.get_token(),
+            )
+        except Exception as e:
+            warnings.warn(
+                f"Failed to upload files: {e}. "
+                "Files may not be available in the dashboard."
+            )
+
+    return str(utils.MEDIA_DIR / project / "files")
+
+
+def show(
+    project: str | None = None,
+    *,
+    theme: str | ThemeClass | None = None,
+    mcp_server: bool | None = None,
+    footer: bool = True,
+    color_palette: list[str] | None = None,
+    open_browser: bool = True,
+    block_thread: bool | None = None,
+    host: str | None = None,
+):
+    """
+    Launches the Trackio dashboard.
+
+    Args:
+        project (`str`, *optional*):
+            The name of the project whose runs to show. If not provided, all projects
+            will be shown and the user can select one.
+        theme (`str` or `ThemeClass`, *optional*):
+            A Gradio Theme to use for the dashboard instead of the default Gradio theme,
+            can be a built-in theme (e.g. `'soft'`, `'citrus'`), a theme from the Hub
+            (e.g. `"gstaff/xkcd"`), or a custom Theme class. If not provided, the
+            `TRACKIO_THEME` environment variable will be used, or if that is not set,
+            the default Gradio theme will be used.
+        mcp_server (`bool`, *optional*):
+            If `True`, the Trackio dashboard will be set up as an MCP server and certain
+            functions will be added as MCP tools. If `None` (default behavior), then the
+            `GRADIO_MCP_SERVER` environment variable will be used to determine if the
+            MCP server should be enabled (which is `"True"` on Hugging Face Spaces).
+        footer (`bool`, *optional*, defaults to `True`):
+            Whether to show the Gradio footer. When `False`, the footer will be hidden.
+            This can also be controlled via the `footer` query parameter in the URL.
+        color_palette (`list[str]`, *optional*):
+            A list of hex color codes to use for plot lines. If not provided, the
+            `TRACKIO_COLOR_PALETTE` environment variable will be used (comma-separated
+            hex codes), or if that is not set, the default color palette will be used.
+            Example: `['#FF0000', '#00FF00', '#0000FF']`
+        open_browser (`bool`, *optional*, defaults to `True`):
+            If `True` and not in a notebook, a new browser tab will be opened with the
+            dashboard. If `False`, the browser will not be opened.
+        block_thread (`bool`, *optional*):
+            If `True`, the main thread will be blocked until the dashboard is closed.
+            If `None` (default behavior), then the main thread will not be blocked if the
+            dashboard is launched in a notebook, otherwise the main thread will be blocked.
+        host (`str`, *optional*):
+            The host to bind the server to. If not provided, defaults to `'127.0.0.1'`
+            (localhost only). Set to `'0.0.0.0'` to allow remote access.
+
+        Returns:
+            `app`: The Gradio app object corresponding to the dashboard launched by Trackio.
+            `url`: The local URL of the dashboard.
+            `share_url`: The public share URL of the dashboard.
+            `full_url`: The full URL of the dashboard including the write token (will use the public share URL if launched publicly, otherwise the local URL).
+    """
+    if color_palette is not None:
+        os.environ["TRACKIO_COLOR_PALETTE"] = ",".join(color_palette)
+
+    theme = theme or os.environ.get("TRACKIO_THEME")
+
+    _mcp_server = (
+        mcp_server
+        if mcp_server is not None
+        else os.environ.get("GRADIO_MCP_SERVER", "False") == "True"
+    )
+
+    app, url, share_url = demo.launch(
+        css=CSS,
+        head=HEAD,
+        footer_links=["gradio", "settings"] + (["api"] if _mcp_server else []),
+        quiet=True,
+        inline=False,
+        prevent_thread_lock=True,
+        favicon_path=TRACKIO_LOGO_DIR / "trackio_logo_light.png",
+        allowed_paths=[TRACKIO_LOGO_DIR, TRACKIO_DIR],
+        mcp_server=_mcp_server,
+        theme=theme,
+        ssr_mode=False,
+        server_name=host,
+    )
+
+    base_url = share_url + "/" if share_url else url
+    full_url = utils.get_full_url(
+        base_url, project=project, write_token=demo.write_token, footer=footer
+    )
+
+    if not utils.is_in_notebook():
+        print(f"* Trackio UI launched at: {full_url}")
+        if open_browser:
+            webbrowser.open(full_url)
+        block_thread = block_thread if block_thread is not None else True
+    else:
+        utils.embed_url_in_notebook(full_url)
+        block_thread = block_thread if block_thread is not None else False
+
+    if block_thread:
+        utils.block_main_thread_until_keyboard_interrupt()
+    return TupleNoPrint((demo, url, share_url, full_url))

trackio/api.py

@@ -0,0 +1,66 @@
+from typing import Iterator
+
+from trackio.sqlite_storage import SQLiteStorage
+
+
+class Run:
+    def __init__(self, project: str, name: str):
+        self.project = project
+        self.name = name
+        self._config = None
+
+    @property
+    def id(self) -> str:
+        return self.name
+
+    @property
+    def config(self) -> dict | None:
+        if self._config is None:
+            self._config = SQLiteStorage.get_run_config(self.project, self.name)
+        return self._config
+
+    def delete(self) -> bool:
+        return SQLiteStorage.delete_run(self.project, self.name)
+
+    def move(self, new_project: str) -> bool:
+        success = SQLiteStorage.move_run(self.project, self.name, new_project)
+        if success:
+            self.project = new_project
+        return success
+
+    def __repr__(self) -> str:
+        return f"<Run {self.name} in project {self.project}>"
+
+
+class Runs:
+    def __init__(self, project: str):
+        self.project = project
+        self._runs = None
+
+    def _load_runs(self):
+        if self._runs is None:
+            run_names = SQLiteStorage.get_runs(self.project)
+            self._runs = [Run(self.project, name) for name in run_names]
+
+    def __iter__(self) -> Iterator[Run]:
+        self._load_runs()
+        return iter(self._runs)
+
+    def __getitem__(self, index: int) -> Run:
+        self._load_runs()
+        return self._runs[index]
+
+    def __len__(self) -> int:
+        self._load_runs()
+        return len(self._runs)
+
+    def __repr__(self) -> str:
+        self._load_runs()
+        return f"<Runs project={self.project} count={len(self._runs)}>"
+
+
+class Api:
+    def runs(self, project: str) -> Runs:
+        if not SQLiteStorage.get_project_db_path(project).exists():
+            raise ValueError(f"Project '{project}' does not exist")
+        return Runs(project)

trackio/cli.py

@@ -0,0 +1,433 @@
+import argparse
+
+from trackio import show, sync
+from trackio.cli_helpers import (
+    error_exit,
+    format_json,
+    format_list,
+    format_metric_values,
+    format_project_summary,
+    format_run_summary,
+    format_system_metric_names,
+    format_system_metrics,
+)
+from trackio.sqlite_storage import SQLiteStorage
+from trackio.ui.main import get_project_summary, get_run_summary
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Trackio CLI")
+    subparsers = parser.add_subparsers(dest="command")
+
+    ui_parser = subparsers.add_parser(
+        "show", help="Show the Trackio dashboard UI for a project"
+    )
+    ui_parser.add_argument(
+        "--project", required=False, help="Project name to show in the dashboard"
+    )
+    ui_parser.add_argument(
+        "--theme",
+        required=False,
+        default="default",
+        help="A Gradio Theme to use for the dashboard instead of the default, can be a built-in theme (e.g. 'soft', 'citrus'), or a theme from the Hub (e.g. 'gstaff/xkcd').",
+    )
+    ui_parser.add_argument(
+        "--mcp-server",
+        action="store_true",
+        help="Enable MCP server functionality. The Trackio dashboard will be set up as an MCP server and certain functions will be exposed as MCP tools.",
+    )
+    ui_parser.add_argument(
+        "--footer",
+        action="store_true",
+        default=True,
+        help="Show the Gradio footer. Use --no-footer to hide it.",
+    )
+    ui_parser.add_argument(
+        "--no-footer",
+        dest="footer",
+        action="store_false",
+        help="Hide the Gradio footer.",
+    )
+    ui_parser.add_argument(
+        "--color-palette",
+        required=False,
+        help="Comma-separated list of hex color codes for plot lines (e.g. '#FF0000,#00FF00,#0000FF'). If not provided, the TRACKIO_COLOR_PALETTE environment variable will be used, or the default palette if not set.",
+    )
+    ui_parser.add_argument(
+        "--host",
+        required=False,
+        help="Host to bind the server to (e.g. '0.0.0.0' for remote access). If not provided, defaults to '127.0.0.1' (localhost only).",
+    )
+
+    sync_parser = subparsers.add_parser(
+        "sync",
+        help="Sync a local project's database to a Hugging Face Space. If the Space does not exist, it will be created.",
+    )
+    sync_parser.add_argument(
+        "--project", required=True, help="The name of the local project."
+    )
+    sync_parser.add_argument(
+        "--space-id",
+        required=True,
+        help="The Hugging Face Space ID where the project will be synced (e.g. username/space_id).",
+    )
+    sync_parser.add_argument(
+        "--private",
+        action="store_true",
+        help="Make the Hugging Face Space private if creating a new Space. By default, the repo will be public unless the organization's default is private. This value is ignored if the repo already exists.",
+    )
+    sync_parser.add_argument(
+        "--force",
+        action="store_true",
+        help="Overwrite the existing database without prompting for confirmation.",
+    )
+
+    list_parser = subparsers.add_parser(
+        "list",
+        help="List projects, runs, or metrics",
+    )
+    list_subparsers = list_parser.add_subparsers(dest="list_type", required=True)
+
+    list_projects_parser = list_subparsers.add_parser(
+        "projects",
+        help="List all projects",
+    )
+    list_projects_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    list_runs_parser = list_subparsers.add_parser(
+        "runs",
+        help="List runs for a project",
+    )
+    list_runs_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    list_runs_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    list_metrics_parser = list_subparsers.add_parser(
+        "metrics",
+        help="List metrics for a run",
+    )
+    list_metrics_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    list_metrics_parser.add_argument(
+        "--run",
+        required=True,
+        help="Run name",
+    )
+    list_metrics_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    list_system_metrics_parser = list_subparsers.add_parser(
+        "system-metrics",
+        help="List system metrics for a run",
+    )
+    list_system_metrics_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    list_system_metrics_parser.add_argument(
+        "--run",
+        required=True,
+        help="Run name",
+    )
+    list_system_metrics_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    get_parser = subparsers.add_parser(
+        "get",
+        help="Get project, run, or metric information",
+    )
+    get_subparsers = get_parser.add_subparsers(dest="get_type", required=True)
+
+    get_project_parser = get_subparsers.add_parser(
+        "project",
+        help="Get project summary",
+    )
+    get_project_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    get_project_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    get_run_parser = get_subparsers.add_parser(
+        "run",
+        help="Get run summary",
+    )
+    get_run_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    get_run_parser.add_argument(
+        "--run",
+        required=True,
+        help="Run name",
+    )
+    get_run_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    get_metric_parser = get_subparsers.add_parser(
+        "metric",
+        help="Get metric values for a run",
+    )
+    get_metric_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    get_metric_parser.add_argument(
+        "--run",
+        required=True,
+        help="Run name",
+    )
+    get_metric_parser.add_argument(
+        "--metric",
+        required=True,
+        help="Metric name",
+    )
+    get_metric_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    get_system_metric_parser = get_subparsers.add_parser(
+        "system-metric",
+        help="Get system metric values for a run",
+    )
+    get_system_metric_parser.add_argument(
+        "--project",
+        required=True,
+        help="Project name",
+    )
+    get_system_metric_parser.add_argument(
+        "--run",
+        required=True,
+        help="Run name",
+    )
+    get_system_metric_parser.add_argument(
+        "--metric",
+        required=False,
+        help="System metric name (optional, if not provided returns all system metrics)",
+    )
+    get_system_metric_parser.add_argument(
+        "--json",
+        action="store_true",
+        help="Output in JSON format",
+    )
+
+    args = parser.parse_args()
+
+    if args.command == "show":
+        color_palette = None
+        if args.color_palette:
+            color_palette = [color.strip() for color in args.color_palette.split(",")]
+        show(
+            project=args.project,
+            theme=args.theme,
+            mcp_server=args.mcp_server,
+            footer=args.footer,
+            color_palette=color_palette,
+            host=args.host,
+        )
+    elif args.command == "sync":
+        sync(
+            project=args.project,
+            space_id=args.space_id,
+            private=args.private,
+            force=args.force,
+        )
+    elif args.command == "list":
+        if args.list_type == "projects":
+            projects = SQLiteStorage.get_projects()
+            if args.json:
+                print(format_json({"projects": projects}))
+            else:
+                print(format_list(projects, "Projects"))
+        elif args.list_type == "runs":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.json:
+                print(format_json({"project": args.project, "runs": runs}))
+            else:
+                print(format_list(runs, f"Runs in '{args.project}'"))
+        elif args.list_type == "metrics":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.run not in runs:
+                error_exit(f"Run '{args.run}' not found in project '{args.project}'.")
+            metrics = SQLiteStorage.get_all_metrics_for_run(args.project, args.run)
+            if args.json:
+                print(
+                    format_json(
+                        {"project": args.project, "run": args.run, "metrics": metrics}
+                    )
+                )
+            else:
+                print(
+                    format_list(
+                        metrics, f"Metrics for '{args.run}' in '{args.project}'"
+                    )
+                )
+        elif args.list_type == "system-metrics":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.run not in runs:
+                error_exit(f"Run '{args.run}' not found in project '{args.project}'.")
+            system_metrics = SQLiteStorage.get_all_system_metrics_for_run(
+                args.project, args.run
+            )
+            if args.json:
+                print(
+                    format_json(
+                        {
+                            "project": args.project,
+                            "run": args.run,
+                            "system_metrics": system_metrics,
+                        }
+                    )
+                )
+            else:
+                print(format_system_metric_names(system_metrics))
+    elif args.command == "get":
+        if args.get_type == "project":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            summary = get_project_summary(args.project)
+            if args.json:
+                print(format_json(summary))
+            else:
+                print(format_project_summary(summary))
+        elif args.get_type == "run":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.run not in runs:
+                error_exit(f"Run '{args.run}' not found in project '{args.project}'.")
+            summary = get_run_summary(args.project, args.run)
+            if args.json:
+                print(format_json(summary))
+            else:
+                print(format_run_summary(summary))
+        elif args.get_type == "metric":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.run not in runs:
+                error_exit(f"Run '{args.run}' not found in project '{args.project}'.")
+            metrics = SQLiteStorage.get_all_metrics_for_run(args.project, args.run)
+            if args.metric not in metrics:
+                error_exit(
+                    f"Metric '{args.metric}' not found in run '{args.run}' of project '{args.project}'."
+                )
+            values = SQLiteStorage.get_metric_values(
+                args.project, args.run, args.metric
+            )
+            if args.json:
+                print(
+                    format_json(
+                        {
+                            "project": args.project,
+                            "run": args.run,
+                            "metric": args.metric,
+                            "values": values,
+                        }
+                    )
+                )
+            else:
+                print(format_metric_values(values))
+        elif args.get_type == "system-metric":
+            db_path = SQLiteStorage.get_project_db_path(args.project)
+            if not db_path.exists():
+                error_exit(f"Project '{args.project}' not found.")
+            runs = SQLiteStorage.get_runs(args.project)
+            if args.run not in runs:
+                error_exit(f"Run '{args.run}' not found in project '{args.project}'.")
+            if args.metric:
+                system_metrics = SQLiteStorage.get_system_logs(args.project, args.run)
+                all_system_metric_names = SQLiteStorage.get_all_system_metrics_for_run(
+                    args.project, args.run
+                )
+                if args.metric not in all_system_metric_names:
+                    error_exit(
+                        f"System metric '{args.metric}' not found in run '{args.run}' of project '{args.project}'."
+                    )
+                filtered_metrics = [
+                    {
+                        k: v
+                        for k, v in entry.items()
+                        if k == "timestamp" or k == args.metric
+                    }
+                    for entry in system_metrics
+                    if args.metric in entry
+                ]
+                if args.json:
+                    print(
+                        format_json(
+                            {
+                                "project": args.project,
+                                "run": args.run,
+                                "metric": args.metric,
+                                "values": filtered_metrics,
+                            }
+                        )
+                    )
+                else:
+                    print(format_system_metrics(filtered_metrics))
+            else:
+                system_metrics = SQLiteStorage.get_system_logs(args.project, args.run)
+                if args.json:
+                    print(
+                        format_json(
+                            {
+                                "project": args.project,
+                                "run": args.run,
+                                "system_metrics": system_metrics,
+                            }
+                        )
+                    )
+                else:
+                    print(format_system_metrics(system_metrics))
+    else:
+        parser.print_help()
+
+
+if __name__ == "__main__":
+    main()

trackio/cli_helpers.py

@@ -0,0 +1,118 @@
+import json
+import sys
+from typing import Any
+
+
+def format_json(data: Any) -> str:
+    """Format data as JSON."""
+    return json.dumps(data, indent=2)
+
+
+def format_list(items: list[str], title: str | None = None) -> str:
+    """Format a list of items in human-readable format."""
+    if not items:
+        return f"No {title.lower() if title else 'items'} found."
+
+    output = []
+    if title:
+        output.append(f"{title}:")
+
+    for item in items:
+        output.append(f"  - {item}")
+
+    return "\n".join(output)
+
+
+def format_project_summary(summary: dict) -> str:
+    """Format project summary in human-readable format."""
+    output = [f"Project: {summary['project']}"]
+    output.append(f"Number of runs: {summary['num_runs']}")
+
+    if summary["runs"]:
+        output.append("\nRuns:")
+        for run in summary["runs"]:
+            output.append(f"  - {run}")
+    else:
+        output.append("\nNo runs found.")
+
+    if summary.get("last_activity"):
+        output.append(f"\nLast activity (max step): {summary['last_activity']}")
+
+    return "\n".join(output)
+
+
+def format_run_summary(summary: dict) -> str:
+    """Format run summary in human-readable format."""
+    output = [f"Project: {summary['project']}"]
+    output.append(f"Run: {summary['run']}")
+    output.append(f"Number of logs: {summary['num_logs']}")
+
+    if summary.get("last_step") is not None:
+        output.append(f"Last step: {summary['last_step']}")
+
+    if summary.get("metrics"):
+        output.append("\nMetrics:")
+        for metric in summary["metrics"]:
+            output.append(f"  - {metric}")
+    else:
+        output.append("\nNo metrics found.")
+
+    config = summary.get("config")
+    if config:
+        output.append("\nConfig:")
+        config_display = {k: v for k, v in config.items() if not k.startswith("_")}
+        if config_display:
+            for key, value in config_display.items():
+                output.append(f"  {key}: {value}")
+        else:
+            output.append("  (no config)")
+    else:
+        output.append("\nConfig: (no config)")
+
+    return "\n".join(output)
+
+
+def format_metric_values(values: list[dict]) -> str:
+    """Format metric values in human-readable format."""
+    if not values:
+        return "No metric values found."
+
+    output = [f"Found {len(values)} value(s):\n"]
+    output.append("Step | Timestamp | Value")
+    output.append("-" * 50)
+
+    for value in values:
+        step = value.get("step", "N/A")
+        timestamp = value.get("timestamp", "N/A")
+        val = value.get("value", "N/A")
+        output.append(f"{step} | {timestamp} | {val}")
+
+    return "\n".join(output)
+
+
+def format_system_metrics(metrics: list[dict]) -> str:
+    """Format system metrics in human-readable format."""
+    if not metrics:
+        return "No system metrics found."
+
+    output = [f"Found {len(metrics)} system metric entry/entries:\n"]
+
+    for i, entry in enumerate(metrics):
+        timestamp = entry.get("timestamp", "N/A")
+        output.append(f"\nEntry {i + 1} (Timestamp: {timestamp}):")
+        for key, value in entry.items():
+            if key != "timestamp":
+                output.append(f"  {key}: {value}")
+
+    return "\n".join(output)
+
+
+def format_system_metric_names(names: list[str]) -> str:
+    """Format system metric names in human-readable format."""
+    return format_list(names, "System Metrics")
+
+
+def error_exit(message: str, code: int = 1) -> None:
+    """Print error message and exit."""
+    print(f"Error: {message}", file=sys.stderr)
+    sys.exit(code)

trackio/commit_scheduler.py

@@ -0,0 +1,391 @@
+# Originally copied from https://github.com/huggingface/huggingface_hub/blob/d0a948fc2a32ed6e557042a95ef3e4af97ec4a7c/src/huggingface_hub/_commit_scheduler.py
+
+import atexit
+import logging
+import os
+import time
+from concurrent.futures import Future
+from dataclasses import dataclass
+from io import SEEK_END, SEEK_SET, BytesIO
+from pathlib import Path
+from threading import Lock, Thread
+from typing import Callable, Dict, List, Union
+
+from huggingface_hub.hf_api import (
+    DEFAULT_IGNORE_PATTERNS,
+    CommitInfo,
+    CommitOperationAdd,
+    HfApi,
+)
+from huggingface_hub.utils import filter_repo_objects
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass(frozen=True)
+class _FileToUpload:
+    """Temporary dataclass to store info about files to upload. Not meant to be used directly."""
+
+    local_path: Path
+    path_in_repo: str
+    size_limit: int
+    last_modified: float
+
+
+class CommitScheduler:
+    """
+    Scheduler to upload a local folder to the Hub at regular intervals (e.g. push to hub every 5 minutes).
+
+    The recommended way to use the scheduler is to use it as a context manager. This ensures that the scheduler is
+    properly stopped and the last commit is triggered when the script ends. The scheduler can also be stopped manually
+    with the `stop` method. Checkout the [upload guide](https://huggingface.co/docs/huggingface_hub/guides/upload#scheduled-uploads)
+    to learn more about how to use it.
+
+    Args:
+        repo_id (`str`):
+            The id of the repo to commit to.
+        folder_path (`str` or `Path`):
+            Path to the local folder to upload regularly.
+        every (`int` or `float`, *optional*):
+            The number of minutes between each commit. Defaults to 5 minutes.
+        path_in_repo (`str`, *optional*):
+            Relative path of the directory in the repo, for example: `"checkpoints/"`. Defaults to the root folder
+            of the repository.
+        repo_type (`str`, *optional*):
+            The type of the repo to commit to. Defaults to `model`.
+        revision (`str`, *optional*):
+            The revision of the repo to commit to. Defaults to `main`.
+        private (`bool`, *optional*):
+            Whether to make the repo private. If `None` (default), the repo will be public unless the organization's default is private. This value is ignored if the repo already exists.
+        token (`str`, *optional*):
+            The token to use to commit to the repo. Defaults to the token saved on the machine.
+        allow_patterns (`List[str]` or `str`, *optional*):
+            If provided, only files matching at least one pattern are uploaded.
+        ignore_patterns (`List[str]` or `str`, *optional*):
+            If provided, files matching any of the patterns are not uploaded.
+        squash_history (`bool`, *optional*):
+            Whether to squash the history of the repo after each commit. Defaults to `False`. Squashing commits is
+            useful to avoid degraded performances on the repo when it grows too large.
+        hf_api (`HfApi`, *optional*):
+            The [`HfApi`] client to use to commit to the Hub. Can be set with custom settings (user agent, token,...).
+        on_before_commit (`Callable[[], None]`, *optional*):
+            If specified, a function that will be called before the CommitScheduler lists files to create a commit.
+
+    Example:
+    ```py
+    >>> from pathlib import Path
+    >>> from huggingface_hub import CommitScheduler
+
+    # Scheduler uploads every 10 minutes
+    >>> csv_path = Path("watched_folder/data.csv")
+    >>> CommitScheduler(repo_id="test_scheduler", repo_type="dataset", folder_path=csv_path.parent, every=10)
+
+    >>> with csv_path.open("a") as f:
+    ...     f.write("first line")
+
+    # Some time later (...)
+    >>> with csv_path.open("a") as f:
+    ...     f.write("second line")
+    ```
+
+    Example using a context manager:
+    ```py
+    >>> from pathlib import Path
+    >>> from huggingface_hub import CommitScheduler
+
+    >>> with CommitScheduler(repo_id="test_scheduler", repo_type="dataset", folder_path="watched_folder", every=10) as scheduler:
+    ...     csv_path = Path("watched_folder/data.csv")
+    ...     with csv_path.open("a") as f:
+    ...         f.write("first line")
+    ...     (...)
+    ...     with csv_path.open("a") as f:
+    ...         f.write("second line")
+
+    # Scheduler is now stopped and last commit have been triggered
+    ```
+    """
+
+    def __init__(
+        self,
+        *,
+        repo_id: str,
+        folder_path: Union[str, Path],
+        every: Union[int, float] = 5,
+        path_in_repo: str | None = None,
+        repo_type: str | None = None,
+        revision: str | None = None,
+        private: bool | None = None,
+        token: str | None = None,
+        allow_patterns: list[str] | str | None = None,
+        ignore_patterns: list[str] | str | None = None,
+        squash_history: bool = False,
+        hf_api: HfApi | None = None,
+        on_before_commit: Callable[[], None] | None = None,
+    ) -> None:
+        self.api = hf_api or HfApi(token=token)
+        self.on_before_commit = on_before_commit
+
+        # Folder
+        self.folder_path = Path(folder_path).expanduser().resolve()
+        self.path_in_repo = path_in_repo or ""
+        self.allow_patterns = allow_patterns
+
+        if ignore_patterns is None:
+            ignore_patterns = []
+        elif isinstance(ignore_patterns, str):
+            ignore_patterns = [ignore_patterns]
+        self.ignore_patterns = ignore_patterns + DEFAULT_IGNORE_PATTERNS
+
+        if self.folder_path.is_file():
+            raise ValueError(
+                f"'folder_path' must be a directory, not a file: '{self.folder_path}'."
+            )
+        self.folder_path.mkdir(parents=True, exist_ok=True)
+
+        # Repository
+        repo_url = self.api.create_repo(
+            repo_id=repo_id, private=private, repo_type=repo_type, exist_ok=True
+        )
+        self.repo_id = repo_url.repo_id
+        self.repo_type = repo_type
+        self.revision = revision
+        self.token = token
+
+        self.last_uploaded: Dict[Path, float] = {}
+        self.last_push_time: float | None = None
+
+        if not every > 0:
+            raise ValueError(f"'every' must be a positive integer, not '{every}'.")
+        self.lock = Lock()
+        self.every = every
+        self.squash_history = squash_history
+
+        logger.info(
+            f"Scheduled job to push '{self.folder_path}' to '{self.repo_id}' every {self.every} minutes."
+        )
+        self._scheduler_thread = Thread(target=self._run_scheduler, daemon=True)
+        self._scheduler_thread.start()
+        atexit.register(self._push_to_hub)
+
+        self.__stopped = False
+
+    def stop(self) -> None:
+        """Stop the scheduler.
+
+        A stopped scheduler cannot be restarted. Mostly for tests purposes.
+        """
+        self.__stopped = True
+
+    def __enter__(self) -> "CommitScheduler":
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback) -> None:
+        # Upload last changes before exiting
+        self.trigger().result()
+        self.stop()
+        return
+
+    def _run_scheduler(self) -> None:
+        """Dumb thread waiting between each scheduled push to Hub."""
+        while True:
+            self.last_future = self.trigger()
+            time.sleep(self.every * 60)
+            if self.__stopped:
+                break
+
+    def trigger(self) -> Future:
+        """Trigger a `push_to_hub` and return a future.
+
+        This method is automatically called every `every` minutes. You can also call it manually to trigger a commit
+        immediately, without waiting for the next scheduled commit.
+        """
+        return self.api.run_as_future(self._push_to_hub)
+
+    def _push_to_hub(self) -> CommitInfo | None:
+        if self.__stopped:  # If stopped, already scheduled commits are ignored
+            return None
+
+        logger.info("(Background) scheduled commit triggered.")
+        try:
+            value = self.push_to_hub()
+            if self.squash_history:
+                logger.info("(Background) squashing repo history.")
+                self.api.super_squash_history(
+                    repo_id=self.repo_id, repo_type=self.repo_type, branch=self.revision
+                )
+            return value
+        except Exception as e:
+            logger.error(
+                f"Error while pushing to Hub: {e}"
+            )  # Depending on the setup, error might be silenced
+            raise
+
+    def push_to_hub(self) -> CommitInfo | None:
+        """
+        Push folder to the Hub and return the commit info.
+
+        <Tip warning={true}>
+
+        This method is not meant to be called directly. It is run in the background by the scheduler, respecting a
+        queue mechanism to avoid concurrent commits. Making a direct call to the method might lead to concurrency
+        issues.
+
+        </Tip>
+
+        The default behavior of `push_to_hub` is to assume an append-only folder. It lists all files in the folder and
+        uploads only changed files. If no changes are found, the method returns without committing anything. If you want
+        to change this behavior, you can inherit from [`CommitScheduler`] and override this method. This can be useful
+        for example to compress data together in a single file before committing. For more details and examples, check
+        out our [integration guide](https://huggingface.co/docs/huggingface_hub/main/en/guides/upload#scheduled-uploads).
+        """
+        # Check files to upload (with lock)
+        with self.lock:
+            if self.on_before_commit is not None:
+                self.on_before_commit()
+
+            logger.debug("Listing files to upload for scheduled commit.")
+
+            # List files from folder (taken from `_prepare_upload_folder_additions`)
+            relpath_to_abspath = {
+                path.relative_to(self.folder_path).as_posix(): path
+                for path in sorted(
+                    self.folder_path.glob("**/*")
+                )  # sorted to be deterministic
+                if path.is_file()
+            }
+            prefix = f"{self.path_in_repo.strip('/')}/" if self.path_in_repo else ""
+
+            # Filter with pattern + filter out unchanged files + retrieve current file size
+            files_to_upload: List[_FileToUpload] = []
+            for relpath in filter_repo_objects(
+                relpath_to_abspath.keys(),
+                allow_patterns=self.allow_patterns,
+                ignore_patterns=self.ignore_patterns,
+            ):
+                local_path = relpath_to_abspath[relpath]
+                stat = local_path.stat()
+                if (
+                    self.last_uploaded.get(local_path) is None
+                    or self.last_uploaded[local_path] != stat.st_mtime
+                ):
+                    files_to_upload.append(
+                        _FileToUpload(
+                            local_path=local_path,
+                            path_in_repo=prefix + relpath,
+                            size_limit=stat.st_size,
+                            last_modified=stat.st_mtime,
+                        )
+                    )
+
+        # Return if nothing to upload
+        if len(files_to_upload) == 0:
+            logger.debug("Dropping schedule commit: no changed file to upload.")
+            return None
+
+        # Convert `_FileToUpload` as `CommitOperationAdd` (=> compute file shas + limit to file size)
+        logger.debug("Removing unchanged files since previous scheduled commit.")
+        add_operations = [
+            CommitOperationAdd(
+                # TODO: Cap the file to its current size, even if the user append data to it while a scheduled commit is happening
+                # (requires an upstream fix for XET-535: `hf_xet` should support `BinaryIO` for upload)
+                path_or_fileobj=file_to_upload.local_path,
+                path_in_repo=file_to_upload.path_in_repo,
+            )
+            for file_to_upload in files_to_upload
+        ]
+
+        # Upload files (append mode expected - no need for lock)
+        logger.debug("Uploading files for scheduled commit.")
+        commit_info = self.api.create_commit(
+            repo_id=self.repo_id,
+            repo_type=self.repo_type,
+            operations=add_operations,
+            commit_message="Scheduled Commit",
+            revision=self.revision,
+        )
+
+        for file in files_to_upload:
+            self.last_uploaded[file.local_path] = file.last_modified
+
+        self.last_push_time = time.time()
+
+        return commit_info
+
+
+class PartialFileIO(BytesIO):
+    """A file-like object that reads only the first part of a file.
+
+    Useful to upload a file to the Hub when the user might still be appending data to it. Only the first part of the
+    file is uploaded (i.e. the part that was available when the filesystem was first scanned).
+
+    In practice, only used internally by the CommitScheduler to regularly push a folder to the Hub with minimal
+    disturbance for the user. The object is passed to `CommitOperationAdd`.
+
+    Only supports `read`, `tell` and `seek` methods.
+
+    Args:
+        file_path (`str` or `Path`):
+            Path to the file to read.
+        size_limit (`int`):
+            The maximum number of bytes to read from the file. If the file is larger than this, only the first part
+            will be read (and uploaded).
+    """
+
+    def __init__(self, file_path: Union[str, Path], size_limit: int) -> None:
+        self._file_path = Path(file_path)
+        self._file = self._file_path.open("rb")
+        self._size_limit = min(size_limit, os.fstat(self._file.fileno()).st_size)
+
+    def __del__(self) -> None:
+        self._file.close()
+        return super().__del__()
+
+    def __repr__(self) -> str:
+        return (
+            f"<PartialFileIO file_path={self._file_path} size_limit={self._size_limit}>"
+        )
+
+    def __len__(self) -> int:
+        return self._size_limit
+
+    def __getattribute__(self, name: str):
+        if name.startswith("_") or name in (
+            "read",
+            "tell",
+            "seek",
+        ):  # only 3 public methods supported
+            return super().__getattribute__(name)
+        raise NotImplementedError(f"PartialFileIO does not support '{name}'.")
+
+    def tell(self) -> int:
+        """Return the current file position."""
+        return self._file.tell()
+
+    def seek(self, __offset: int, __whence: int = SEEK_SET) -> int:
+        """Change the stream position to the given offset.
+
+        Behavior is the same as a regular file, except that the position is capped to the size limit.
+        """
+        if __whence == SEEK_END:
+            # SEEK_END => set from the truncated end
+            __offset = len(self) + __offset
+            __whence = SEEK_SET
+
+        pos = self._file.seek(__offset, __whence)
+        if pos > self._size_limit:
+            return self._file.seek(self._size_limit)
+        return pos
+
+    def read(self, __size: int | None = -1) -> bytes:
+        """Read at most `__size` bytes from the file.
+
+        Behavior is the same as a regular file, except that it is capped to the size limit.
+        """
+        current = self._file.tell()
+        if __size is None or __size < 0:
+            # Read until file limit
+            truncated_size = self._size_limit - current
+        else:
+            # Read until file limit or __size
+            truncated_size = min(__size, self._size_limit - current)
+        return self._file.read(truncated_size)

trackio/context_vars.py

@@ -0,0 +1,21 @@
+import contextvars
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from trackio.run import Run
+
+current_run: contextvars.ContextVar["Run | None"] = contextvars.ContextVar(
+    "current_run", default=None
+)
+current_project: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "current_project", default=None
+)
+current_server: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "current_server", default=None
+)
+current_space_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "current_space_id", default=None
+)
+current_share_server: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "current_share_server", default=None
+)

trackio/deploy.py

@@ -0,0 +1,340 @@
+import importlib.metadata
+import io
+import os
+import sys
+import threading
+import time
+from importlib.resources import files
+from pathlib import Path
+
+if sys.version_info >= (3, 11):
+    import tomllib
+else:
+    import tomli as tomllib
+
+import gradio
+import huggingface_hub
+from gradio_client import Client, handle_file
+from httpx import ReadTimeout
+from huggingface_hub.errors import HfHubHTTPError, RepositoryNotFoundError
+
+import trackio
+from trackio.sqlite_storage import SQLiteStorage
+from trackio.utils import get_or_create_project_hash, preprocess_space_and_dataset_ids
+
+SPACE_HOST_URL = "https://{user_name}-{space_name}.hf.space/"
+SPACE_URL = "https://huggingface.co/spaces/{space_id}"
+
+
+def _get_source_install_dependencies() -> str:
+    """Get trackio dependencies from pyproject.toml for source installs."""
+    trackio_path = files("trackio")
+    pyproject_path = Path(trackio_path).parent / "pyproject.toml"
+    with open(pyproject_path, "rb") as f:
+        pyproject = tomllib.load(f)
+    deps = pyproject["project"]["dependencies"]
+    spaces_deps = (
+        pyproject["project"].get("optional-dependencies", {}).get("spaces", [])
+    )
+    return "\n".join(deps + spaces_deps)
+
+
+def _is_trackio_installed_from_source() -> bool:
+    """Check if trackio is installed from source/editable install vs PyPI."""
+    try:
+        trackio_file = trackio.__file__
+        if "site-packages" not in trackio_file:
+            return True
+
+        dist = importlib.metadata.distribution("trackio")
+        if dist.files:
+            files = list(dist.files)
+            has_pth = any(".pth" in str(f) for f in files)
+            if has_pth:
+                return True
+
+        return False
+    except (
+        AttributeError,
+        importlib.metadata.PackageNotFoundError,
+        importlib.metadata.MetadataError,
+        ValueError,
+        TypeError,
+    ):
+        return True
+
+
+def deploy_as_space(
+    space_id: str,
+    space_storage: huggingface_hub.SpaceStorage | None = None,
+    dataset_id: str | None = None,
+    private: bool | None = None,
+):
+    if (
+        os.getenv("SYSTEM") == "spaces"
+    ):  # in case a repo with this function is uploaded to spaces
+        return
+
+    trackio_path = files("trackio")
+
+    hf_api = huggingface_hub.HfApi()
+
+    try:
+        huggingface_hub.create_repo(
+            space_id,
+            private=private,
+            space_sdk="gradio",
+            space_storage=space_storage,
+            repo_type="space",
+            exist_ok=True,
+        )
+    except HfHubHTTPError as e:
+        if e.response.status_code in [401, 403]:  # unauthorized or forbidden
+            print("Need 'write' access token to create a Spaces repo.")
+            huggingface_hub.login(add_to_git_credential=False)
+            huggingface_hub.create_repo(
+                space_id,
+                private=private,
+                space_sdk="gradio",
+                space_storage=space_storage,
+                repo_type="space",
+                exist_ok=True,
+            )
+        else:
+            raise ValueError(f"Failed to create Space: {e}")
+
+    # We can assume pandas, gradio, and huggingface-hub are already installed in a Gradio Space.
+    # Make sure necessary dependencies are installed by creating a requirements.txt.
+    is_source_install = _is_trackio_installed_from_source()
+
+    with open(Path(trackio_path, "README.md"), "r") as f:
+        readme_content = f.read()
+        readme_content = readme_content.replace("{GRADIO_VERSION}", gradio.__version__)
+        if is_source_install:
+            readme_content = readme_content.replace("{APP_FILE}", "trackio/ui/main.py")
+        else:
+            readme_content = readme_content.replace("{APP_FILE}", "app.py")
+        readme_buffer = io.BytesIO(readme_content.encode("utf-8"))
+        hf_api.upload_file(
+            path_or_fileobj=readme_buffer,
+            path_in_repo="README.md",
+            repo_id=space_id,
+            repo_type="space",
+        )
+
+    if is_source_install:
+        requirements_content = _get_source_install_dependencies()
+    else:
+        requirements_content = f"trackio[spaces]=={trackio.__version__}"
+
+    requirements_buffer = io.BytesIO(requirements_content.encode("utf-8"))
+    hf_api.upload_file(
+        path_or_fileobj=requirements_buffer,
+        path_in_repo="requirements.txt",
+        repo_id=space_id,
+        repo_type="space",
+    )
+
+    huggingface_hub.utils.disable_progress_bars()
+
+    if is_source_install:
+        hf_api.upload_folder(
+            repo_id=space_id,
+            repo_type="space",
+            folder_path=trackio_path,
+            path_in_repo="trackio",
+            ignore_patterns=["README.md"],
+        )
+
+    app_file_content = """import trackio
+trackio.show()"""
+    app_file_buffer = io.BytesIO(app_file_content.encode("utf-8"))
+    hf_api.upload_file(
+        path_or_fileobj=app_file_buffer,
+        path_in_repo="app.py",
+        repo_id=space_id,
+        repo_type="space",
+    )
+
+    if hf_token := huggingface_hub.utils.get_token():
+        huggingface_hub.add_space_secret(space_id, "HF_TOKEN", hf_token)
+    if dataset_id is not None:
+        huggingface_hub.add_space_variable(space_id, "TRACKIO_DATASET_ID", dataset_id)
+    if logo_light_url := os.environ.get("TRACKIO_LOGO_LIGHT_URL"):
+        huggingface_hub.add_space_variable(
+            space_id, "TRACKIO_LOGO_LIGHT_URL", logo_light_url
+        )
+    if logo_dark_url := os.environ.get("TRACKIO_LOGO_DARK_URL"):
+        huggingface_hub.add_space_variable(
+            space_id, "TRACKIO_LOGO_DARK_URL", logo_dark_url
+        )
+    if plot_order := os.environ.get("TRACKIO_PLOT_ORDER"):
+        huggingface_hub.add_space_variable(space_id, "TRACKIO_PLOT_ORDER", plot_order)
+    if theme := os.environ.get("TRACKIO_THEME"):
+        huggingface_hub.add_space_variable(space_id, "TRACKIO_THEME", theme)
+    huggingface_hub.add_space_variable(space_id, "GRADIO_MCP_SERVER", "True")
+
+
+def create_space_if_not_exists(
+    space_id: str,
+    space_storage: huggingface_hub.SpaceStorage | None = None,
+    dataset_id: str | None = None,
+    private: bool | None = None,
+) -> None:
+    """
+    Creates a new Hugging Face Space if it does not exist.
+
+    Args:
+        space_id (`str`):
+            The ID of the Space to create.
+        space_storage ([`~huggingface_hub.SpaceStorage`], *optional*):
+            Choice of persistent storage tier for the Space.
+        dataset_id (`str`, *optional*):
+            The ID of the Dataset to add to the Space as a space variable.
+        private (`bool`, *optional*):
+            Whether to make the Space private. If `None` (default), the repo will be
+            public unless the organization's default is private. This value is ignored
+            if the repo already exists.
+    """
+    if "/" not in space_id:
+        raise ValueError(
+            f"Invalid space ID: {space_id}. Must be in the format: username/reponame or orgname/reponame."
+        )
+    if dataset_id is not None and "/" not in dataset_id:
+        raise ValueError(
+            f"Invalid dataset ID: {dataset_id}. Must be in the format: username/datasetname or orgname/datasetname."
+        )
+    try:
+        huggingface_hub.repo_info(space_id, repo_type="space")
+        print(f"* Found existing space: {SPACE_URL.format(space_id=space_id)}")
+        return
+    except RepositoryNotFoundError:
+        pass
+    except HfHubHTTPError as e:
+        if e.response.status_code in [401, 403]:  # unauthorized or forbidden
+            print("Need 'write' access token to create a Spaces repo.")
+            huggingface_hub.login(add_to_git_credential=False)
+        else:
+            raise ValueError(f"Failed to create Space: {e}")
+
+    print(f"* Creating new space: {SPACE_URL.format(space_id=space_id)}")
+    deploy_as_space(space_id, space_storage, dataset_id, private)
+
+
+def wait_until_space_exists(
+    space_id: str,
+) -> None:
+    """
+    Blocks the current thread until the Space exists.
+
+    Args:
+        space_id (`str`):
+            The ID of the Space to wait for.
+
+    Raises:
+        `TimeoutError`: If waiting for the Space takes longer than expected.
+    """
+    hf_api = huggingface_hub.HfApi()
+    delay = 1
+    for _ in range(30):
+        try:
+            hf_api.space_info(space_id)
+            return
+        except (huggingface_hub.utils.HfHubHTTPError, ReadTimeout):
+            time.sleep(delay)
+            delay = min(delay * 2, 60)
+    raise TimeoutError("Waiting for space to exist took longer than expected")
+
+
+def upload_db_to_space(project: str, space_id: str, force: bool = False) -> None:
+    """
+    Uploads the database of a local Trackio project to a Hugging Face Space.
+
+    This uses the Gradio Client to upload since we do not want to trigger a new build of
+    the Space, which would happen if we used `huggingface_hub.upload_file`.
+
+    Args:
+        project (`str`):
+            The name of the project to upload.
+        space_id (`str`):
+            The ID of the Space to upload to.
+        force (`bool`, *optional*, defaults to `False`):
+            If `True`, overwrites the existing database without prompting. If `False`,
+            prompts for confirmation.
+    """
+    db_path = SQLiteStorage.get_project_db_path(project)
+    client = Client(space_id, verbose=False, httpx_kwargs={"timeout": 90})
+
+    if not force:
+        try:
+            existing_projects = client.predict(api_name="/get_all_projects")
+            if project in existing_projects:
+                response = input(
+                    f"Database for project '{project}' already exists on Space '{space_id}'. "
+                    f"Overwrite it? (y/N): "
+                )
+                if response.lower() not in ["y", "yes"]:
+                    print("* Upload cancelled.")
+                    return
+        except Exception as e:
+            print(f"* Warning: Could not check if project exists on Space: {e}")
+            print("* Proceeding with upload...")
+
+    client.predict(
+        api_name="/upload_db_to_space",
+        project=project,
+        uploaded_db=handle_file(db_path),
+        hf_token=huggingface_hub.utils.get_token(),
+    )
+
+
+def sync(
+    project: str,
+    space_id: str | None = None,
+    private: bool | None = None,
+    force: bool = False,
+    run_in_background: bool = False,
+) -> str:
+    """
+    Syncs a local Trackio project's database to a Hugging Face Space.
+    If the Space does not exist, it will be created.
+
+    Args:
+        project (`str`): The name of the project to upload.
+        space_id (`str`, *optional*): The ID of the Space to upload to (e.g., `"username/space_id"`).
+            If not provided, a random space_id (e.g. "username/project-2ac3z2aA") will be used.
+        private (`bool`, *optional*):
+            Whether to make the Space private. If None (default), the repo will be
+            public unless the organization's default is private. This value is ignored
+            if the repo already exists.
+        force (`bool`, *optional*, defaults to `False`):
+            If `True`, overwrite the existing database without prompting for confirmation.
+            If `False`, prompt the user before overwriting an existing database.
+        run_in_background (`bool`, *optional*, defaults to `False`):
+            If `True`, the Space creation and database upload will be run in a background thread.
+            If `False`, all the steps will be run synchronously.
+    Returns:
+        `str`: The Space ID of the synced project.
+    """
+    if space_id is None:
+        space_id = f"{project}-{get_or_create_project_hash(project)}"
+    space_id, _ = preprocess_space_and_dataset_ids(space_id, None)
+
+    def space_creation_and_upload(
+        space_id: str, private: bool | None = None, force: bool = False
+    ):
+        print(
+            f"* Syncing local Trackio project to: {SPACE_URL.format(space_id=space_id)} (please wait...)"
+        )
+        create_space_if_not_exists(space_id, private=private)
+        wait_until_space_exists(space_id)
+        upload_db_to_space(project, space_id, force=force)
+        print(f"* Synced successfully to space: {SPACE_URL.format(space_id=space_id)}")
+
+    if run_in_background:
+        threading.Thread(
+            target=space_creation_and_upload, args=(space_id, private, force)
+        ).start()
+    else:
+        space_creation_and_upload(space_id, private, force)
+    return space_id

trackio/dummy_commit_scheduler.py

@@ -0,0 +1,12 @@
+# A dummy object to fit the interface of huggingface_hub's CommitScheduler
+class DummyCommitSchedulerLock:
+    def __enter__(self):
+        return None
+
+    def __exit__(self, exception_type, exception_value, exception_traceback):
+        pass
+
+
+class DummyCommitScheduler:
+    def __init__(self):
+        self.lock = DummyCommitSchedulerLock()

trackio/gpu.py

@@ -0,0 +1,368 @@
+import os
+import threading
+import warnings
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    from trackio.run import Run
+
+pynvml: Any = None
+PYNVML_AVAILABLE = False
+_nvml_initialized = False
+_nvml_lock = threading.Lock()
+_energy_baseline: dict[int, float] = {}
+
+
+def _ensure_pynvml():
+    global PYNVML_AVAILABLE, pynvml
+    if PYNVML_AVAILABLE:
+        return pynvml
+    try:
+        import pynvml as _pynvml
+
+        pynvml = _pynvml
+        PYNVML_AVAILABLE = True
+        return pynvml
+    except ImportError:
+        raise ImportError(
+            "nvidia-ml-py is required for GPU monitoring. "
+            "Install it with: pip install nvidia-ml-py"
+        )
+
+
+def _init_nvml() -> bool:
+    global _nvml_initialized
+    with _nvml_lock:
+        if _nvml_initialized:
+            return True
+        try:
+            nvml = _ensure_pynvml()
+            nvml.nvmlInit()
+            _nvml_initialized = True
+            return True
+        except Exception:
+            return False
+
+
+def _shutdown_nvml():
+    global _nvml_initialized
+    with _nvml_lock:
+        if _nvml_initialized and pynvml is not None:
+            try:
+                pynvml.nvmlShutdown()
+            except Exception:
+                pass
+            _nvml_initialized = False
+
+
+def get_gpu_count() -> tuple[int, list[int]]:
+    """
+    Get the number of GPUs visible to this process and their physical indices.
+    Respects CUDA_VISIBLE_DEVICES environment variable.
+
+    Returns:
+        Tuple of (count, physical_indices) where:
+        - count: Number of visible GPUs
+        - physical_indices: List mapping logical index to physical GPU index.
+          e.g., if CUDA_VISIBLE_DEVICES=2,3 returns (2, [2, 3])
+          meaning logical GPU 0 = physical GPU 2, logical GPU 1 = physical GPU 3
+    """
+    if not _init_nvml():
+        return 0, []
+
+    cuda_visible = os.environ.get("CUDA_VISIBLE_DEVICES")
+    if cuda_visible is not None and cuda_visible.strip():
+        try:
+            indices = [int(x.strip()) for x in cuda_visible.split(",") if x.strip()]
+            return len(indices), indices
+        except ValueError:
+            pass
+
+    try:
+        total = pynvml.nvmlDeviceGetCount()
+        return total, list(range(total))
+    except Exception:
+        return 0, []
+
+
+def gpu_available() -> bool:
+    """
+    Check if GPU monitoring is available.
+
+    Returns True if nvidia-ml-py is installed and at least one NVIDIA GPU is detected.
+    This is used for auto-detection of GPU logging.
+    """
+    try:
+        _ensure_pynvml()
+        count, _ = get_gpu_count()
+        return count > 0
+    except ImportError:
+        return False
+    except Exception:
+        return False
+
+
+def reset_energy_baseline():
+    """Reset the energy baseline for all GPUs. Called when a new run starts."""
+    global _energy_baseline
+    _energy_baseline = {}
+
+
+def collect_gpu_metrics(device: int | None = None) -> dict:
+    """
+    Collect GPU metrics for visible GPUs.
+
+    Args:
+        device: CUDA device index to collect metrics from. If None, collects
+                from all GPUs visible to this process (respects CUDA_VISIBLE_DEVICES).
+                The device index is the logical CUDA index (0, 1, 2...), not the
+                physical GPU index.
+
+    Returns:
+        Dictionary of GPU metrics. Keys use logical device indices (gpu/0/, gpu/1/, etc.)
+        which correspond to CUDA device indices, not physical GPU indices.
+    """
+    if not _init_nvml():
+        return {}
+
+    gpu_count, visible_gpus = get_gpu_count()
+    if gpu_count == 0:
+        return {}
+
+    if device is not None:
+        if device < 0 or device >= gpu_count:
+            return {}
+        gpu_indices = [(device, visible_gpus[device])]
+    else:
+        gpu_indices = list(enumerate(visible_gpus))
+
+    metrics = {}
+    total_util = 0.0
+    total_mem_used_gib = 0.0
+    total_power = 0.0
+    max_temp = 0.0
+    valid_util_count = 0
+
+    for logical_idx, physical_idx in gpu_indices:
+        prefix = f"gpu/{logical_idx}"
+        try:
+            handle = pynvml.nvmlDeviceGetHandleByIndex(physical_idx)
+
+            try:
+                util = pynvml.nvmlDeviceGetUtilizationRates(handle)
+                metrics[f"{prefix}/utilization"] = util.gpu
+                metrics[f"{prefix}/memory_utilization"] = util.memory
+                total_util += util.gpu
+                valid_util_count += 1
+            except Exception:
+                pass
+
+            try:
+                mem = pynvml.nvmlDeviceGetMemoryInfo(handle)
+                mem_used_gib = mem.used / (1024**3)
+                mem_total_gib = mem.total / (1024**3)
+                metrics[f"{prefix}/allocated_memory"] = mem_used_gib
+                metrics[f"{prefix}/total_memory"] = mem_total_gib
+                if mem.total > 0:
+                    metrics[f"{prefix}/memory_usage"] = mem.used / mem.total
+                total_mem_used_gib += mem_used_gib
+            except Exception:
+                pass
+
+            try:
+                power_mw = pynvml.nvmlDeviceGetPowerUsage(handle)
+                power_w = power_mw / 1000.0
+                metrics[f"{prefix}/power"] = power_w
+                total_power += power_w
+            except Exception:
+                pass
+
+            try:
+                power_limit_mw = pynvml.nvmlDeviceGetPowerManagementLimit(handle)
+                power_limit_w = power_limit_mw / 1000.0
+                metrics[f"{prefix}/power_limit"] = power_limit_w
+                if power_limit_w > 0 and f"{prefix}/power" in metrics:
+                    metrics[f"{prefix}/power_percent"] = (
+                        metrics[f"{prefix}/power"] / power_limit_w
+                    ) * 100
+            except Exception:
+                pass
+
+            try:
+                temp = pynvml.nvmlDeviceGetTemperature(
+                    handle, pynvml.NVML_TEMPERATURE_GPU
+                )
+                metrics[f"{prefix}/temp"] = temp
+                max_temp = max(max_temp, temp)
+            except Exception:
+                pass
+
+            try:
+                sm_clock = pynvml.nvmlDeviceGetClockInfo(handle, pynvml.NVML_CLOCK_SM)
+                metrics[f"{prefix}/sm_clock"] = sm_clock
+            except Exception:
+                pass
+
+            try:
+                mem_clock = pynvml.nvmlDeviceGetClockInfo(handle, pynvml.NVML_CLOCK_MEM)
+                metrics[f"{prefix}/memory_clock"] = mem_clock
+            except Exception:
+                pass
+
+            try:
+                fan_speed = pynvml.nvmlDeviceGetFanSpeed(handle)
+                metrics[f"{prefix}/fan_speed"] = fan_speed
+            except Exception:
+                pass
+
+            try:
+                pstate = pynvml.nvmlDeviceGetPerformanceState(handle)
+                metrics[f"{prefix}/performance_state"] = pstate
+            except Exception:
+                pass
+
+            try:
+                energy_mj = pynvml.nvmlDeviceGetTotalEnergyConsumption(handle)
+                if logical_idx not in _energy_baseline:
+                    _energy_baseline[logical_idx] = energy_mj
+                energy_consumed_mj = energy_mj - _energy_baseline[logical_idx]
+                metrics[f"{prefix}/energy_consumed"] = energy_consumed_mj / 1000.0
+            except Exception:
+                pass
+
+            try:
+                pcie_tx = pynvml.nvmlDeviceGetPcieThroughput(
+                    handle, pynvml.NVML_PCIE_UTIL_TX_BYTES
+                )
+                pcie_rx = pynvml.nvmlDeviceGetPcieThroughput(
+                    handle, pynvml.NVML_PCIE_UTIL_RX_BYTES
+                )
+                metrics[f"{prefix}/pcie_tx"] = pcie_tx / 1024.0
+                metrics[f"{prefix}/pcie_rx"] = pcie_rx / 1024.0
+            except Exception:
+                pass
+
+            try:
+                throttle = pynvml.nvmlDeviceGetCurrentClocksThrottleReasons(handle)
+                metrics[f"{prefix}/throttle_thermal"] = int(
+                    bool(throttle & pynvml.nvmlClocksThrottleReasonSwThermalSlowdown)
+                )
+                metrics[f"{prefix}/throttle_power"] = int(
+                    bool(throttle & pynvml.nvmlClocksThrottleReasonSwPowerCap)
+                )
+                metrics[f"{prefix}/throttle_hw_slowdown"] = int(
+                    bool(throttle & pynvml.nvmlClocksThrottleReasonHwSlowdown)
+                )
+                metrics[f"{prefix}/throttle_apps"] = int(
+                    bool(
+                        throttle
+                        & pynvml.nvmlClocksThrottleReasonApplicationsClocksSetting
+                    )
+                )
+            except Exception:
+                pass
+
+            try:
+                ecc_corrected = pynvml.nvmlDeviceGetTotalEccErrors(
+                    handle,
+                    pynvml.NVML_MEMORY_ERROR_TYPE_CORRECTED,
+                    pynvml.NVML_VOLATILE_ECC,
+                )
+                metrics[f"{prefix}/corrected_memory_errors"] = ecc_corrected
+            except Exception:
+                pass
+
+            try:
+                ecc_uncorrected = pynvml.nvmlDeviceGetTotalEccErrors(
+                    handle,
+                    pynvml.NVML_MEMORY_ERROR_TYPE_UNCORRECTED,
+                    pynvml.NVML_VOLATILE_ECC,
+                )
+                metrics[f"{prefix}/uncorrected_memory_errors"] = ecc_uncorrected
+            except Exception:
+                pass
+
+        except Exception:
+            continue
+
+    if valid_util_count > 0:
+        metrics["gpu/mean_utilization"] = total_util / valid_util_count
+    if total_mem_used_gib > 0:
+        metrics["gpu/total_allocated_memory"] = total_mem_used_gib
+    if total_power > 0:
+        metrics["gpu/total_power"] = total_power
+    if max_temp > 0:
+        metrics["gpu/max_temp"] = max_temp
+
+    return metrics
+
+
+class GpuMonitor:
+    def __init__(self, run: "Run", interval: float = 10.0):
+        self._run = run
+        self._interval = interval
+        self._stop_flag = threading.Event()
+        self._thread: "threading.Thread | None" = None
+
+    def start(self):
+        count, _ = get_gpu_count()
+        if count == 0:
+            warnings.warn(
+                "auto_log_gpu=True but no NVIDIA GPUs detected. GPU logging disabled."
+            )
+            return
+
+        reset_energy_baseline()
+        self._thread = threading.Thread(target=self._monitor_loop, daemon=True)
+        self._thread.start()
+
+    def stop(self):
+        self._stop_flag.set()
+        if self._thread is not None:
+            self._thread.join(timeout=2.0)
+
+    def _monitor_loop(self):
+        while not self._stop_flag.is_set():
+            try:
+                metrics = collect_gpu_metrics()
+                if metrics:
+                    self._run.log_system(metrics)
+            except Exception:
+                pass
+
+            self._stop_flag.wait(timeout=self._interval)
+
+
+def log_gpu(run: "Run | None" = None, device: int | None = None) -> dict:
+    """
+    Log GPU metrics to the current or specified run as system metrics.
+
+    Args:
+        run: Optional Run instance. If None, uses current run from context.
+        device: CUDA device index to collect metrics from. If None, collects
+                from all GPUs visible to this process (respects CUDA_VISIBLE_DEVICES).
+
+    Returns:
+        dict: The GPU metrics that were logged.
+
+    Example:
+        ```python
+        import trackio
+
+        run = trackio.init(project="my-project")
+        trackio.log({"loss": 0.5})
+        trackio.log_gpu()  # logs all visible GPUs
+        trackio.log_gpu(device=0)  # logs only CUDA device 0
+        ```
+    """
+    from trackio import context_vars
+
+    if run is None:
+        run = context_vars.current_run.get()
+        if run is None:
+            raise RuntimeError("Call trackio.init() before trackio.log_gpu().")
+
+    metrics = collect_gpu_metrics(device=device)
+    if metrics:
+        run.log_system(metrics)
+    return metrics

trackio/histogram.py

@@ -0,0 +1,71 @@
+from typing import Sequence
+
+import numpy as np
+
+
+class Histogram:
+    """
+    Histogram data type for Trackio, compatible with wandb.Histogram.
+
+    Args:
+        sequence (`np.ndarray` or `Sequence[float]` or `Sequence[int]`, *optional*):
+            Sequence of values to create the histogram from.
+        np_histogram (`tuple`, *optional*):
+            Pre-computed NumPy histogram as a `(hist, bins)` tuple.
+        num_bins (`int`, *optional*, defaults to `64`):
+            Number of bins for the histogram (maximum `512`).
+
+    Example:
+        ```python
+        import trackio
+        import numpy as np
+
+        # Create histogram from sequence
+        data = np.random.randn(1000)
+        trackio.log({"distribution": trackio.Histogram(data)})
+
+        # Create histogram from numpy histogram
+        hist, bins = np.histogram(data, bins=30)
+        trackio.log({"distribution": trackio.Histogram(np_histogram=(hist, bins))})
+
+        # Specify custom number of bins
+        trackio.log({"distribution": trackio.Histogram(data, num_bins=50)})
+        ```
+    """
+
+    TYPE = "trackio.histogram"
+
+    def __init__(
+        self,
+        sequence: np.ndarray | Sequence[float] | Sequence[int] | None = None,
+        np_histogram: tuple | None = None,
+        num_bins: int = 64,
+    ):
+        if sequence is None and np_histogram is None:
+            raise ValueError("Must provide either sequence or np_histogram")
+
+        if sequence is not None and np_histogram is not None:
+            raise ValueError("Cannot provide both sequence and np_histogram")
+
+        num_bins = min(num_bins, 512)
+
+        if np_histogram is not None:
+            self.histogram, self.bins = np_histogram
+            self.histogram = np.asarray(self.histogram)
+            self.bins = np.asarray(self.bins)
+        else:
+            data = np.asarray(sequence).flatten()
+            data = data[np.isfinite(data)]
+            if len(data) == 0:
+                self.histogram = np.array([])
+                self.bins = np.array([])
+            else:
+                self.histogram, self.bins = np.histogram(data, bins=num_bins)
+
+    def _to_dict(self) -> dict:
+        """Convert histogram to dictionary for storage."""
+        return {
+            "_type": self.TYPE,
+            "bins": self.bins.tolist(),
+            "values": self.histogram.tolist(),
+        }

trackio/imports.py

@@ -0,0 +1,304 @@
+import os
+from pathlib import Path
+
+import pandas as pd
+
+from trackio import deploy, utils
+from trackio.sqlite_storage import SQLiteStorage
+
+
+def import_csv(
+    csv_path: str | Path,
+    project: str,
+    name: str | None = None,
+    space_id: str | None = None,
+    dataset_id: str | None = None,
+    private: bool | None = None,
+    force: bool = False,
+) -> None:
+    """
+    Imports a CSV file into a Trackio project. The CSV file must contain a `"step"`
+    column, may optionally contain a `"timestamp"` column, and any other columns will be
+    treated as metrics. It should also include a header row with the column names.
+
+    TODO: call init() and return a Run object so that the user can continue to log metrics to it.
+
+    Args:
+        csv_path (`str` or `Path`):
+            The str or Path to the CSV file to import.
+        project (`str`):
+            The name of the project to import the CSV file into. Must not be an existing
+            project.
+        name (`str`, *optional*):
+            The name of the Run to import the CSV file into. If not provided, a default
+            name will be generated.
+        name (`str`, *optional*):
+            The name of the run (if not provided, a default name will be generated).
+        space_id (`str`, *optional*):
+            If provided, the project will be logged to a Hugging Face Space instead of a
+            local directory. Should be a complete Space name like `"username/reponame"`
+            or `"orgname/reponame"`, or just `"reponame"` in which case the Space will
+            be created in the currently-logged-in Hugging Face user's namespace. If the
+            Space does not exist, it will be created. If the Space already exists, the
+            project will be logged to it.
+        dataset_id (`str`, *optional*):
+            If provided, a persistent Hugging Face Dataset will be created and the
+            metrics will be synced to it every 5 minutes. Should be a complete Dataset
+            name like `"username/datasetname"` or `"orgname/datasetname"`, or just
+            `"datasetname"` in which case the Dataset will be created in the
+            currently-logged-in Hugging Face user's namespace. If the Dataset does not
+            exist, it will be created. If the Dataset already exists, the project will
+            be appended to it. If not provided, the metrics will be logged to a local
+            SQLite database, unless a `space_id` is provided, in which case a Dataset
+            will be automatically created with the same name as the Space but with the
+            `"_dataset"` suffix.
+        private (`bool`, *optional*):
+            Whether to make the Space private. If None (default), the repo will be
+            public unless the organization's default is private. This value is ignored
+            if the repo already exists.
+    """
+    if SQLiteStorage.get_runs(project):
+        raise ValueError(
+            f"Project '{project}' already exists. Cannot import CSV into existing project."
+        )
+
+    csv_path = Path(csv_path)
+    if not csv_path.exists():
+        raise FileNotFoundError(f"CSV file not found: {csv_path}")
+
+    df = pd.read_csv(csv_path)
+    if df.empty:
+        raise ValueError("CSV file is empty")
+
+    column_mapping = utils.simplify_column_names(df.columns.tolist())
+    df = df.rename(columns=column_mapping)
+
+    step_column = None
+    for col in df.columns:
+        if col.lower() == "step":
+            step_column = col
+            break
+
+    if step_column is None:
+        raise ValueError("CSV file must contain a 'step' or 'Step' column")
+
+    if name is None:
+        name = csv_path.stem
+
+    metrics_list = []
+    steps = []
+    timestamps = []
+
+    numeric_columns = []
+    for column in df.columns:
+        if column == step_column:
+            continue
+        if column == "timestamp":
+            continue
+
+        try:
+            pd.to_numeric(df[column], errors="raise")
+            numeric_columns.append(column)
+        except (ValueError, TypeError):
+            continue
+
+    for _, row in df.iterrows():
+        metrics = {}
+        for column in numeric_columns:
+            value = row[column]
+            if bool(pd.notna(value)):
+                metrics[column] = float(value)
+
+        if metrics:
+            metrics_list.append(metrics)
+            steps.append(int(row[step_column]))
+
+            if "timestamp" in df.columns and bool(pd.notna(row["timestamp"])):
+                timestamps.append(str(row["timestamp"]))
+            else:
+                timestamps.append("")
+
+    if metrics_list:
+        SQLiteStorage.bulk_log(
+            project=project,
+            run=name,
+            metrics_list=metrics_list,
+            steps=steps,
+            timestamps=timestamps,
+        )
+
+    print(
+        f"* Imported {len(metrics_list)} rows from {csv_path} into project '{project}' as run '{name}'"
+    )
+    print(f"* Metrics found: {', '.join(metrics_list[0].keys())}")
+
+    space_id, dataset_id = utils.preprocess_space_and_dataset_ids(space_id, dataset_id)
+    if dataset_id is not None:
+        os.environ["TRACKIO_DATASET_ID"] = dataset_id
+        print(f"* Trackio metrics will be synced to Hugging Face Dataset: {dataset_id}")
+
+    if space_id is None:
+        utils.print_dashboard_instructions(project)
+    else:
+        deploy.create_space_if_not_exists(
+            space_id=space_id, dataset_id=dataset_id, private=private
+        )
+        deploy.wait_until_space_exists(space_id=space_id)
+        deploy.upload_db_to_space(project=project, space_id=space_id, force=force)
+        print(
+            f"* View dashboard by going to: {deploy.SPACE_URL.format(space_id=space_id)}"
+        )
+
+
+def import_tf_events(
+    log_dir: str | Path,
+    project: str,
+    name: str | None = None,
+    space_id: str | None = None,
+    dataset_id: str | None = None,
+    private: bool | None = None,
+    force: bool = False,
+) -> None:
+    """
+    Imports TensorFlow Events files from a directory into a Trackio project. Each
+    subdirectory in the log directory will be imported as a separate run.
+
+    Args:
+        log_dir (`str` or `Path`):
+            The str or Path to the directory containing TensorFlow Events files.
+        project (`str`):
+            The name of the project to import the TensorFlow Events files into. Must not
+            be an existing project.
+        name (`str`, *optional*):
+            The name prefix for runs (if not provided, will use directory names). Each
+            subdirectory will create a separate run.
+        space_id (`str`, *optional*):
+            If provided, the project will be logged to a Hugging Face Space instead of a
+            local directory. Should be a complete Space name like `"username/reponame"`
+            or `"orgname/reponame"`, or just `"reponame"` in which case the Space will
+            be created in the currently-logged-in Hugging Face user's namespace. If the
+            Space does not exist, it will be created. If the Space already exists, the
+            project will be logged to it.
+        dataset_id (`str`, *optional*):
+            If provided, a persistent Hugging Face Dataset will be created and the
+            metrics will be synced to it every 5 minutes. Should be a complete Dataset
+            name like `"username/datasetname"` or `"orgname/datasetname"`, or just
+            `"datasetname"` in which case the Dataset will be created in the
+            currently-logged-in Hugging Face user's namespace. If the Dataset does not
+            exist, it will be created. If the Dataset already exists, the project will
+            be appended to it. If not provided, the metrics will be logged to a local
+            SQLite database, unless a `space_id` is provided, in which case a Dataset
+            will be automatically created with the same name as the Space but with the
+            `"_dataset"` suffix.
+        private (`bool`, *optional*):
+            Whether to make the Space private. If None (default), the repo will be
+            public unless the organization's default is private. This value is ignored
+            if the repo already exists.
+    """
+    try:
+        from tbparse import SummaryReader
+    except ImportError:
+        raise ImportError(
+            "The `tbparse` package is not installed but is required for `import_tf_events`. Please install trackio with the `tensorboard` extra: `pip install trackio[tensorboard]`."
+        )
+
+    if SQLiteStorage.get_runs(project):
+        raise ValueError(
+            f"Project '{project}' already exists. Cannot import TF events into existing project."
+        )
+
+    path = Path(log_dir)
+    if not path.exists():
+        raise FileNotFoundError(f"TF events directory not found: {path}")
+
+    # Use tbparse to read all tfevents files in the directory structure
+    reader = SummaryReader(str(path), extra_columns={"dir_name"})
+    df = reader.scalars
+
+    if df.empty:
+        raise ValueError(f"No TensorFlow events data found in {path}")
+
+    total_imported = 0
+    imported_runs = []
+
+    # Group by dir_name to create separate runs
+    for dir_name, group_df in df.groupby("dir_name"):
+        try:
+            # Determine run name based on directory name
+            if dir_name == "":
+                run_name = "main"  # For files in the root directory
+            else:
+                run_name = dir_name  # Use directory name
+
+            if name:
+                run_name = f"{name}_{run_name}"
+
+            if group_df.empty:
+                print(f"* Skipping directory {dir_name}: no scalar data found")
+                continue
+
+            metrics_list = []
+            steps = []
+            timestamps = []
+
+            for _, row in group_df.iterrows():
+                # Convert row values to appropriate types
+                tag = str(row["tag"])
+                value = float(row["value"])
+                step = int(row["step"])
+
+                metrics = {tag: value}
+                metrics_list.append(metrics)
+                steps.append(step)
+
+                # Use wall_time if present, else fallback
+                if "wall_time" in group_df.columns and not bool(
+                    pd.isna(row["wall_time"])
+                ):
+                    timestamps.append(str(row["wall_time"]))
+                else:
+                    timestamps.append("")
+
+            if metrics_list:
+                SQLiteStorage.bulk_log(
+                    project=project,
+                    run=str(run_name),
+                    metrics_list=metrics_list,
+                    steps=steps,
+                    timestamps=timestamps,
+                )
+
+                total_imported += len(metrics_list)
+                imported_runs.append(run_name)
+
+                print(
+                    f"* Imported {len(metrics_list)} scalar events from directory '{dir_name}' as run '{run_name}'"
+                )
+                print(f"* Metrics in this run: {', '.join(set(group_df['tag']))}")
+
+        except Exception as e:
+            print(f"* Error processing directory {dir_name}: {e}")
+            continue
+
+    if not imported_runs:
+        raise ValueError("No valid TensorFlow events data could be imported")
+
+    print(f"* Total imported events: {total_imported}")
+    print(f"* Created runs: {', '.join(imported_runs)}")
+
+    space_id, dataset_id = utils.preprocess_space_and_dataset_ids(space_id, dataset_id)
+    if dataset_id is not None:
+        os.environ["TRACKIO_DATASET_ID"] = dataset_id
+        print(f"* Trackio metrics will be synced to Hugging Face Dataset: {dataset_id}")
+
+    if space_id is None:
+        utils.print_dashboard_instructions(project)
+    else:
+        deploy.create_space_if_not_exists(
+            space_id, dataset_id=dataset_id, private=private
+        )
+        deploy.wait_until_space_exists(space_id)
+        deploy.upload_db_to_space(project, space_id, force=force)
+        print(
+            f"* View dashboard by going to: {deploy.SPACE_URL.format(space_id=space_id)}"
+        )

trackio/media/__init__.py

@@ -0,0 +1,27 @@
+"""
+Media module for Trackio.
+
+This module contains all media-related functionality including:
+- TrackioImage, TrackioVideo, TrackioAudio classes
+- Video writing utilities
+- Audio conversion utilities
+"""
+
+from trackio.media.audio import TrackioAudio
+from trackio.media.image import TrackioImage
+from trackio.media.media import TrackioMedia
+from trackio.media.utils import get_project_media_path
+from trackio.media.video import TrackioVideo
+
+write_audio = TrackioAudio.write_audio
+write_video = TrackioVideo.write_video
+
+__all__ = [
+    "TrackioMedia",
+    "TrackioImage",
+    "TrackioVideo",
+    "TrackioAudio",
+    "get_project_media_path",
+    "write_video",
+    "write_audio",
+]

trackio/media/audio.py

@@ -0,0 +1,167 @@
+import os
+import shutil
+import warnings
+from pathlib import Path
+from typing import Literal
+
+import numpy as np
+from pydub import AudioSegment
+
+from trackio.media.media import TrackioMedia
+from trackio.media.utils import check_ffmpeg_installed, check_path
+
+SUPPORTED_FORMATS = ["wav", "mp3"]
+AudioFormatType = Literal["wav", "mp3"]
+TrackioAudioSourceType = str | Path | np.ndarray
+
+
+class TrackioAudio(TrackioMedia):
+    """
+    Initializes an Audio object.
+
+    Example:
+        ```python
+        import trackio
+        import numpy as np
+
+        # Generate a 1-second 440 Hz sine wave (mono)
+        sr = 16000
+        t = np.linspace(0, 1, sr, endpoint=False)
+        wave = 0.2 * np.sin(2 * np.pi * 440 * t)
+        audio = trackio.Audio(wave, caption="A4 sine", sample_rate=sr, format="wav")
+        trackio.log({"tone": audio})
+
+        # Stereo from numpy array (shape: samples, 2)
+        stereo = np.stack([wave, wave], axis=1)
+        audio = trackio.Audio(stereo, caption="Stereo", sample_rate=sr, format="mp3")
+        trackio.log({"stereo": audio})
+
+        # From an existing file
+        audio = trackio.Audio("path/to/audio.wav", caption="From file")
+        trackio.log({"file_audio": audio})
+        ```
+
+    Args:
+        value (`str`, `Path`, or `numpy.ndarray`, *optional*):
+            A path to an audio file, or a numpy array.
+            The array should be shaped `(samples,)` for mono or `(samples, 2)` for stereo.
+            Float arrays will be peak-normalized and converted to 16-bit PCM; integer arrays will be converted to 16-bit PCM as needed.
+        caption (`str`, *optional*):
+            A string caption for the audio.
+        sample_rate (`int`, *optional*):
+            Sample rate in Hz. Required when `value` is a numpy array.
+        format (`Literal["wav", "mp3"]`, *optional*):
+            Audio format used when `value` is a numpy array. Default is "wav".
+    """
+
+    TYPE = "trackio.audio"
+
+    def __init__(
+        self,
+        value: TrackioAudioSourceType,
+        caption: str | None = None,
+        sample_rate: int | None = None,
+        format: AudioFormatType | None = None,
+    ):
+        super().__init__(value, caption)
+        if isinstance(value, np.ndarray):
+            if sample_rate is None:
+                raise ValueError("Sample rate is required when value is an ndarray")
+            if format is None:
+                format = "wav"
+        self._format = format
+        self._sample_rate = sample_rate
+
+    def _save_media(self, file_path: Path):
+        if isinstance(self._value, np.ndarray):
+            TrackioAudio.write_audio(
+                data=self._value,
+                sample_rate=self._sample_rate,
+                filename=file_path,
+                format=self._format,
+            )
+        elif isinstance(self._value, str | Path):
+            if os.path.isfile(self._value):
+                shutil.copy(self._value, file_path)
+            else:
+                raise ValueError(f"File not found: {self._value}")
+
+    @staticmethod
+    def ensure_int16_pcm(data: np.ndarray) -> np.ndarray:
+        """
+        Convert input audio array to contiguous int16 PCM.
+        Peak normalization is applied to floating inputs.
+        """
+        arr = np.asarray(data)
+        if arr.ndim not in (1, 2):
+            raise ValueError("Audio data must be 1D (mono) or 2D ([samples, channels])")
+
+        if arr.dtype != np.int16:
+            warnings.warn(
+                f"Converting {arr.dtype} audio to int16 PCM; pass int16 to avoid conversion.",
+                stacklevel=2,
+            )
+
+        arr = np.nan_to_num(arr, copy=False)
+
+        # Floating types: normalize to peak 1.0, then scale to int16
+        if np.issubdtype(arr.dtype, np.floating):
+            max_abs = float(np.max(np.abs(arr))) if arr.size else 0.0
+            if max_abs > 0.0:
+                arr = arr / max_abs
+            out = (arr * 32767.0).clip(-32768, 32767).astype(np.int16, copy=False)
+            return np.ascontiguousarray(out)
+
+        converters: dict[np.dtype, callable] = {
+            np.dtype(np.int16): lambda a: a,
+            np.dtype(np.int32): lambda a: (
+                (a.astype(np.int32) // 65536).astype(np.int16, copy=False)
+            ),
+            np.dtype(np.uint16): lambda a: (
+                (a.astype(np.int32) - 32768).astype(np.int16, copy=False)
+            ),
+            np.dtype(np.uint8): lambda a: (
+                (a.astype(np.int32) * 257 - 32768).astype(np.int16, copy=False)
+            ),
+            np.dtype(np.int8): lambda a: (
+                (a.astype(np.int32) * 256).astype(np.int16, copy=False)
+            ),
+        }
+
+        conv = converters.get(arr.dtype)
+        if conv is not None:
+            out = conv(arr)
+            return np.ascontiguousarray(out)
+        raise TypeError(f"Unsupported audio dtype: {arr.dtype}")
+
+    @staticmethod
+    def write_audio(
+        data: np.ndarray,
+        sample_rate: int,
+        filename: str | Path,
+        format: AudioFormatType = "wav",
+    ) -> None:
+        if not isinstance(sample_rate, int) or sample_rate <= 0:
+            raise ValueError(f"Invalid sample_rate: {sample_rate}")
+        if format not in SUPPORTED_FORMATS:
+            raise ValueError(
+                f"Unsupported format: {format}. Supported: {SUPPORTED_FORMATS}"
+            )
+
+        check_path(filename)
+
+        pcm = TrackioAudio.ensure_int16_pcm(data)
+
+        if format != "wav":
+            check_ffmpeg_installed()
+
+        channels = 1 if pcm.ndim == 1 else pcm.shape[1]
+        audio = AudioSegment(
+            pcm.tobytes(),
+            frame_rate=sample_rate,
+            sample_width=2,  # int16
+            channels=channels,
+        )
+
+        file = audio.export(str(filename), format=format)
+        file.close()

trackio/media/image.py

@@ -0,0 +1,84 @@
+import os
+import shutil
+from pathlib import Path
+
+import numpy as np
+from PIL import Image as PILImage
+
+from trackio.media.media import TrackioMedia
+
+TrackioImageSourceType = str | Path | np.ndarray | PILImage.Image
+
+
+class TrackioImage(TrackioMedia):
+    """
+    Initializes an Image object.
+
+    Example:
+        ```python
+        import trackio
+        import numpy as np
+        from PIL import Image
+
+        # Create an image from numpy array
+        image_data = np.random.randint(0, 255, (64, 64, 3), dtype=np.uint8)
+        image = trackio.Image(image_data, caption="Random image")
+        trackio.log({"my_image": image})
+
+        # Create an image from PIL Image
+        pil_image = Image.new('RGB', (100, 100), color='red')
+        image = trackio.Image(pil_image, caption="Red square")
+        trackio.log({"red_image": image})
+
+        # Create an image from file path
+        image = trackio.Image("path/to/image.jpg", caption="Photo from file")
+        trackio.log({"file_image": image})
+        ```
+
+    Args:
+        value (`str`, `Path`, `numpy.ndarray`, or `PIL.Image`, *optional*):
+            A path to an image, a PIL Image, or a numpy array of shape (height, width, channels).
+            If numpy array, should be of type `np.uint8` with RGB values in the range `[0, 255]`.
+        caption (`str`, *optional*):
+            A string caption for the image.
+    """
+
+    TYPE = "trackio.image"
+
+    def __init__(self, value: TrackioImageSourceType, caption: str | None = None):
+        super().__init__(value, caption)
+        self._format: str | None = None
+
+        if not isinstance(self._value, TrackioImageSourceType):
+            raise ValueError(
+                f"Invalid value type, expected {TrackioImageSourceType}, got {type(self._value)}"
+            )
+        if isinstance(self._value, np.ndarray) and self._value.dtype != np.uint8:
+            raise ValueError(
+                f"Invalid value dtype, expected np.uint8, got {self._value.dtype}"
+            )
+        if (
+            isinstance(self._value, np.ndarray | PILImage.Image)
+            and self._format is None
+        ):
+            self._format = "png"
+
+    def _as_pil(self) -> PILImage.Image | None:
+        try:
+            if isinstance(self._value, np.ndarray):
+                arr = np.asarray(self._value).astype("uint8")
+                return PILImage.fromarray(arr).convert("RGBA")
+            if isinstance(self._value, PILImage.Image):
+                return self._value.convert("RGBA")
+        except Exception as e:
+            raise ValueError(f"Failed to process image data: {self._value}") from e
+        return None
+
+    def _save_media(self, file_path: Path):
+        if pil := self._as_pil():
+            pil.save(file_path, format=self._format)
+        elif isinstance(self._value, str | Path):
+            if os.path.isfile(self._value):
+                shutil.copy(self._value, file_path)
+            else:
+                raise ValueError(f"File not found: {self._value}")

trackio/media/media.py

@@ -0,0 +1,79 @@
+import os
+import uuid
+from abc import ABC, abstractmethod
+from pathlib import Path
+
+from trackio.media.utils import get_project_media_path
+from trackio.utils import MEDIA_DIR
+
+
+class TrackioMedia(ABC):
+    """
+    Abstract base class for Trackio media objects
+    Provides shared functionality for file handling and serialization.
+    """
+
+    TYPE: str
+
+    def __init_subclass__(cls, **kwargs):
+        """Ensure subclasses define the TYPE attribute."""
+        super().__init_subclass__(**kwargs)
+        if not hasattr(cls, "TYPE") or cls.TYPE is None:
+            raise TypeError(f"Class {cls.__name__} must define TYPE attribute")
+
+    def __init__(self, value, caption: str | None = None):
+        """
+        Saves the value and caption, and if the value is a file path, checks if the file exists.
+        """
+        self.caption = caption
+        self._value = value
+        self._file_path: Path | None = None
+
+        if isinstance(self._value, str | Path):
+            if not os.path.isfile(self._value):
+                raise ValueError(f"File not found: {self._value}")
+
+    def _file_extension(self) -> str:
+        if self._file_path:
+            return self._file_path.suffix[1:].lower()
+        if isinstance(self._value, str | Path):
+            path = Path(self._value)
+            return path.suffix[1:].lower()
+        if hasattr(self, "_format") and self._format:
+            return self._format
+        return "unknown"
+
+    def _get_relative_file_path(self) -> Path | None:
+        return self._file_path
+
+    def _get_absolute_file_path(self) -> Path | None:
+        if self._file_path:
+            return MEDIA_DIR / self._file_path
+        return None
+
+    def _save(self, project: str, run: str, step: int = 0):
+        if self._file_path:
+            return
+
+        media_dir = get_project_media_path(project=project, run=run, step=step)
+        filename = f"{uuid.uuid4()}.{self._file_extension()}"
+        file_path = media_dir / filename
+
+        self._save_media(file_path)
+        self._file_path = file_path.relative_to(MEDIA_DIR)
+
+    @abstractmethod
+    def _save_media(self, file_path: Path):
+        """
+        Performs the actual media saving logic.
+        """
+        pass
+
+    def _to_dict(self) -> dict:
+        if not self._file_path:
+            raise ValueError("Media must be saved to file before serialization")
+        return {
+            "_type": self.TYPE,
+            "file_path": str(self._get_relative_file_path()),
+            "caption": self.caption,
+        }

trackio/media/utils.py

@@ -0,0 +1,60 @@
+import shutil
+from pathlib import Path
+
+from trackio.utils import MEDIA_DIR
+
+
+def check_path(file_path: str | Path) -> None:
+    """Raise an error if the parent directory does not exist."""
+    file_path = Path(file_path)
+    if not file_path.parent.exists():
+        try:
+            file_path.parent.mkdir(parents=True, exist_ok=True)
+        except OSError as e:
+            raise ValueError(
+                f"Failed to create parent directory {file_path.parent}: {e}"
+            )
+
+
+def check_ffmpeg_installed() -> None:
+    """Raise an error if ffmpeg is not available on the system PATH."""
+    if shutil.which("ffmpeg") is None:
+        raise RuntimeError(
+            "ffmpeg is required to write video but was not found on your system. "
+            "Please install ffmpeg and ensure it is available on your PATH."
+        )
+
+
+def get_project_media_path(
+    project: str,
+    run: str | None = None,
+    step: int | None = None,
+    relative_path: str | Path | None = None,
+) -> Path:
+    """
+    Get the full path where uploaded files are stored for a Trackio project (and create the directory if it doesn't exist).
+    If a run is not provided, the files are stored in a project-level directory with the given relative path.
+
+    Args:
+        project: The project name
+        run: The run name
+        step: The step number
+        relative_path: The relative path within the directory (only used if run is not provided)
+
+    Returns:
+        The full path to the media file
+    """
+    if step is not None and run is None:
+        raise ValueError("Uploading files at a specific step requires a run")
+
+    path = MEDIA_DIR / project
+    if run:
+        path /= run
+        if step is not None:
+            path /= str(step)
+    else:
+        path /= "files"
+        if relative_path:
+            path /= relative_path
+    path.mkdir(parents=True, exist_ok=True)
+    return path

trackio/media/video.py

@@ -0,0 +1,246 @@
+import os
+import shutil
+import subprocess
+from pathlib import Path
+from typing import Literal
+
+import numpy as np
+
+from trackio.media.media import TrackioMedia
+from trackio.media.utils import check_ffmpeg_installed, check_path
+
+TrackioVideoSourceType = str | Path | np.ndarray
+TrackioVideoFormatType = Literal["gif", "mp4", "webm"]
+VideoCodec = Literal["h264", "vp9", "gif"]
+
+
+class TrackioVideo(TrackioMedia):
+    """
+    Initializes a Video object.
+
+    Example:
+        ```python
+        import trackio
+        import numpy as np
+
+        # Create a simple video from numpy array
+        frames = np.random.randint(0, 255, (10, 3, 64, 64), dtype=np.uint8)
+        video = trackio.Video(frames, caption="Random video", fps=30)
+
+        # Create a batch of videos
+        batch_frames = np.random.randint(0, 255, (3, 10, 3, 64, 64), dtype=np.uint8)
+        batch_video = trackio.Video(batch_frames, caption="Batch of videos", fps=15)
+
+        # Create video from file path
+        video = trackio.Video("path/to/video.mp4", caption="Video from file")
+        ```
+
+    Args:
+        value (`str`, `Path`, or `numpy.ndarray`, *optional*):
+            A path to a video file, or a numpy array.
+            If numpy array, should be of type `np.uint8` with RGB values in the range `[0, 255]`.
+            It is expected to have shape of either (frames, channels, height, width) or (batch, frames, channels, height, width).
+            For the latter, the videos will be tiled into a grid.
+        caption (`str`, *optional*):
+            A string caption for the video.
+        fps (`int`, *optional*):
+            Frames per second for the video. Only used when value is an ndarray. Default is `24`.
+        format (`Literal["gif", "mp4", "webm"]`, *optional*):
+            Video format ("gif", "mp4", or "webm"). Only used when value is an ndarray. Default is "gif".
+    """
+
+    TYPE = "trackio.video"
+
+    def __init__(
+        self,
+        value: TrackioVideoSourceType,
+        caption: str | None = None,
+        fps: int | None = None,
+        format: TrackioVideoFormatType | None = None,
+    ):
+        super().__init__(value, caption)
+
+        if not isinstance(self._value, TrackioVideoSourceType):
+            raise ValueError(
+                f"Invalid value type, expected {TrackioVideoSourceType}, got {type(self._value)}"
+            )
+        if isinstance(self._value, np.ndarray):
+            if self._value.dtype != np.uint8:
+                raise ValueError(
+                    f"Invalid value dtype, expected np.uint8, got {self._value.dtype}"
+                )
+            if format is None:
+                format = "gif"
+            if fps is None:
+                fps = 24
+        self._fps = fps
+        self._format = format
+
+    @staticmethod
+    def _check_array_format(video: np.ndarray) -> None:
+        """Raise an error if the array is not in the expected format."""
+        if not (video.ndim == 4 and video.shape[-1] == 3):
+            raise ValueError(
+                f"Expected RGB input shaped (F, H, W, 3), got {video.shape}. "
+                f"Input has {video.ndim} dimensions, expected 4."
+            )
+        if video.dtype != np.uint8:
+            raise TypeError(
+                f"Expected dtype=uint8, got {video.dtype}. "
+                "Please convert your video data to uint8 format."
+            )
+
+    @staticmethod
+    def write_video(
+        file_path: str | Path, video: np.ndarray, fps: float, codec: VideoCodec
+    ) -> None:
+        """RGB uint8 only, shape (F, H, W, 3)."""
+        check_ffmpeg_installed()
+        check_path(file_path)
+
+        if codec not in {"h264", "vp9", "gif"}:
+            raise ValueError("Unsupported codec. Use h264, vp9, or gif.")
+
+        arr = np.asarray(video)
+        TrackioVideo._check_array_format(arr)
+
+        frames = np.ascontiguousarray(arr)
+        _, height, width, _ = frames.shape
+        out_path = str(file_path)
+
+        cmd = [
+            "ffmpeg",
+            "-y",
+            "-f",
+            "rawvideo",
+            "-s",
+            f"{width}x{height}",
+            "-pix_fmt",
+            "rgb24",
+            "-r",
+            str(fps),
+            "-i",
+            "-",
+            "-an",
+        ]
+
+        if codec == "gif":
+            video_filter = "split[s0][s1];[s0]palettegen[p];[s1][p]paletteuse"
+            cmd += [
+                "-vf",
+                video_filter,
+                "-loop",
+                "0",
+            ]
+        elif codec == "h264":
+            cmd += [
+                "-vcodec",
+                "libx264",
+                "-pix_fmt",
+                "yuv420p",
+                "-movflags",
+                "+faststart",
+            ]
+        elif codec == "vp9":
+            bpp = 0.08
+            bps = int(width * height * fps * bpp)
+            if bps >= 1_000_000:
+                bitrate = f"{round(bps / 1_000_000)}M"
+            elif bps >= 1_000:
+                bitrate = f"{round(bps / 1_000)}k"
+            else:
+                bitrate = str(max(bps, 1))
+            cmd += [
+                "-vcodec",
+                "libvpx-vp9",
+                "-b:v",
+                bitrate,
+                "-pix_fmt",
+                "yuv420p",
+            ]
+        cmd += [out_path]
+        proc = subprocess.Popen(cmd, stdin=subprocess.PIPE, stderr=subprocess.PIPE)
+        try:
+            for frame in frames:
+                proc.stdin.write(frame.tobytes())
+        finally:
+            if proc.stdin:
+                proc.stdin.close()
+            stderr = (
+                proc.stderr.read().decode("utf-8", errors="ignore")
+                if proc.stderr
+                else ""
+            )
+            ret = proc.wait()
+            if ret != 0:
+                raise RuntimeError(f"ffmpeg failed with code {ret}\n{stderr}")
+
+    @property
+    def _codec(self) -> str:
+        match self._format:
+            case "gif":
+                return "gif"
+            case "mp4":
+                return "h264"
+            case "webm":
+                return "vp9"
+            case _:
+                raise ValueError(f"Unsupported format: {self._format}")
+
+    def _save_media(self, file_path: Path):
+        if isinstance(self._value, np.ndarray):
+            video = TrackioVideo._process_ndarray(self._value)
+            TrackioVideo.write_video(file_path, video, fps=self._fps, codec=self._codec)
+        elif isinstance(self._value, str | Path):
+            if os.path.isfile(self._value):
+                shutil.copy(self._value, file_path)
+            else:
+                raise ValueError(f"File not found: {self._value}")
+
+    @staticmethod
+    def _process_ndarray(value: np.ndarray) -> np.ndarray:
+        # Verify value is either 4D (single video) or 5D array (batched videos).
+        # Expected format: (frames, channels, height, width) or (batch, frames, channels, height, width)
+        if value.ndim < 4:
+            raise ValueError(
+                "Video requires at least 4 dimensions (frames, channels, height, width)"
+            )
+        if value.ndim > 5:
+            raise ValueError(
+                "Videos can have at most 5 dimensions (batch, frames, channels, height, width)"
+            )
+        if value.ndim == 4:
+            # Reshape to 5D with single batch: (1, frames, channels, height, width)
+            value = value[np.newaxis, ...]
+
+        value = TrackioVideo._tile_batched_videos(value)
+        return value
+
+    @staticmethod
+    def _tile_batched_videos(video: np.ndarray) -> np.ndarray:
+        """
+        Tiles a batch of videos into a grid of videos.
+
+        Input format: (batch, frames, channels, height, width) - original FCHW format
+        Output format: (frames, total_height, total_width, channels)
+        """
+        batch_size, frames, channels, height, width = video.shape
+
+        next_pow2 = 1 << (batch_size - 1).bit_length()
+        if batch_size != next_pow2:
+            pad_len = next_pow2 - batch_size
+            pad_shape = (pad_len, frames, channels, height, width)
+            padding = np.zeros(pad_shape, dtype=video.dtype)
+            video = np.concatenate((video, padding), axis=0)
+            batch_size = next_pow2
+
+        n_rows = 1 << ((batch_size.bit_length() - 1) // 2)
+        n_cols = batch_size // n_rows
+
+        # Reshape to grid layout: (n_rows, n_cols, frames, channels, height, width)
+        video = video.reshape(n_rows, n_cols, frames, channels, height, width)
+
+        # Rearrange dimensions to (frames, total_height, total_width, channels)
+        video = video.transpose(2, 0, 4, 1, 5, 3)
+        video = video.reshape(frames, n_rows * height, n_cols * width, channels)
+        return video

trackio/package.json

@@ -0,0 +1,6 @@
+{
+	"name": "trackio",
+	"version": "0.15.0",
+	"description": "",
+	"python": "true"
+}

trackio/run.py

@@ -0,0 +1,283 @@
+import threading
+import time
+import warnings
+from datetime import datetime, timezone
+
+import huggingface_hub
+from gradio_client import Client, handle_file
+
+from trackio import utils
+from trackio.gpu import GpuMonitor
+from trackio.histogram import Histogram
+from trackio.media import TrackioMedia
+from trackio.sqlite_storage import SQLiteStorage
+from trackio.table import Table
+from trackio.typehints import LogEntry, SystemLogEntry, UploadEntry
+from trackio.utils import _get_default_namespace
+
+BATCH_SEND_INTERVAL = 0.5
+
+
+class Run:
+    def __init__(
+        self,
+        url: str,
+        project: str,
+        client: Client | None,
+        name: str | None = None,
+        group: str | None = None,
+        config: dict | None = None,
+        space_id: str | None = None,
+        auto_log_gpu: bool = False,
+        gpu_log_interval: float = 10.0,
+    ):
+        self.url = url
+        self.project = project
+        self._client_lock = threading.Lock()
+        self._client_thread = None
+        self._client = client
+        self._space_id = space_id
+        self.name = name or utils.generate_readable_name(
+            SQLiteStorage.get_runs(project), space_id
+        )
+        self.group = group
+        self.config = utils.to_json_safe(config or {})
+
+        if isinstance(self.config, dict):
+            for key in self.config:
+                if key.startswith("_"):
+                    raise ValueError(
+                        f"Config key '{key}' is reserved (keys starting with '_' are reserved for internal use)"
+                    )
+
+        self.config["_Username"] = self._get_username()
+        self.config["_Created"] = datetime.now(timezone.utc).isoformat()
+        self.config["_Group"] = self.group
+
+        self._queued_logs: list[LogEntry] = []
+        self._queued_system_logs: list[SystemLogEntry] = []
+        self._queued_uploads: list[UploadEntry] = []
+        self._stop_flag = threading.Event()
+        self._config_logged = False
+
+        self._client_thread = threading.Thread(target=self._init_client_background)
+        self._client_thread.daemon = True
+        self._client_thread.start()
+
+        self._gpu_monitor: "GpuMonitor | None" = None
+        if auto_log_gpu:
+            self._gpu_monitor = GpuMonitor(self, interval=gpu_log_interval)
+            self._gpu_monitor.start()
+
+    def _get_username(self) -> str | None:
+        """Get the current HuggingFace username if logged in, otherwise None."""
+        try:
+            return _get_default_namespace()
+        except Exception:
+            return None
+
+    def _batch_sender(self):
+        """Send batched logs every BATCH_SEND_INTERVAL."""
+        while (
+            not self._stop_flag.is_set()
+            or len(self._queued_logs) > 0
+            or len(self._queued_system_logs) > 0
+        ):
+            if not self._stop_flag.is_set():
+                time.sleep(BATCH_SEND_INTERVAL)
+
+            with self._client_lock:
+                if self._client is None:
+                    return
+                if self._queued_logs:
+                    logs_to_send = self._queued_logs.copy()
+                    self._queued_logs.clear()
+                    self._client.predict(
+                        api_name="/bulk_log",
+                        logs=logs_to_send,
+                        hf_token=huggingface_hub.utils.get_token(),
+                    )
+                if self._queued_system_logs:
+                    system_logs_to_send = self._queued_system_logs.copy()
+                    self._queued_system_logs.clear()
+                    self._client.predict(
+                        api_name="/bulk_log_system",
+                        logs=system_logs_to_send,
+                        hf_token=huggingface_hub.utils.get_token(),
+                    )
+                if self._queued_uploads:
+                    uploads_to_send = self._queued_uploads.copy()
+                    self._queued_uploads.clear()
+                    self._client.predict(
+                        api_name="/bulk_upload_media",
+                        uploads=uploads_to_send,
+                        hf_token=huggingface_hub.utils.get_token(),
+                    )
+
+    def _init_client_background(self):
+        if self._client is None:
+            fib = utils.fibo()
+            for sleep_coefficient in fib:
+                try:
+                    client = Client(self.url, verbose=False)
+
+                    with self._client_lock:
+                        self._client = client
+                    break
+                except Exception:
+                    pass
+                if sleep_coefficient is not None:
+                    time.sleep(0.1 * sleep_coefficient)
+
+        self._batch_sender()
+
+    def _queue_upload(
+        self,
+        file_path,
+        step: int | None,
+        relative_path: str | None = None,
+        use_run_name: bool = True,
+    ):
+        """
+        Queues a media file for upload to a Space.
+
+        Args:
+            file_path:
+                The path to the file to upload.
+            step (`int` or `None`, *optional*):
+                The step number associated with this upload.
+            relative_path (`str` or `None`, *optional*):
+                The relative path within the project's files directory. Used when
+                uploading files via `trackio.save()`.
+            use_run_name (`bool`, *optional*):
+                Whether to use the run name for the uploaded file. This is set to
+                `False` when uploading files via `trackio.save()`.
+        """
+        upload_entry: UploadEntry = {
+            "project": self.project,
+            "run": self.name if use_run_name else None,
+            "step": step,
+            "relative_path": relative_path,
+            "uploaded_file": handle_file(file_path),
+        }
+        with self._client_lock:
+            self._queued_uploads.append(upload_entry)
+
+    def _process_media(self, value: TrackioMedia, step: int | None) -> dict:
+        """
+        Serialize media in metrics and upload to space if needed.
+        """
+        value._save(self.project, self.name, step if step is not None else 0)
+        if self._space_id:
+            self._queue_upload(value._get_absolute_file_path(), step)
+        return value._to_dict()
+
+    def _scan_and_queue_media_uploads(self, table_dict: dict, step: int | None):
+        """
+        Scan a serialized table for media objects and queue them for upload to space.
+        """
+        if not self._space_id:
+            return
+
+        table_data = table_dict.get("_value", [])
+        for row in table_data:
+            for value in row.values():
+                if isinstance(value, dict) and value.get("_type") in [
+                    "trackio.image",
+                    "trackio.video",
+                    "trackio.audio",
+                ]:
+                    file_path = value.get("file_path")
+                    if file_path:
+                        from trackio.utils import MEDIA_DIR
+
+                        absolute_path = MEDIA_DIR / file_path
+                        self._queue_upload(absolute_path, step)
+                elif isinstance(value, list):
+                    for item in value:
+                        if isinstance(item, dict) and item.get("_type") in [
+                            "trackio.image",
+                            "trackio.video",
+                            "trackio.audio",
+                        ]:
+                            file_path = item.get("file_path")
+                            if file_path:
+                                from trackio.utils import MEDIA_DIR
+
+                                absolute_path = MEDIA_DIR / file_path
+                                self._queue_upload(absolute_path, step)
+
+    def log(self, metrics: dict, step: int | None = None):
+        renamed_keys = []
+        new_metrics = {}
+
+        for k, v in metrics.items():
+            if k in utils.RESERVED_KEYS or k.startswith("__"):
+                new_key = f"__{k}"
+                renamed_keys.append(k)
+                new_metrics[new_key] = v
+            else:
+                new_metrics[k] = v
+
+        if renamed_keys:
+            warnings.warn(f"Reserved keys renamed: {renamed_keys} → '__{{key}}'")
+
+        metrics = new_metrics
+        for key, value in metrics.items():
+            if isinstance(value, Table):
+                metrics[key] = value._to_dict(
+                    project=self.project, run=self.name, step=step
+                )
+                self._scan_and_queue_media_uploads(metrics[key], step)
+            elif isinstance(value, Histogram):
+                metrics[key] = value._to_dict()
+            elif isinstance(value, TrackioMedia):
+                metrics[key] = self._process_media(value, step)
+        metrics = utils.serialize_values(metrics)
+
+        config_to_log = None
+        if not self._config_logged and self.config:
+            config_to_log = utils.to_json_safe(self.config)
+            self._config_logged = True
+
+        log_entry: LogEntry = {
+            "project": self.project,
+            "run": self.name,
+            "metrics": metrics,
+            "step": step,
+            "config": config_to_log,
+        }
+
+        with self._client_lock:
+            self._queued_logs.append(log_entry)
+
+    def log_system(self, metrics: dict):
+        """
+        Log system metrics (GPU, etc.) without a step number.
+        These metrics use timestamps for the x-axis instead of steps.
+        """
+        metrics = utils.serialize_values(metrics)
+        timestamp = datetime.now(timezone.utc).isoformat()
+
+        system_log_entry: SystemLogEntry = {
+            "project": self.project,
+            "run": self.name,
+            "metrics": metrics,
+            "timestamp": timestamp,
+        }
+
+        with self._client_lock:
+            self._queued_system_logs.append(system_log_entry)
+
+    def finish(self):
+        """Cleanup when run is finished."""
+        if self._gpu_monitor is not None:
+            self._gpu_monitor.stop()
+
+        self._stop_flag.set()
+
+        time.sleep(2 * BATCH_SEND_INTERVAL)
+
+        if self._client_thread is not None:
+            print("* Run finished. Uploading logs to Trackio (please wait...)")
+            self._client_thread.join()