tensorbored 2.21.0rc1769988660__py3-none-any.whl → 2.21.0rc1770021042__py3-none-any.whl

tensorbored/plugins/core/color_sampler.py

@@ -0,0 +1,451 @@
+# Copyright 2026 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Perceptually uniform color sampling for TensorBoard run colors.
+
+This module provides utilities for generating visually distinguishable colors
+using the OKLCH color space, which is perceptually uniform. This means equal
+steps in the color space correspond to equal perceived differences.
+
+Example usage:
+
+    from tensorbored.plugins.core import color_sampler
+
+    # Get 5 evenly-spaced colors
+    colors = color_sampler.sample_colors(5)
+    # ['#dc8a78', '#a4b93e', '#40c4aa', '#7aa6f5', '#d898d5']
+
+    # Use with run_colors
+    run_ids = ['train', 'eval', 'test', 'baseline', 'experiment']
+    run_colors = {rid: color_sampler.sample_colors(len(run_ids))[i]
+                  for i, rid in enumerate(run_ids)}
+
+    # Or use the ColorMap class for cleaner syntax
+    cm = color_sampler.ColorMap(len(run_ids))
+    run_colors = {rid: cm(i) for i, rid in enumerate(run_ids)}
+
+    # Even simpler - auto-assign from list
+    run_colors = color_sampler.colors_for_runs(run_ids)
+"""
+
+import math
+from typing import List, Tuple
+
+# =============================================================================
+# OKLCH Color Space Implementation
+# =============================================================================
+# OKLCH is a perceptually uniform color space where:
+#   L = Lightness (0 = black, 1 = white)
+#   C = Chroma (0 = gray, higher = more saturated)
+#   H = Hue angle in degrees (0-360)
+#
+# We convert OKLCH → OKLAB → Linear sRGB → sRGB → Hex
+
+
+def _oklch_to_oklab(L: float, C: float, H: float) -> Tuple[float, float, float]:
+    """Convert OKLCH to OKLAB."""
+    h_rad = math.radians(H)
+    a = C * math.cos(h_rad)
+    b = C * math.sin(h_rad)
+    return (L, a, b)
+
+
+def _oklab_to_linear_srgb(
+    L: float, a: float, b: float
+) -> Tuple[float, float, float]:
+    """Convert OKLAB to linear sRGB."""
+    # OKLAB to LMS (approximate)
+    l_ = L + 0.3963377774 * a + 0.2158037573 * b
+    m_ = L - 0.1055613458 * a - 0.0638541728 * b
+    s_ = L - 0.0894841775 * a - 1.2914855480 * b
+
+    # Cube the values
+    l = l_ * l_ * l_
+    m = m_ * m_ * m_
+    s = s_ * s_ * s_
+
+    # LMS to linear sRGB
+    r = +4.0767416621 * l - 3.3077115913 * m + 0.2309699292 * s
+    g = -1.2684380046 * l + 2.6097574011 * m - 0.3413193965 * s
+    b = -0.0041960863 * l - 0.7034186147 * m + 1.7076147010 * s
+
+    return (r, g, b)
+
+
+def _linear_to_srgb(x: float) -> float:
+    """Convert linear RGB component to sRGB (gamma correction)."""
+    if x <= 0.0031308:
+        return 12.92 * x
+    return 1.055 * (x ** (1 / 2.4)) - 0.055
+
+
+def _clamp(x: float, lo: float = 0.0, hi: float = 1.0) -> float:
+    """Clamp value to range."""
+    return max(lo, min(hi, x))
+
+
+def _oklch_to_hex(L: float, C: float, H: float) -> str:
+    """Convert OKLCH color to hex string."""
+    # OKLCH → OKLAB → Linear sRGB → sRGB
+    lab = _oklch_to_oklab(L, C, H)
+    linear = _oklab_to_linear_srgb(*lab)
+    srgb = tuple(_clamp(_linear_to_srgb(c)) for c in linear)
+
+    # Convert to 8-bit and format as hex
+    r = int(round(srgb[0] * 255))
+    g = int(round(srgb[1] * 255))
+    b = int(round(srgb[2] * 255))
+
+    return f"#{r:02x}{g:02x}{b:02x}"
+
+
+# =============================================================================
+# Public API
+# =============================================================================
+
+
+def sample_colors(
+    n: int,
+    lightness: float = 0.7,
+    chroma: float = 0.15,
+    hue_start: float = 0.0,
+    hue_range: float = 360.0,
+) -> List[str]:
+    """Generate n perceptually uniform, evenly-spaced colors.
+
+    Uses the OKLCH color space to ensure colors are visually distinguishable.
+    Colors are spaced evenly around the hue wheel while maintaining consistent
+    lightness and chroma for uniform appearance.
+
+    Args:
+        n: Number of colors to generate.
+        lightness: OKLCH lightness (0-1). Default 0.7 works well on white
+            backgrounds. Use ~0.65 for dark backgrounds.
+        chroma: OKLCH chroma (0-0.4). Higher = more saturated. Default 0.15
+            gives vivid but not garish colors.
+        hue_start: Starting hue angle in degrees (0-360). Shifts the color
+            palette around the wheel.
+        hue_range: Range of hues to use (default 360 = full wheel). Use less
+            to restrict to a portion of the spectrum.
+
+    Returns:
+        List of n hex color strings (e.g., ['#dc8a78', '#40c4aa', ...]).
+
+    Example:
+        >>> sample_colors(3)
+        ['#dc8a78', '#5fba72', '#7a9ef7']
+
+        >>> sample_colors(5, lightness=0.6, chroma=0.2)
+        ['#d96a5c', '#8ba600', '#00ab9e', '#4d95f2', '#c87ed4']
+    """
+    if n <= 0:
+        return []
+
+    colors = []
+    for i in range(n):
+        # Evenly space hues, leaving a gap so first and last aren't too close
+        hue = (hue_start + (i * hue_range / n)) % 360
+        colors.append(_oklch_to_hex(lightness, chroma, hue))
+
+    return colors
+
+
+def sample_colors_varied(
+    n: int,
+    lightness_range: Tuple[float, float] = (0.55, 0.8),
+    chroma_range: Tuple[float, float] = (0.12, 0.18),
+) -> List[str]:
+    """Generate n colors with varied lightness and chroma for maximum distinction.
+
+    When you have many colors (>8), varying lightness and chroma in addition
+    to hue helps distinguish them. This function maximizes perceptual distance
+    between colors.
+
+    Args:
+        n: Number of colors to generate.
+        lightness_range: (min, max) lightness values.
+        chroma_range: (min, max) chroma values.
+
+    Returns:
+        List of n hex color strings optimized for visual distinction.
+
+    Example:
+        >>> sample_colors_varied(10)  # Good for 10+ runs
+    """
+    if n <= 0:
+        return []
+
+    colors = []
+    l_min, l_max = lightness_range
+    c_min, c_max = chroma_range
+
+    for i in range(n):
+        # Primary variation: hue
+        hue = (i * 360 / n) % 360
+
+        # Secondary variation: alternate lightness and chroma
+        # This creates a "zigzag" pattern in L-C space
+        t = i / max(n - 1, 1)
+
+        if i % 2 == 0:
+            lightness = l_min + (l_max - l_min) * (1 - t * 0.5)
+            chroma = c_min + (c_max - c_min) * t
+        else:
+            lightness = l_min + (l_max - l_min) * (0.5 + t * 0.5)
+            chroma = c_max - (c_max - c_min) * t * 0.5
+
+        colors.append(_oklch_to_hex(lightness, chroma, hue))
+
+    return colors
+
+
+class ColorMap:
+    """A callable color map that returns colors by index.
+
+    Convenient for use with enumerate() or dict comprehensions.
+
+    Example:
+        >>> cm = ColorMap(5)
+        >>> cm(0)
+        '#dc8a78'
+        >>> cm(2)
+        '#40c4aa'
+        >>> run_colors = {rid: cm(i) for i, rid in enumerate(run_ids)}
+    """
+
+    def __init__(
+        self,
+        n: int,
+        lightness: float = 0.7,
+        chroma: float = 0.15,
+        hue_start: float = 0.0,
+        varied: bool = False,
+    ):
+        """Create a color map with n colors.
+
+        Args:
+            n: Number of colors in the palette.
+            lightness: OKLCH lightness (ignored if varied=True).
+            chroma: OKLCH chroma (ignored if varied=True).
+            hue_start: Starting hue angle.
+            varied: If True, use sample_colors_varied() for better distinction
+                with many colors (>8).
+        """
+        if varied:
+            self._colors = sample_colors_varied(n)
+        else:
+            self._colors = sample_colors(n, lightness, chroma, hue_start)
+
+    def __call__(self, index: int) -> str:
+        """Get color at index (wraps around if out of bounds)."""
+        if not self._colors:
+            return "#808080"  # Gray fallback
+        return self._colors[index % len(self._colors)]
+
+    def __len__(self) -> int:
+        return len(self._colors)
+
+    def __iter__(self):
+        return iter(self._colors)
+
+    def __getitem__(self, index: int) -> str:
+        return self._colors[index]
+
+
+def colors_for_runs(
+    run_ids: List[str],
+    lightness: float = 0.7,
+    chroma: float = 0.15,
+    varied: bool = False,
+) -> dict:
+    """Generate a run_colors dict for a list of run IDs.
+
+    Convenience function that creates a complete run_colors mapping.
+
+    Args:
+        run_ids: List of run identifiers.
+        lightness: OKLCH lightness.
+        chroma: OKLCH chroma.
+        varied: Use varied lightness/chroma for many runs.
+
+    Returns:
+        Dict mapping run IDs to hex color strings.
+
+    Example:
+        >>> colors_for_runs(['train', 'eval', 'test'])
+        {'train': '#dc8a78', 'eval': '#5fba72', 'test': '#7a9ef7'}
+    """
+    n = len(run_ids)
+    if varied or n > 8:
+        colors = sample_colors_varied(n)
+    else:
+        colors = sample_colors(n, lightness, chroma)
+
+    return {rid: colors[i] for i, rid in enumerate(run_ids)}
+
+
+# =============================================================================
+# Preset Palettes
+# =============================================================================
+
+
+def palette_categorical(n: int) -> List[str]:
+    """Generate a categorical palette optimized for charts.
+
+    Uses high chroma and medium lightness for maximum pop on white backgrounds.
+    """
+    return sample_colors(n, lightness=0.65, chroma=0.18)
+
+
+def palette_sequential(n: int, hue: float = 250) -> List[str]:
+    """Generate a sequential palette (light to dark) for ordered data.
+
+    All colors have the same hue but vary in lightness.
+
+    Args:
+        n: Number of colors.
+        hue: Base hue (default 250 = blue).
+
+    Returns:
+        List of colors from light to dark.
+    """
+    if n <= 0:
+        return []
+
+    colors = []
+    for i in range(n):
+        # Lightness from 0.9 (light) to 0.35 (dark)
+        lightness = 0.9 - (i / max(n - 1, 1)) * 0.55
+        # Chroma increases slightly with darkness
+        chroma = 0.08 + (i / max(n - 1, 1)) * 0.12
+        colors.append(_oklch_to_hex(lightness, chroma, hue))
+
+    return colors
+
+
+def palette_diverging(
+    n: int, hue_low: float = 250, hue_high: float = 30
+) -> List[str]:
+    """Generate a diverging palette for data with a meaningful midpoint.
+
+    Goes from one hue through neutral to another hue.
+
+    Args:
+        n: Number of colors (odd numbers work best).
+        hue_low: Hue for low values (default 250 = blue).
+        hue_high: Hue for high values (default 30 = orange).
+
+    Returns:
+        List of colors diverging from center.
+    """
+    if n <= 0:
+        return []
+
+    colors = []
+    mid = (n - 1) / 2
+
+    for i in range(n):
+        if i < mid:
+            # Low side: blue-ish
+            t = i / mid if mid > 0 else 0
+            lightness = 0.45 + t * 0.45  # Dark to light
+            chroma = 0.18 * (1 - t)  # Saturated to neutral
+            hue = hue_low
+        elif i > mid:
+            # High side: orange-ish
+            t = (i - mid) / (n - 1 - mid) if n - 1 > mid else 0
+            lightness = 0.9 - t * 0.45  # Light to dark
+            chroma = 0.18 * t  # Neutral to saturated
+            hue = hue_high
+        else:
+            # Midpoint: neutral
+            lightness = 0.9
+            chroma = 0.0
+            hue = 0
+
+        colors.append(_oklch_to_hex(lightness, chroma, hue))
+
+    return colors
+
+
+# =============================================================================
+# Color Utilities
+# =============================================================================
+
+
+def lighten(hex_color: str, amount: float = 0.1) -> str:
+    """Lighten a hex color by increasing its OKLCH lightness.
+
+    Args:
+        hex_color: Input color as hex string (e.g., '#dc8a78').
+        amount: How much to lighten (0-1).
+
+    Returns:
+        Lightened hex color string.
+    """
+    l, c, h = _hex_to_oklch(hex_color)
+    l = min(1.0, l + amount)
+    return _oklch_to_hex(l, c, h)
+
+
+def darken(hex_color: str, amount: float = 0.1) -> str:
+    """Darken a hex color by decreasing its OKLCH lightness.
+
+    Args:
+        hex_color: Input color as hex string.
+        amount: How much to darken (0-1).
+
+    Returns:
+        Darkened hex color string.
+    """
+    l, c, h = _hex_to_oklch(hex_color)
+    l = max(0.0, l - amount)
+    return _oklch_to_hex(l, c, h)
+
+
+def _hex_to_oklch(hex_color: str) -> Tuple[float, float, float]:
+    """Convert hex color to OKLCH (approximate reverse conversion)."""
+    # Parse hex
+    hex_color = hex_color.lstrip("#")
+    r = int(hex_color[0:2], 16) / 255
+    g = int(hex_color[2:4], 16) / 255
+    b = int(hex_color[4:6], 16) / 255
+
+    # sRGB to linear
+    def to_linear(c):
+        return c / 12.92 if c <= 0.04045 else ((c + 0.055) / 1.055) ** 2.4
+
+    r_lin = to_linear(r)
+    g_lin = to_linear(g)
+    b_lin = to_linear(b)
+
+    # Linear sRGB to LMS
+    l = 0.4122214708 * r_lin + 0.5363325363 * g_lin + 0.0514459929 * b_lin
+    m = 0.2119034982 * r_lin + 0.6806995451 * g_lin + 0.1073969566 * b_lin
+    s = 0.0883024619 * r_lin + 0.2817188376 * g_lin + 0.6299787005 * b_lin
+
+    # LMS to OKLAB
+    l_ = l ** (1 / 3) if l >= 0 else -((-l) ** (1 / 3))
+    m_ = m ** (1 / 3) if m >= 0 else -((-m) ** (1 / 3))
+    s_ = s ** (1 / 3) if s >= 0 else -((-s) ** (1 / 3))
+
+    L = 0.2104542553 * l_ + 0.7936177850 * m_ - 0.0040720468 * s_
+    a = 1.9779984951 * l_ - 2.4285922050 * m_ + 0.4505937099 * s_
+    b_val = 0.0259040371 * l_ + 0.7827717662 * m_ - 0.8086757660 * s_
+
+    # OKLAB to OKLCH
+    C = math.sqrt(a * a + b_val * b_val)
+    H = math.degrees(math.atan2(b_val, a)) % 360
+
+    return (L, C, H)

tensorbored/plugins/core/profile_writer.py

@@ -0,0 +1,280 @@
+# Copyright 2026 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Utility for writing TensorBoard default profiles from Python.
+
+This module provides a simple API for training scripts to set default
+TensorBoard dashboard configurations. When users load TensorBoard, the
+default profile will be automatically applied.
+
+Example usage:
+
+    from tensorbored.plugins.core import profile_writer
+
+    # Create a profile with pinned cards and run colors
+    profile = profile_writer.create_profile(
+        name="Training Dashboard",
+        pinned_cards=[
+            {"plugin": "scalars", "tag": "train/loss"},
+            {"plugin": "scalars", "tag": "train/accuracy"},
+            {"plugin": "scalars", "tag": "eval/loss"},
+        ],
+        run_colors={
+            "train": "#2196F3",  # Blue
+            "eval": "#4CAF50",   # Green
+        },
+        tag_filter="loss|accuracy",
+        smoothing=0.8,
+    )
+
+    # Write the profile to the logdir
+    profile_writer.write_profile(logdir, profile)
+
+    # Or use the convenience function
+    profile_writer.set_default_profile(
+        logdir,
+        pinned_cards=[{"plugin": "scalars", "tag": "train/loss"}],
+        run_colors={"train": "#ff0000"},
+    )
+"""
+
+import json
+import os
+import time
+from typing import Any, Dict, List, Optional
+
+# Profile format version
+PROFILE_VERSION = 1
+
+
+def create_profile(
+    name: str = "Default Profile",
+    pinned_cards: Optional[List[Dict[str, Any]]] = None,
+    run_colors: Optional[Dict[str, str]] = None,
+    group_colors: Optional[List[Dict[str, Any]]] = None,
+    superimposed_cards: Optional[List[Dict[str, Any]]] = None,
+    tag_filter: str = "",
+    run_filter: str = "",
+    smoothing: float = 0.6,
+    group_by: Optional[Dict[str, Any]] = None,
+) -> Dict[str, Any]:
+    """Create a TensorBoard profile dictionary.
+
+    Args:
+        name: User-friendly name for the profile.
+        pinned_cards: List of cards to pin. Each card is a dict with:
+            - plugin: str (e.g., "scalars", "images", "histograms")
+            - tag: str (the tag name)
+            - runId: str (optional, for single-run plugins)
+            - sample: int (optional, for sampled plugins like images)
+        run_colors: Dict mapping run names/IDs to hex color strings
+            (e.g., {"run1": "#ff0000", "run2": "#00ff00"}).
+        group_colors: List of group color assignments. Each entry is a dict:
+            - groupKey: str
+            - colorId: int
+        superimposed_cards: List of superimposed card definitions. Each is:
+            - id: str (unique identifier)
+            - title: str (display title)
+            - tags: List[str] (scalar tags to combine)
+            - runId: Optional[str] (run filter, or None for all runs)
+        tag_filter: Regex pattern to filter tags.
+        run_filter: Regex pattern to filter runs.
+        smoothing: Scalar smoothing value (0.0 to 0.999).
+        group_by: Grouping configuration dict with:
+            - key: str ("RUN", "EXPERIMENT", "REGEX", or "REGEX_BY_EXP")
+            - regexString: str (optional, for REGEX/REGEX_BY_EXP)
+
+    Returns:
+        A profile dictionary ready to be written to the logdir.
+    """
+    # Convert run_colors dict to list format
+    run_color_entries = []
+    if run_colors:
+        for run_id, color in run_colors.items():
+            run_color_entries.append({"runId": run_id, "color": color})
+
+    return {
+        "version": PROFILE_VERSION,
+        "data": {
+            "version": PROFILE_VERSION,
+            "name": name,
+            "lastModifiedTimestamp": int(time.time() * 1000),
+            "pinnedCards": pinned_cards or [],
+            "runColors": run_color_entries,
+            "groupColors": group_colors or [],
+            "superimposedCards": superimposed_cards or [],
+            "tagFilter": tag_filter,
+            "runFilter": run_filter,
+            "smoothing": smoothing,
+            "groupBy": group_by,
+        },
+    }
+
+
+def write_profile(logdir: str, profile: Dict[str, Any]) -> str:
+    """Write a profile to the logdir.
+
+    The profile will be written to `<logdir>/.tensorboard/default_profile.json`.
+    When TensorBoard starts with this logdir, it will offer this profile
+    as the default dashboard configuration.
+
+    Args:
+        logdir: The TensorBoard log directory.
+        profile: A profile dictionary (from create_profile or manually created).
+
+    Returns:
+        The path to the written profile file.
+
+    Raises:
+        ValueError: If the profile is missing required fields.
+        OSError: If unable to write to the logdir.
+    """
+    if "version" not in profile:
+        raise ValueError("Profile must have a 'version' field")
+
+    profile_dir = os.path.join(logdir, ".tensorboard")
+    os.makedirs(profile_dir, exist_ok=True)
+
+    profile_path = os.path.join(profile_dir, "default_profile.json")
+    with open(profile_path, "w", encoding="utf-8") as f:
+        json.dump(profile, f, indent=2)
+
+    return profile_path
+
+
+def read_profile(logdir: str) -> Optional[Dict[str, Any]]:
+    """Read the default profile from a logdir.
+
+    Args:
+        logdir: The TensorBoard log directory.
+
+    Returns:
+        The profile dictionary, or None if no profile exists.
+    """
+    profile_path = os.path.join(logdir, ".tensorboard", "default_profile.json")
+    if not os.path.exists(profile_path):
+        return None
+
+    try:
+        with open(profile_path, "r", encoding="utf-8") as f:
+            return json.load(f)
+    except (json.JSONDecodeError, OSError):
+        return None
+
+
+def set_default_profile(
+    logdir: str,
+    name: str = "Default Profile",
+    pinned_cards: Optional[List[Dict[str, Any]]] = None,
+    run_colors: Optional[Dict[str, str]] = None,
+    group_colors: Optional[List[Dict[str, Any]]] = None,
+    superimposed_cards: Optional[List[Dict[str, Any]]] = None,
+    tag_filter: str = "",
+    run_filter: str = "",
+    smoothing: float = 0.6,
+    group_by: Optional[Dict[str, Any]] = None,
+) -> str:
+    """Convenience function to create and write a profile in one call.
+
+    Args:
+        logdir: The TensorBoard log directory.
+        name: User-friendly name for the profile.
+        pinned_cards: List of cards to pin (see create_profile).
+        run_colors: Dict mapping run names to hex colors.
+        group_colors: List of group color assignments.
+        superimposed_cards: List of superimposed card definitions.
+        tag_filter: Regex pattern to filter tags.
+        run_filter: Regex pattern to filter runs.
+        smoothing: Scalar smoothing value.
+        group_by: Grouping configuration.
+
+    Returns:
+        The path to the written profile file.
+    """
+    profile = create_profile(
+        name=name,
+        pinned_cards=pinned_cards,
+        run_colors=run_colors,
+        group_colors=group_colors,
+        superimposed_cards=superimposed_cards,
+        tag_filter=tag_filter,
+        run_filter=run_filter,
+        smoothing=smoothing,
+        group_by=group_by,
+    )
+    return write_profile(logdir, profile)
+
+
+def pin_scalar(tag: str) -> Dict[str, str]:
+    """Helper to create a pinned scalar card entry.
+
+    Args:
+        tag: The scalar tag name (e.g., "train/loss").
+
+    Returns:
+        A dict suitable for the pinned_cards list.
+    """
+    return {"plugin": "scalars", "tag": tag}
+
+
+def pin_histogram(tag: str, run_id: str) -> Dict[str, str]:
+    """Helper to create a pinned histogram card entry.
+
+    Args:
+        tag: The histogram tag name.
+        run_id: The run ID (required for histograms).
+
+    Returns:
+        A dict suitable for the pinned_cards list.
+    """
+    return {"plugin": "histograms", "tag": tag, "runId": run_id}
+
+
+def pin_image(tag: str, run_id: str, sample: int = 0) -> Dict[str, Any]:
+    """Helper to create a pinned image card entry.
+
+    Args:
+        tag: The image tag name.
+        run_id: The run ID (required for images).
+        sample: The sample index (default 0).
+
+    Returns:
+        A dict suitable for the pinned_cards list.
+    """
+    return {"plugin": "images", "tag": tag, "runId": run_id, "sample": sample}
+
+
+def create_superimposed_card(
+    title: str,
+    tags: List[str],
+    run_id: Optional[str] = None,
+) -> Dict[str, Any]:
+    """Helper to create a superimposed card entry.
+
+    Superimposed cards combine multiple scalar tags on a single plot.
+
+    Args:
+        title: Display title for the card.
+        tags: List of scalar tag names to superimpose.
+        run_id: Optional run ID filter (None shows all runs).
+
+    Returns:
+        A dict suitable for the superimposed_cards list.
+    """
+    return {
+        "id": f"superimposed-{int(time.time() * 1000)}",
+        "title": title,
+        "tags": tags,
+        "runId": run_id,
+    }

tensorbored/version.py

@@ -15,7 +15,7 @@
 
 """Contains the version string."""
 
-VERSION = "2.21.0rc1769988660"
+VERSION = "2.21.0rc1770021042"
 
 if __name__ == "__main__":
     print(VERSION)

{tensorbored-2.21.0rc1769988660.dist-info → tensorbored-2.21.0rc1770021042.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tensorbored
-Version: 2.21.0rc1769988660
+Version: 2.21.0rc1770021042
 Summary: TensorBored: a PyTorch-first TensorBoard fork
 Home-page: https://github.com/Demonstrandum/tensorbored
 Author: TensorBored authors

{tensorbored-2.21.0rc1769988660.dist-info → tensorbored-2.21.0rc1770021042.dist-info}/RECORD

@@ -13,7 +13,7 @@ tensorbored/manager.py,sha256=6LKijMpiKUtPTpApEJnQby113MXM55-G_guhulGQsdA,16300
 tensorbored/notebook.py,sha256=zO8ruL979Lc1a6mGOW0J-o97RjbmIZ2nWci3l0y1wLA,14698
 tensorbored/plugin_util.py,sha256=DLJzEy9_50Iq-o14JGz1Q0DwaND9b64Q7BmDROyqZyQ,8376
 tensorbored/program.py,sha256=GHDG5FqtdJtAknzQekWGTYFXwpEnHXWHlAZoidN8Kmo,35500
-tensorbored/version.py,sha256=3XqH730qPLYju5oDzZGmQiFuZ4EuyHAc_FO_9Rcj4jM,804
+tensorbored/version.py,sha256=gIqOTSQDg2IAuXrWCI90-ktmVw68maIXrNjnEo1zqBQ,804
 tensorbored/webfiles.zip,sha256=1bIiTAQHXahMebNZdnQdctde_hd9Ws8oF69o-dFT_1g,4999505
 tensorbored/_vendor/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 tensorbored/_vendor/bleach/__init__.py,sha256=T_iQHpeDCQpz55IVNEsktpOzw3QHvwj7rfTj34bWWKc,3689
@@ -156,7 +156,9 @@ tensorbored/plugins/audio/plugin_data_pb2.py,sha256=YwUPbBSQhslWWQKbUINEX8PSitC5
 tensorbored/plugins/audio/summary.py,sha256=w1xrRP-GFXWzgrY1MS9xxKWiAD0tBe8xYWZZ7dXM7A8,8572
 tensorbored/plugins/audio/summary_v2.py,sha256=LxtBrrU3qvHjmD7UjylFTIZHhcaSNrkdhzL9LaZX3oY,5453
 tensorbored/plugins/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+tensorbored/plugins/core/color_sampler.py,sha256=5xq5IFr9-n3azp044qAH4QwdLUiS0feuZT9puBZUDak,14035
 tensorbored/plugins/core/core_plugin.py,sha256=ID72vqZrJ--Ww8MIH4jHmJdWFQt1lxC7QAyQbKHG9Ag,35920
+tensorbored/plugins/core/profile_writer.py,sha256=y4x8D1nzsFxwgizD9exjsv-RB2RSulMJ3rwVtUx6-AE,9320
 tensorbored/plugins/custom_scalar/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 tensorbored/plugins/custom_scalar/custom_scalars_plugin.py,sha256=vMJS3hBPBXxpd5zw4-LS1HxkHgB12-18cHZf3TXKmuM,11753
 tensorbored/plugins/custom_scalar/layout_pb2.py,sha256=9VfNLBWhLxEQvoOS-r0hgc9VabMBMr4hPiy1oVqwJgE,4340
@@ -263,9 +265,9 @@ tensorbored/util/platform_util.py,sha256=5jr42kvi6GqRzFx6YponSDOFYg7_RL-DTlk6dDe
 tensorbored/util/tb_logging.py,sha256=LNfFN1qZqAoZd6OaBWE_L_bqOqA_i1m9xj8lIZGQ_G4,780
 tensorbored/util/tensor_util.py,sha256=KfR_OMxxUjb8mp7yUH-x8CnEd0C8yxwHZ-vBQYDrdyE,21752
 tensorbored/util/timing.py,sha256=DduPKf8izD47D-UJ6D3xlcLDqI5WTAXK2d3WDmb3Abg,3734
-tensorbored-2.21.0rc1769988660.dist-info/licenses/LICENSE,sha256=1Y0t0nxgtnadUpTCqZlfVLWaaCju6MmbbODuxXOy-Rg,37111
-tensorbored-2.21.0rc1769988660.dist-info/METADATA,sha256=GuSjsby0TciC0EMTbKzhJYEXDs5IBrg1e-Z1g3ODJi4,1781
-tensorbored-2.21.0rc1769988660.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-tensorbored-2.21.0rc1769988660.dist-info/entry_points.txt,sha256=WiMjmdrTDkZXBlZmKWt1Kz0FCymATWUZjyTHEW-v3LU,196
-tensorbored-2.21.0rc1769988660.dist-info/top_level.txt,sha256=F6oc1TBDSM0Hq_5vyIWijSnNewbXEBnBMS15t2S1jaY,12
-tensorbored-2.21.0rc1769988660.dist-info/RECORD,,
+tensorbored-2.21.0rc1770021042.dist-info/licenses/LICENSE,sha256=1Y0t0nxgtnadUpTCqZlfVLWaaCju6MmbbODuxXOy-Rg,37111
+tensorbored-2.21.0rc1770021042.dist-info/METADATA,sha256=fqNWOS3n1BB93t6wk1ENAR-EniFsdgEFR8c7gB6YnGg,1781
+tensorbored-2.21.0rc1770021042.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+tensorbored-2.21.0rc1770021042.dist-info/entry_points.txt,sha256=WiMjmdrTDkZXBlZmKWt1Kz0FCymATWUZjyTHEW-v3LU,196
+tensorbored-2.21.0rc1770021042.dist-info/top_level.txt,sha256=F6oc1TBDSM0Hq_5vyIWijSnNewbXEBnBMS15t2S1jaY,12
+tensorbored-2.21.0rc1770021042.dist-info/RECORD,,