SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.00566 · NeurIPS · 2023

CROMA: Remote Sensing Representations with Contrastive Radar-Optical Masked Autoencoders

Anthony Fuller, Koreen Millard, James Green, Multimodal Encoder, Mmm Mm, Multimodal Decoder, Radar Encoder, R Er

arXiv · PDF · Open in the Atlas

Code that ran

12 of 23 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; 2 honoured the contract we drafted, 2 violated it and 6 have no contract verdict; 2 failures of our instrument; 11 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2311.00566.json · badge: /api/ran/2311.00566.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
antofuller/croma canonical (classifier label)
classifier basis: classifier_2025
9 ran · 9 no recorded run
antofuller/CROMA canonical (classifier label)
classifier basis: classifier_2025
2 ran · 2 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
Attention Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("1fcfdd6a3153b5d2")
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 3, dtype family float, shape (2, 3, 8) 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
from torch import nn, einsum
from einops import rearrange

def exists(val):
    return val is not None

class Attention(nn.Module):

    def __init__(self, dim, attention_heads=8):
        super().__init__()
        self.attention_heads = attention_heads
        dim_head = int(dim / attention_heads)
        self.scale = dim_head ** (-0.5)
        self.create_qkv = nn.Linear(dim, dim * 3, bias=False)
        self.out = nn.Linear(dim, dim)
        self.input_norm = nn.LayerNorm(dim)

    def forward(self, x, alibi):
        x = self.input_norm(x)
        q, k, v = self.create_qkv(x).chunk(3, dim=-1)
        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h=self.attention_heads), (q, k, v))
        attention_scores = einsum('b h i d, b h j d -> b h i j', q, k) * self.scale
        if exists(alibi):
            attention_scores = attention_scores + alibi
        attn = attention_scores.softmax(dim=-1)
        out = einsum('b h i j, b h j d -> b h i d', attn, v)
        return self.out(rearrange(out, 'b h n d -> b n (h d)'))

INIT_ARGS = {
    'dim': 8,
    'attention_heads': 2,
}

ARGS = {
    'x': torch.randn(2, 3, 8),
    'alibi': torch.randn(2, 2, 3, 3),
}

if __name__ == '__main__':
    attention = Attention(**INIT_ARGS)
    output = attention(**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
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 8) — shape (2, 3, 8) 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)
BaseTransformer Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("4e0949bd5a6c4d3a")
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 3, dtype family float, shape (2, 5, 8) 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
from torch import nn, einsum
from einops import rearrange

def exists(val):
    return val is not None

class Attention(nn.Module):

    def __init__(self, dim, attention_heads=8):
        super().__init__()
        self.attention_heads = attention_heads
        dim_head = int(dim / attention_heads)
        self.scale = dim_head ** (-0.5)
        self.create_qkv = nn.Linear(dim, dim * 3, bias=False)
        self.out = nn.Linear(dim, dim)
        self.input_norm = nn.LayerNorm(dim)

    def forward(self, x, alibi):
        x = self.input_norm(x)
        q, k, v = self.create_qkv(x).chunk(3, dim=-1)
        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h=self.attention_heads), (q, k, v))
        attention_scores = einsum('b h i d, b h j d -> b h i j', q, k) * self.scale
        if exists(alibi):
            attention_scores = attention_scores + alibi
        attn = attention_scores.softmax(dim=-1)
        out = einsum('b h i j, b h j d -> b h i d', attn, v)
        return self.out(rearrange(out, 'b h n d -> b n (h d)'))

class FFN(nn.Module):

    def __init__(self, dim, mult=4):
        super().__init__()
        inner_dim = int(dim * mult)
        self.net = nn.Sequential(nn.Linear(dim, inner_dim), nn.GELU(), nn.Linear(inner_dim, dim))
        self.input_norm = nn.LayerNorm(dim)

    def forward(self, x):
        x = self.input_norm(x)
        return self.net(x)

class BaseTransformer(nn.Module):

    def __init__(self, dim, layers, attention_heads=8, ff_mult=4, final_norm=True):
        super().__init__()
        self.final_norm = final_norm
        self.layers = nn.ModuleList([])
        for _ in range(layers):
            self.layers.append(nn.ModuleList([Attention(dim=dim, attention_heads=attention_heads), FFN(dim=dim, mult=ff_mult)]))
        if self.final_norm:
            self.norm_out = nn.LayerNorm(dim)

    def forward(self, x, alibi=None):
        for self_attn, ffn in self.layers:
            x = self_attn(x, alibi) + x
            x = ffn(x) + x
        if self.final_norm:
            return self.norm_out(x)
        else:
            return x

INIT_ARGS = {
    'dim': 8,
    'layers': 2,
    'attention_heads': 2,
    'ff_mult': 4,
    'final_norm': True,
}

ARGS = {
    'x': torch.randn(2, 5, 8),
    'alibi': None,
}

if __name__ == '__main__':
    model = BaseTransformer(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 5, 8)")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 5, 8) — shape (2, 5, 8) 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)
BaseTransformerCrossAttn Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("d1017efaaf6dbc46")
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 3, 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 numpy as np
from torch import nn

INIT_ARGS = {
    'dim': 4,
    'layers': 2,
    'attention_heads': 2,
    'ff_mult': 2
}

# Create small tensors for input and context
batch_size = 2
seq_len = 3
dim = INIT_ARGS['dim']

x = torch.randn(batch_size, seq_len, dim)
context = torch.randn(batch_size, seq_len, dim)

# Create alibi tensor (small stand-in)
alibi = torch.randn(1, INIT_ARGS['attention_heads'], seq_len, seq_len) * 0.1

ARGS = {
    'x': x,
    'context': context,
    'alibi': alibi
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
einops, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, 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)
ContrastLossInput Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("9a8c2a02087bbb20")
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 0, 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 numpy as np

# Mock distributed environment for single-process testing
class MockDist:
    @staticmethod
    def all_gather(tensor_list, tensor):
        # Simulate gathering by just copying the tensor to each position
        for i in range(len(tensor_list)):
            tensor_list[i].copy_(tensor)

# Patch the distributed module
import torch.distributed as dist
original_all_gather = dist.all_gather if hasattr(dist, 'all_gather') else None
dist.all_gather = MockDist.all_gather

INIT_ARGS = {
    'projection_input': 6,
    'projection_output': 8
}

batch_size = 2
world_size = 2
rank = 0

ARGS = {
    'radar_features': torch.randn(batch_size, 6),
    'optical_features': torch.randn(batch_size, 6),
    'world_size': world_size,
    'rank': rank
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, 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)
CrossAttention Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("ba24c9bdfa918f05")
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 3, dtype family float, shape (2, 3, 8) 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
from torch import nn, einsum
from einops import rearrange

class CrossAttention(nn.Module):

    def __init__(self, dim, attention_heads=8):
        super().__init__()
        self.attention_heads = attention_heads
        dim_head = int(dim / attention_heads)
        self.scale = dim_head ** (-0.5)
        self.create_q = nn.Linear(dim, dim, bias=False)
        self.create_k = nn.Linear(dim, dim, bias=False)
        self.create_v = nn.Linear(dim, dim, bias=False)
        self.to_out = nn.Linear(dim, dim)
        self.input_norm = nn.LayerNorm(dim)

    def forward(self, x, context, alibi):
        x = self.input_norm(x)
        context = self.input_norm(context)
        q = self.create_q(x)
        k = self.create_k(context)
        v = self.create_v(context)
        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h=self.attention_heads), (q, k, v))
        attention_scores = einsum('b h i d, b h j d -> b h i j', q, k) * self.scale
        attention_scores = attention_scores + alibi
        attn = attention_scores.softmax(dim=-1)
        out = einsum('b h i j, b h j d -> b h i d', attn, v)
        out = rearrange(out, 'b h n d -> b n (h d)')
        return self.to_out(out)


INIT_ARGS = {
    'dim': 8,
    'attention_heads': 2,
}

ARGS = {
    'x': torch.randn(2, 3, 8),
    'context': torch.randn(2, 5, 8),
    'alibi': torch.randn(2, 2, 3, 5),
}

if __name__ == '__main__':
    model = CrossAttention(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 3, 8)")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 8) — shape (2, 3, 8) 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)
FFN Ran
ran on a synthesized input
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("2346039a18d33db2")
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, 4) 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
from torch import nn

class FFN(nn.Module):

    def __init__(self, dim, mult=4):
        super().__init__()
        inner_dim = int(dim * mult)
        self.net = nn.Sequential(nn.Linear(dim, inner_dim), nn.GELU(), nn.Linear(inner_dim, dim))
        self.input_norm = nn.LayerNorm(dim)

    def forward(self, x):
        x = self.input_norm(x)
        return self.net(x)


INIT_ARGS = {
    'dim': 4,
    'mult': 4,
}

ARGS = {
    'x': torch.randn(2, 4),
}

if __name__ == '__main__':
    model = FFN(**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, shape (2, 4) — shape (2, 4) 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)
apply_mask_to_alibi Ran
ran; our contract draft was wrong, not the code
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("be7db6871909bd33")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (2, 4, 3, 3) 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
from einops import rearrange

torch.manual_seed(0)

# Analyze the function to understand tensor shapes:
# - ids_keep_queries: shape (b, masked_seq_len) - indices of kept query positions
# - ids_keep_keys: shape (b, masked_seq_len) - indices of kept key positions
# - alibi: shape (1, attention_heads, orig_seq_len, orig_seq_len) - ALiBi bias
# - batch_size: int
# - orig_seq_len: int - original sequence length
# - masked_seq_len: int - length after masking
# - attention_heads: int

batch_size = 2
orig_seq_len = 5
masked_seq_len = 3
attention_heads = 4

# Create ids_keep_queries and ids_keep_keys
# These should contain indices in range [0, orig_seq_len)
ids_keep_queries = torch.tensor([
    [0, 2, 4],
    [1, 3, 4]
], dtype=torch.long)

ids_keep_keys = torch.tensor([
    [0, 1, 3],
    [0, 2, 4]
], dtype=torch.long)

# Create alibi tensor with shape (1, attention_heads, orig_seq_len, orig_seq_len)
alibi = torch.randn(1, attention_heads, orig_seq_len, orig_seq_len)

ARGS = {
    "alibi": alibi,
    "ids_keep_queries": ids_keep_queries,
    "ids_keep_keys": ids_keep_keys,
    "batch_size": batch_size,
    "orig_seq_len": orig_seq_len,
    "masked_seq_len": masked_seq_len,
    "attention_heads": attention_heads,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 4, 3, 3) — shape (2, 4, 3, 3) 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: none by the harness; any seed is in the setup block
apply_mask_to_sequence Ran
ran; our contract draft was wrong, not the code
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("e52be538e2263534")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 3, dtype family float, shape (2, 2, 3) 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

torch.manual_seed(0)

# x: input tensor with shape (batch_size, seq_len, feature_dim)
# ids_keep: indices tensor with shape (batch_size, num_keep)
# The function gathers along dim=1 using ids_keep expanded to match feature dimensions

x = torch.randn(2, 4, 3)  # batch=2, seq_len=4, feature_dim=3
ids_keep = torch.tensor([[0, 2], [1, 3]])  # batch=2, num_keep=2

ARGS = {
    "x": x,
    "ids_keep": ids_keep,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 3, dtype family float, shape (2, 2, 3) — shape (2, 2, 3) 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: none by the harness; any seed is in the setup block
exists Ran
ran, violated the contract we drafted
antofuller/CROMA/pretrain_croma.py
code served (permissive licence) · get_code("689b78ad9d312313")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned bool, ndim 0, dtype family float, shape () 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 numpy as np

torch.manual_seed(0)

def exists(val):
    return val is not None

# The function 'exists' takes a single parameter 'val' and returns True if val is not None
# We need to provide test cases for this simple function

ARGS = {
    'val': torch.tensor([1, 2, 3])
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
einops, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () 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: none by the harness; any seed is in the setup block
exists Ran
ran, violated the contract we drafted
this paper's copy was not recorded; identical code first harvested from ThomasMrY/VCT
pointer only; the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("aa5486a3650902d8")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_2dalibi Ran
ran, honoured the contract we drafted
antofuller/CROMA/use_croma.py
code served (permissive licence) · get_code("6861b66287490d06")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (1, 4, 9, 9) 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 math

torch.manual_seed(0)

ARGS = {
    "num_heads": 4,
    "num_patches": 9,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (1, 4, 9, 9) — shape (1, 4, 9, 9) 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: none by the harness; any seed is in the setup block
get_alibi Ran
ran, honoured the contract we drafted
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("2c3811714f31ec91")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (1, 8, 16, 16) 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 math
import itertools

torch.manual_seed(0)

ARGS = {
    "attention_heads": 8,
    "num_patches": 16,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 4, dtype family float, shape (1, 8, 16, 16) — shape (1, 8, 16, 16) 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: none by the harness; any seed is in the setup block
CROMA Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("bfd179d5cf8b21fb")
How it ran

no invocation recorded.

DecoderMAE Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("8571409c484a1238")
How it ran

no invocation recorded.

ViT Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("9f7fe5abf47f2d58")
How it ran

no invocation recorded.

gather_features Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("816756cd7d531829")
How it ran

no invocation recorded.

get_1d_sincos_pos_embed_from_grid Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("2fae31fd41fc4c69")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("d599b95b94498131")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed Not yet run
unverified (harvested, no recorded run)
antofuller/CROMA/pretrain_croma.py
code served (permissive licence) · get_code("2ac385d924e863c5")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("1b10b35db2b36f35")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed_from_grid Not yet run
unverified (harvested, no recorded run)
antofuller/croma/pretrain_croma.py
code served (permissive licence) · get_code("90bcf667c7d798b4")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed_from_grid Not yet run
unverified (harvested, no recorded run)
antofuller/CROMA/pretrain_croma.py
code served (permissive licence) · get_code("be98aaf46e9dea53")
How it ran

no invocation recorded.

get_2dalibi Not yet run
unverified (harvested, no recorded run)
antofuller/croma/use_croma.py
code served (permissive licence) · get_code("aeec3aa501ca98c5")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

Also present in other papers' code

Of 23 samples lifted from this paper's linked repositories, 23 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 4 the same month, 4 later) and 14 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
exists
brick of this paper's assembly: helper
486 other papers
247 dated earlier, 14 the same month, 219 later, 4 same year, order not determinable, 2 undated · 78 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1611.07004 (2016-11)
Image-to-Image Translation with Conditional Adversarial Networks
1804.04637 (2018-04)
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
and 483 more not named here
get_2d_sincos_pos_embed_from_grid
brick of this paper's assembly: helper
190 other papers
53 dated earlier, 2 the same month, 134 later, 1 same year, order not determinable · 65 only as an assembly brick
1807.10221 (2018-07)
Unified Perceptual Parsing for Scene Understanding
2105.12723 (2021-05)
Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
and 187 more not named here
get_2d_sincos_pos_embed
brick of this paper's assembly: helper
95 other papers
38 dated earlier, 2 the same month, 54 later, 1 same year, order not determinable · 33 only as an assembly brick
1807.10221 (2018-07)
Unified Perceptual Parsing for Scene Understanding
2105.12723 (2021-05)
Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
and 92 more not named here
get_1d_sincos_pos_embed_from_grid
brick of this paper's assembly: helper
39 other papers
25 dated earlier, 1 the same month, 12 later, 1 same year, order not determinable · 25 only as an assembly brick
2105.12723 (2021-05)
Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
2204.09435 (2022-04)
Hephaestus: A large scale multitask dataset towards InSAR understanding
and 36 more not named here

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

A vital and rapidly growing application, remote sensing offers vast yet sparsely labeled, spatially aligned multimodal data; this makes self-supervised learning algorithms invaluable. We present CROMA: a framework that combines contrastive and reconstruction self-supervised objectives to learn rich unimodal and multimodal representations. Our method separately encodes masked-out multispectral optical and synthetic aperture radar samples-aligned in space and time-and performs cross-modal contrastive learning. Another encoder fuses these sensors, producing joint multimodal encodings that are used to predict the masked patches via a lightweight decoder. We show that these objectives are complementary when leveraged on spatially aligned multimodal data. We also introduce X-and 2D-ALiBi, which spatially biases our cross-and self-attention matrices. These strategies improve representations and allow our models to effectively extrapolate to images up to 17.6× larger at test-time. CROMA outperforms the current SoTA multispectral model, evaluated on: four classification benchmarksfinetuning (avg.↑ 1.8%), linear (avg.↑ 2.4%) and nonlinear (avg.↑ 1.4%) probing, kNN classification (avg.↑ 3.5%), and K-means clustering (avg.↑ 8.4%); and three segmentation benchmarks (avg.↑ 6.4%). CROMA's rich, optionally multimodal representations can be widely leveraged across remote sensing applications.

For agents

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

get_harvested_code_for_paper("2311.00566")
get_code_for_paper("2311.00566")
have("2311.00566")

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

curl https://syntology.ai/api/ran/2311.00566.json

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

[![Syntology run record](https://syntology.ai/api/ran/2311.00566.svg)](https://syntology.ai/paper/2311.00566)

Connect an agent — have() is free.