SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2504.05089 · ICLR · 2025

Climplicit: Climatic Implicit Embeddings for Global Ecological Tasks

Jan Wegner, Johannes Dollinger, Damien Robert, Elena Plekhanova, Lukas Drees

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2504.05089.json · badge: /api/ran/2504.05089.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ecovision-uzh/climplicit canonical (classifier label)
classifier basis: classifier_2025
3 ran · 1 no recorded run
FunctionStatusWhere it lives
Direct Ran
ran on a synthesized input
ecovision-uzh/climplicit/quickstart/climplicit.py
code served (permissive licence) · get_code("44c5914ef6783287")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

class Direct(nn.Module):
    """
    Just turn it into a [-1,1] scaling based on the input range.
    """

    def __init__(self, lon_min, lon_max, lat_min, lat_max):
        """
        Args:
        """
        super(Direct, self).__init__()
        self.lon_min = lon_min
        self.lon_max = lon_max
        self.lat_min = lat_min
        self.lat_max = lat_max

    def forward(self, coords):
        """"""
        lon, lat = (coords[:, 0], coords[:, 1])
        lon = 2 * (lon - self.lon_min) / (self.lon_max - self.lon_min) - 1
        lat = 2 * (lat - self.lat_min) / (self.lat_max - self.lat_min) - 1
        return torch.stack([lon, lat], dim=1).float()


INIT_ARGS = {
    'lon_min': -180.0,
    'lon_max': 180.0,
    'lat_min': -90.0,
    'lat_max': 90.0,
}

ARGS = {
    'coords': torch.tensor([
        [-90.0, 45.0],
        [0.0, 0.0],
    ], dtype=torch.float32),
}

if __name__ == '__main__':
    model = Direct(**INIT_ARGS)
    output = model(**ARGS)
    print(output)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 2) — shape (2, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Siren Ran
ran on a synthesized input
ecovision-uzh/climplicit/quickstart/climplicit.py
code served (permissive licence) · get_code("4f8e8e2bc5cff333")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import math
import torch.nn.functional as F

class Sine(nn.Module):

    def __init__(self, w0=1.0):
        super().__init__()
        self.w0 = w0

    def forward(self, x):
        return torch.sin(self.w0 * x)

class Siren(nn.Module):

    def __init__(self, dim_in, dim_out, w0=1.0, c=6.0, is_first=False, use_bias=True, activation=None, dropout=False, residual_connections=False, h_siren=False):
        super().__init__()
        self.dim_in = dim_in
        self.is_first = is_first
        self.dim_out = dim_out
        self.dropout = dropout
        self.h_siren = h_siren
        weight = torch.zeros(dim_out, dim_in)
        bias = torch.zeros(dim_out) if use_bias else None
        self.init_(weight, bias, c=c, w0=w0)
        self.weight = nn.Parameter(weight)
        self.bias = nn.Parameter(bias) if use_bias else None
        self.activation = Sine(w0) if activation is None else activation
        self.residual_connections = residual_connections

    def init_(self, weight, bias, c, w0):
        dim = self.dim_in
        w_std = 1 / dim if self.is_first else math.sqrt(c / dim) / w0
        weight.uniform_(-w_std, w_std)
        if not bias is None:
            bias.uniform_(-w_std, w_std)

    def forward(self, x, prev_gaussian=None):
        out = F.linear(x, self.weight, self.bias)
        if self.dropout:
            out = F.dropout(out, training=self.training)
        if self.residual_connections and prev_gaussian is not None and (out.shape == prev_gaussian.shape):
            out = (out + prev_gaussian) / 2
        gaussian = out
        if self.h_siren and self.is_first:
            out = torch.sinh(2 * out)
        out = self.activation(out)
        return (out, gaussian)


INIT_ARGS = {
    'dim_in': 4,
    'dim_out': 3,
    'w0': 1.0,
    'c': 6.0,
    'is_first': True,
    'use_bias': True,
    'activation': None,
    'dropout': False,
    'residual_connections': False,
    'h_siren': False,
}

ARGS = {
    'x': torch.randn(2, 4),
    'prev_gaussian': None,
}

if __name__ == '__main__':
    model = Siren(**INIT_ARGS)
    output, gaussian = model(**ARGS)
    print("Output shape:", output.shape)
    print("Gaussian shape:", gaussian.shape)
    print("Output:", output)
    print("Gaussian:", gaussian)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
SirenNet Ran
ran on a synthesized input
ecovision-uzh/climplicit/quickstart/climplicit.py
code served (permissive licence) · get_code("b157c2c44fab825f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import math
import torch.nn.functional as F

# Copy the classes
class Sine(nn.Module):

    def __init__(self, w0=1.0):
        super().__init__()
        self.w0 = w0

    def forward(self, x):
        return torch.sin(self.w0 * x)

class Siren(nn.Module):

    def __init__(self, dim_in, dim_out, w0=1.0, c=6.0, is_first=False, use_bias=True, activation=None, dropout=False, residual_connections=False, h_siren=False):
        super().__init__()
        self.dim_in = dim_in
        self.is_first = is_first
        self.dim_out = dim_out
        self.dropout = dropout
        self.h_siren = h_siren
        weight = torch.zeros(dim_out, dim_in)
        bias = torch.zeros(dim_out) if use_bias else None
        self.init_(weight, bias, c=c, w0=w0)
        self.weight = nn.Parameter(weight)
        self.bias = nn.Parameter(bias) if use_bias else None
        self.activation = Sine(w0) if activation is None else activation
        self.residual_connections = residual_connections

    def init_(self, weight, bias, c, w0):
        dim = self.dim_in
        w_std = 1 / dim if self.is_first else math.sqrt(c / dim) / w0
        weight.uniform_(-w_std, w_std)
        if not bias is None:
            bias.uniform_(-w_std, w_std)

    def forward(self, x, prev_gaussian=None):
        out = F.linear(x, self.weight, self.bias)
        if self.dropout:
            out = F.dropout(out, training=self.training)
        if self.residual_connections and prev_gaussian is not None and (out.shape == prev_gaussian.shape):
            out = (out + prev_gaussian) / 2
        gaussian = out
        if self.h_siren and self.is_first:
            out = torch.sinh(2 * out)
        out = self.activation(out)
        return (out, gaussian)

class SirenNet(nn.Module):
    """Sinusoidal Representation Network (SIREN) with residual connections (ReSIREN).
    Adapted from: https://github.com/microsoft/satclip/blob/main/satclip/location_encoder.py

        residual_connections: Turn the network into a ReSIREN model
        h_siren: Use the H-SIREN activation after the first layer
        return_hidden_embs: Append lower-resolution embeddings from lower layers to the output
    """

    def __init__(self, dim_in, dim_hidden, dim_out, num_layers, w0=1.0, w0_initial=30.0, use_bias=True, final_activation=None, dropout=False, residual_connections=False, h_siren=False, return_hidden_embs=None):
        super().__init__()
        self.num_layers = num_layers
        self.dim_hidden = dim_hidden
        self.return_hidden_embs = return_hidden_embs
        self.layers = nn.ModuleList([])
        for ind in range(num_layers):
            is_first = ind == 0
            layer_w0 = w0_initial if is_first else w0
            layer_dim_in = dim_in if is_first else dim_hidden
            self.layers.append(Siren(dim_in=layer_dim_in, dim_out=dim_hidden, w0=layer_w0, use_bias=use_bias, is_first=is_first, h_siren=h_siren, dropout=dropout, residual_connections=residual_connections))
        final_activation = nn.Identity() if not final_activation else final_activation
        self.last_layer = Siren(dim_in=dim_hidden, dim_out=dim_out, w0=w0, use_bias=use_bias, activation=final_activation, dropout=False)

    def forward(self, x, mods=None):
        res = []
        gaussian = None
        for i in range(len(self.layers)):
            layer = self.layers[i]
            x, gaussian = layer(x, gaussian)
            if self.return_hidden_embs is not None:
                if i in self.return_hidden_embs:
                    res.append(x)
        x, _ = self.last_layer(x, gaussian)
        if len(res) > 0:
            res.append(x)
            x = torch.cat(res, dim=1)
        return x

INIT_ARGS = {
    'dim_in': 4,
    'dim_hidden': 8,
    'dim_out': 3,
    'num_layers': 2,
    'w0': 1.0,
    'w0_initial': 30.0,
    'use_bias': True,
    'final_activation': None,
    'dropout': False,
    'residual_connections': False,
    'h_siren': False,
    'return_hidden_embs': None,
}

ARGS = {
    'x': torch.randn(2, 4),
    'mods': None,
}

if __name__ == '__main__':
    model = SirenNet(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Climplicit Not yet run
unverified (harvested, no recorded run)
ecovision-uzh/climplicit/quickstart/climplicit.py
code served (permissive licence) · get_code("5ae0117a26357563")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 2 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
Siren
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2006.09661 (2020-06)
Implicit Neural Representations with Periodic Activation Functions
SirenNet
brick of this paper's assembly: model
1 other paper
1 dated earlier
2006.09661 (2020-06)
Implicit Neural Representations with Periodic Activation Functions

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Deep learning on climatic data holds potential for macroecological applications. However, its adoption remains limited among scientists outside the deep learning community due to storage, compute, and technical expertise barriers. To address this, we introduce Climplicit, a spatio-temporal geolocation encoder pretrained to generate implicit climatic representations anywhere on Earth. By bypassing the need to download raw climatic rasters and train feature extractors, our model uses ×3500 less disk space and significantly reduces computational needs for downstream tasks. We evaluate our Climplicit embeddings on biomes classification, species distribution modeling, and plant trait regression. We find that single-layer probing our Climplicit embeddings consistently performs better or on par with training a model from scratch on downstream tasks and overall better than alternative geolocation encoding models.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2504.05089")
get_code_for_paper("2504.05089")
have("2504.05089")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2504.05089.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2504.05089.svg)](https://syntology.ai/paper/2504.05089)

Connect an agent — have() is free.