SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2404.13478 · ICLR · 2024

Deep SE(3)-Equivariant Geometric Reasoning for Precise Placement Tasks

Yi Yang, David Held, Jon Scholz, Ben Eisner, Todor Davchev, Mel Veceric

arXiv · PDF · Open in the Atlas

Code that ran

13 of 22 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; 1 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 2 failures of our instrument; 9 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/2404.13478.json · badge: /api/ran/2404.13478.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
r-pad/taxpose community
Papers with Code archive
12 ran · 9 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
DGCNN Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("0256b22bbf6fc0bd")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 torch.nn.functional as F

def get_graph_feature(x, k=20):
    idx = knn(x, k=k)
    batch_size, num_points, _ = idx.size()
    device = x.device
    idx_base = torch.arange(0, batch_size, device=device).view(-1, 1, 1) * num_points
    idx = idx + idx_base
    idx = idx.view(-1)
    _, num_dims, _ = x.size()
    x = x.transpose(2, 1).contiguous()
    feature = x.view(batch_size * num_points, -1)[idx, :]
    feature = feature.view(batch_size, num_points, k, num_dims)
    x = x.view(batch_size, num_points, 1, num_dims).repeat(1, 1, k, 1)
    feature = torch.cat((feature, x), dim=3).permute(0, 3, 1, 2)
    return feature

def knn(x, k):
    inner = -2 * torch.matmul(x.transpose(2, 1).contiguous(), x)
    xx = torch.sum(x ** 2, dim=1, keepdim=True)
    pairwise_distance = -xx - inner - xx.transpose(2, 1).contiguous()
    idx = pairwise_distance.topk(k=k, dim=-1)[1]
    return idx

class DGCNN(nn.Module):

    def __init__(self, emb_dims=512):
        super(DGCNN, self).__init__()
        self.conv1 = nn.Conv2d(6, 64, kernel_size=1, bias=False)
        self.conv2 = nn.Conv2d(64, 64, kernel_size=1, bias=False)
        self.conv3 = nn.Conv2d(64, 128, kernel_size=1, bias=False)
        self.conv4 = nn.Conv2d(128, 256, kernel_size=1, bias=False)
        self.conv5 = nn.Conv2d(512, emb_dims, kernel_size=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.bn2 = nn.BatchNorm2d(64)
        self.bn3 = nn.BatchNorm2d(128)
        self.bn4 = nn.BatchNorm2d(256)
        self.bn5 = nn.BatchNorm2d(emb_dims)

    def forward(self, x):
        batch_size, num_dims, num_points = x.size()
        x = get_graph_feature(x)
        x = F.relu(self.bn1(self.conv1(x)))
        x1 = x.max(dim=-1, keepdim=True)[0]
        x = F.relu(self.bn2(self.conv2(x)))
        x2 = x.max(dim=-1, keepdim=True)[0]
        x = F.relu(self.bn3(self.conv3(x)))
        x3 = x.max(dim=-1, keepdim=True)[0]
        x = F.relu(self.bn4(self.conv4(x)))
        x4 = x.max(dim=-1, keepdim=True)[0]
        x = torch.cat((x1, x2, x3, x4), dim=1)
        x = F.relu(self.bn5(self.conv5(x))).view(batch_size, -1, num_points)
        return x

INIT_ARGS = {
    'emb_dims': 32
}

ARGS = {
    'x': torch.randn(2, 3, 25)
}
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 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)
Decoder Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("511f66707327f995")
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, 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
import torch.nn as nn
import copy

class LayerNorm(nn.Module):

    def __init__(self, features, eps=1e-06):
        super(LayerNorm, self).__init__()
        self.a_2 = nn.Parameter(torch.ones(features))
        self.b_2 = nn.Parameter(torch.zeros(features))
        self.eps = eps

    def forward(self, x):
        mean = x.mean(-1, keepdim=True)
        std = x.std(-1, keepdim=True)
        return self.a_2 * (x - mean) / (std + self.eps) + self.b_2

def clones(module, N):
    return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])

class Decoder(nn.Module):
    """Generic N layer decoder with masking."""

    def __init__(self, layer, N):
        super(Decoder, self).__init__()
        self.layers = clones(layer, N)
        self.norm = LayerNorm(layer.size)

    def forward(self, x, memory, src_mask, tgt_mask):
        for layer in self.layers:
            x = layer(x, memory, src_mask, tgt_mask)
        return self.norm(x)


class MockDecoderLayer(nn.Module):
    def __init__(self, size):
        super(MockDecoderLayer, self).__init__()
        self.size = size
        self.linear = nn.Linear(size, size)
    
    def forward(self, x, memory, src_mask, tgt_mask):
        return self.linear(x)


mock_layer = MockDecoderLayer(size=4)

INIT_ARGS = {
    'layer': mock_layer,
    'N': 2
}

ARGS = {
    'x': torch.randn(2, 3, 4),
    'memory': torch.randn(2, 5, 4),
    'src_mask': None,
    'tgt_mask': None
}

if __name__ == '__main__':
    decoder = Decoder(**INIT_ARGS)
    output = decoder(**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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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)
DecoderLayer Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("6fa6bcc8cc01f9be")
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 torch.nn as nn
import copy

class LayerNorm(nn.Module):

    def __init__(self, features, eps=1e-06):
        super(LayerNorm, self).__init__()
        self.a_2 = nn.Parameter(torch.ones(features))
        self.b_2 = nn.Parameter(torch.zeros(features))
        self.eps = eps

    def forward(self, x):
        mean = x.mean(-1, keepdim=True)
        std = x.std(-1, keepdim=True)
        return self.a_2 * (x - mean) / (std + self.eps) + self.b_2

class SublayerConnection(nn.Module):

    def __init__(self, size, dropout=None):
        super(SublayerConnection, self).__init__()
        self.norm = LayerNorm(size)

    def forward(self, x, sublayer):
        return x + sublayer(self.norm(x))

def clones(module, N):
    return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])

class DecoderLayer(nn.Module):
    """Decoder is made of self-attn, src-attn, and feed forward (defined below)"""

    def __init__(self, size, self_attn, src_attn, feed_forward, dropout):
        super(DecoderLayer, self).__init__()
        self.size = size
        self.self_attn = self_attn
        self.src_attn = src_attn
        self.feed_forward = feed_forward
        self.sublayer = clones(SublayerConnection(size, dropout), 3)

    def forward(self, x, memory, src_mask, tgt_mask):
        """Follow Figure 1 (right) for connections."""
        m = memory
        x = self.sublayer[0](x, lambda x: self.self_attn(x, x, x, tgt_mask))
        x = self.sublayer[1](x, lambda x: self.src_attn(x, m, m, src_mask))
        return self.sublayer[2](x, self.feed_forward)


class MockAttention(nn.Module):
    def __init__(self, d_model):
        super().__init__()
        self.d_model = d_model
        self.linear = nn.Linear(d_model, d_model)
    
    def forward(self, query, key, value, mask=None):
        return self.linear(query)


class MockFeedForward(nn.Module):
    def __init__(self, d_model):
        super().__init__()
        self.linear = nn.Linear(d_model, d_model)
    
    def forward(self, x):
        return self.linear(x)


d_model = 8
batch_size = 2
tgt_seq_len = 5
src_seq_len = 7

self_attn = MockAttention(d_model)
src_attn = MockAttention(d_model)
feed_forward = MockFeedForward(d_model)

INIT_ARGS = {
    'size': d_model,
    'self_attn': self_attn,
    'src_attn': src_attn,
    'feed_forward': feed_forward,
    'dropout': 0.1
}

ARGS = {
    'x': torch.randn(batch_size, tgt_seq_len, d_model),
    'memory': torch.randn(batch_size, src_seq_len, d_model),
    'src_mask': torch.ones(batch_size, 1, src_seq_len).bool(),
    'tgt_mask': torch.ones(batch_size, tgt_seq_len, tgt_seq_len).bool()
}
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 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)
Encoder Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("065e92ade1cc3217")
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, 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
import torch.nn as nn
import copy

class LayerNorm(nn.Module):

    def __init__(self, features, eps=1e-06):
        super(LayerNorm, self).__init__()
        self.a_2 = nn.Parameter(torch.ones(features))
        self.b_2 = nn.Parameter(torch.zeros(features))
        self.eps = eps

    def forward(self, x):
        mean = x.mean(-1, keepdim=True)
        std = x.std(-1, keepdim=True)
        return self.a_2 * (x - mean) / (std + self.eps) + self.b_2

def clones(module, N):
    return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])

class Encoder(nn.Module):

    def __init__(self, layer, N):
        super(Encoder, self).__init__()
        self.layers = clones(layer, N)
        self.norm = LayerNorm(layer.size)

    def forward(self, x, mask):
        for layer in self.layers:
            x = layer(x, mask)
        return self.norm(x)


class DummyLayer(nn.Module):
    def __init__(self, size):
        super(DummyLayer, self).__init__()
        self.size = size
        self.linear = nn.Linear(size, size)
    
    def forward(self, x, mask):
        return self.linear(x)


dummy_layer = DummyLayer(size=4)

INIT_ARGS = {
    'layer': dummy_layer,
    'N': 2
}

ARGS = {
    'x': torch.randn(2, 3, 4),
    'mask': None
}
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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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)
EncoderDecoder Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("0b1c98d83e7c5bed")
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 torch.nn as nn

# Define small stand-in modules for encoder, decoder, embeddings, and generator
class SimpleEncoder(nn.Module):
    def __init__(self, d_model):
        super().__init__()
        self.d_model = d_model
    
    def forward(self, x, mask):
        return x

class SimpleDecoder(nn.Module):
    def __init__(self, d_model):
        super().__init__()
        self.d_model = d_model
    
    def forward(self, tgt, memory, src_mask, tgt_mask):
        return tgt

class SimpleEmbedding(nn.Module):
    def __init__(self, vocab_size, d_model):
        super().__init__()
        self.embedding = nn.Embedding(vocab_size, d_model)
    
    def forward(self, x):
        return self.embedding(x)

class SimpleGenerator(nn.Module):
    def __init__(self, d_model, vocab_size):
        super().__init__()
        self.linear = nn.Linear(d_model, vocab_size)
    
    def forward(self, x):
        return self.linear(x)

class EncoderDecoder(nn.Module):
    """
    A standard Encoder-Decoder architecture. Base for this and many
    other models.
    """

    def __init__(self, encoder, decoder, src_embed, tgt_embed, generator):
        super(EncoderDecoder, self).__init__()
        self.encoder = encoder
        self.decoder = decoder
        self.src_embed = src_embed
        self.tgt_embed = tgt_embed
        self.generator = generator

    def forward(self, src, tgt, src_mask, tgt_mask):
        """Take in and process masked src and target sequences."""
        return self.decode(self.encode(src, src_mask), src_mask, tgt, tgt_mask)

    def encode(self, src, src_mask):
        return self.encoder(self.src_embed(src), src_mask)

    def decode(self, memory, src_mask, tgt, tgt_mask):
        return self.generator(self.decoder(self.tgt_embed(tgt), memory, src_mask, tgt_mask))

# Configuration parameters
d_model = 4
src_vocab_size = 10
tgt_vocab_size = 12
batch_size = 2
src_seq_len = 5
tgt_seq_len = 3

# Build components
encoder = SimpleEncoder(d_model)
decoder = SimpleDecoder(d_model)
src_embed = SimpleEmbedding(src_vocab_size, d_model)
tgt_embed = SimpleEmbedding(tgt_vocab_size, d_model)
generator = SimpleGenerator(d_model, tgt_vocab_size)

INIT_ARGS = {
    'encoder': encoder,
    'decoder': decoder,
    'src_embed': src_embed,
    'tgt_embed': tgt_embed,
    'generator': generator,
}

# Create input tensors
src = torch.randint(0, src_vocab_size, (batch_size, src_seq_len))
tgt = torch.randint(0, tgt_vocab_size, (batch_size, tgt_seq_len))
src_mask = torch.ones(batch_size, 1, src_seq_len)
tgt_mask = torch.ones(batch_size, tgt_seq_len, tgt_seq_len)

ARGS = {
    'src': src,
    'tgt': tgt,
    'src_mask': src_mask,
    'tgt_mask': tgt_mask,
}
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 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)
EncoderLayer Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("24ccd00d4adcc3a9")
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, 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
import torch.nn as nn
import copy

class LayerNorm(nn.Module):

    def __init__(self, features, eps=1e-06):
        super(LayerNorm, self).__init__()
        self.a_2 = nn.Parameter(torch.ones(features))
        self.b_2 = nn.Parameter(torch.zeros(features))
        self.eps = eps

    def forward(self, x):
        mean = x.mean(-1, keepdim=True)
        std = x.std(-1, keepdim=True)
        return self.a_2 * (x - mean) / (std + self.eps) + self.b_2

class SublayerConnection(nn.Module):

    def __init__(self, size, dropout=None):
        super(SublayerConnection, self).__init__()
        self.norm = LayerNorm(size)

    def forward(self, x, sublayer):
        return x + sublayer(self.norm(x))

def clones(module, N):
    return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])

class EncoderLayer(nn.Module):

    def __init__(self, size, self_attn, feed_forward, dropout):
        super(EncoderLayer, self).__init__()
        self.self_attn = self_attn
        self.feed_forward = feed_forward
        self.sublayer = clones(SublayerConnection(size, dropout), 2)
        self.size = size

    def forward(self, x, mask):
        x = self.sublayer[0](x, lambda x: self.self_attn(x, x, x, mask))
        return self.sublayer[1](x, self.feed_forward)


class DummyAttention(nn.Module):
    def forward(self, q, k, v, mask):
        return v


class DummyFeedForward(nn.Module):
    def forward(self, x):
        return x


INIT_ARGS = {
    'size': 4,
    'self_attn': DummyAttention(),
    'feed_forward': DummyFeedForward(),
    'dropout': 0.1
}

ARGS = {
    'x': torch.randn(2, 3, 4),
    'mask': None
}
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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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)
MultiHeadedAttention Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("2aaa428ab3d2f019")
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, 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 math
import torch
import torch.nn as nn
import torch.nn.functional as F
import copy

def attention(query, key, value, mask=None, dropout=None):
    d_k = query.size(-1)
    scores = torch.matmul(query, key.transpose(-2, -1).contiguous()) / math.sqrt(d_k)
    if mask is not None:
        scores = scores.masked_fill(mask == 0, -1000000000.0)
    p_attn = F.softmax(scores, dim=-1)
    return (torch.matmul(p_attn, value), p_attn)

def clones(module, N):
    return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])

class MultiHeadedAttention(nn.Module):

    def __init__(self, h, d_model, dropout=0.1):
        """Take in model size and number of heads."""
        super(MultiHeadedAttention, self).__init__()
        assert d_model % h == 0
        self.d_k = d_model // h
        self.h = h
        self.linears = clones(nn.Linear(d_model, d_model), 4)
        self.attn = None
        self.dropout = None

    def forward(self, query, key, value, mask=None):
        """Implements Figure 2"""
        if mask is not None:
            mask = mask.unsqueeze(1)
        nbatches = query.size(0)
        query, key, value = [l(x).view(nbatches, -1, self.h, self.d_k).transpose(1, 2).contiguous() for l, x in zip(self.linears, (query, key, value))]
        x, self.attn = attention(query, key, value, mask=mask, dropout=self.dropout)
        x = x.transpose(1, 2).contiguous().view(nbatches, -1, self.h * self.d_k)
        return self.linears[-1](x)


INIT_ARGS = {
    'h': 2,
    'd_model': 4,
    'dropout': 0.1
}

ARGS = {
    'query': torch.randn(2, 3, 4),
    'key': torch.randn(2, 3, 4),
    'value': torch.randn(2, 3, 4),
    'mask': None
}

if __name__ == '__main__':
    model = MultiHeadedAttention(**INIT_ARGS)
    output = model(**ARGS)
    print("Output shape:", output.shape)
    print("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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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)
PointNet Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("79a4b9f82b00cd47")
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, 64, 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
import torch.nn as nn
import torch.nn.functional as F

class PointNet(nn.Module):

    def __init__(self, layer_dims=[3, 64, 64, 64, 128, 512]):
        super(PointNet, self).__init__()
        convs = []
        norms = []
        for j in range(len(layer_dims) - 1):
            convs.append(nn.Conv1d(layer_dims[j], layer_dims[j + 1], kernel_size=1, bias=False))
            norms.append(nn.BatchNorm1d(layer_dims[j + 1]))
        self.convs = nn.ModuleList(convs)
        self.norms = nn.ModuleList(norms)

    def forward(self, x):
        for bn, conv in zip(self.norms, self.convs):
            x = F.relu(bn(conv(x)))
        return x


INIT_ARGS = {
    'layer_dims': [3, 16, 32, 64]
}

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

if __name__ == '__main__':
    model = PointNet(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
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 3, dtype family float, shape (2, 64, 8) — shape (2, 64, 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)
PositionwiseFeedForward Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("49c1ccb5ad835ec4")
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 torch.nn as nn
import torch.nn.functional as F

class PositionwiseFeedForward(nn.Module):
    """Implements FFN equation."""

    def __init__(self, d_model, d_ff, dropout=0.1):
        super(PositionwiseFeedForward, self).__init__()
        self.w_1 = nn.Linear(d_model, d_ff)
        self.norm = nn.Sequential()
        self.w_2 = nn.Linear(d_ff, d_model)
        self.dropout = None

    def forward(self, x):
        return self.w_2(self.norm(F.relu(self.w_1(x)).transpose(2, 1).contiguous()).transpose(2, 1).contiguous())


INIT_ARGS = {
    'd_model': 4,
    'd_ff': 8,
    'dropout': 0.1
}

ARGS = {
    'x': torch.randn(2, 3, 4)
}
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 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)
SublayerConnection Ran
ran on a synthesized input
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("dd481d1dfb89ace0")
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, 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
import torch.nn as nn

class LayerNorm(nn.Module):

    def __init__(self, features, eps=1e-06):
        super(LayerNorm, self).__init__()
        self.a_2 = nn.Parameter(torch.ones(features))
        self.b_2 = nn.Parameter(torch.zeros(features))
        self.eps = eps

    def forward(self, x):
        mean = x.mean(-1, keepdim=True)
        std = x.std(-1, keepdim=True)
        return self.a_2 * (x - mean) / (std + self.eps) + self.b_2

class SublayerConnection(nn.Module):

    def __init__(self, size, dropout=None):
        super(SublayerConnection, self).__init__()
        self.norm = LayerNorm(size)

    def forward(self, x, sublayer):
        return x + sublayer(self.norm(x))


INIT_ARGS = {
    'size': 4,
    'dropout': None,
}

ARGS = {
    'x': torch.randn(2, 3, 4),
    'sublayer': lambda t: torch.randn_like(t),
}

if __name__ == '__main__':
    model = SublayerConnection(**INIT_ARGS)
    output = model(**ARGS)
    print(output.shape)
    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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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)
attention Ran
ran, honoured the contract we drafted
this paper's copy was not recorded; identical code first harvested from WangYueFt/dcp
pointer only; the stored copy get_code serves, from WangYueFt/dcp, is pointer only (licence: NONE) · get_code("70ba9ecc422a6a48")
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_graph_feature Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("a4df418e59bd929e")
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; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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

torch.manual_seed(0)

# Create a small tensor for point cloud data
# Shape: (batch_size, num_dims, num_points)
# batch_size=2, num_dims=4, num_points=3
x = torch.randn(2, 4, 3)

# k parameter for k-nearest neighbors
k = 2

ARGS = {
    'x': x,
    'k': k
}
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 — no output shape is stated: observe_signature.py could not observe it (call:AssertionError)
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
knn Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("35d6887261f3e988")
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 int, shape (2, 3, 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

torch.manual_seed(0)

# x is used with:
# - transpose(2, 1) and contiguous() -> needs 3D tensor
# - sum(x ** 2, dim=1, keepdim=True) -> needs dim 1 to exist
# - matmul operations
# Shape: (batch, features, num_points) based on the transpose operations
x = torch.randn(2, 4, 3)

# k is used with topk(k=k, dim=-1)
# Should be a positive integer less than the last dimension size
k = 2

ARGS = {
    "x": x,
    "k": k,
}
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 int, shape (2, 3, 2) — shape (2, 3, 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: none by the harness; any seed is in the setup block
CustomTransformer Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("db755fe5ed8d6e0e")
How it ran

no invocation recorded.

MLPKernel Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("06d70af2876337b8")
How it ran

no invocation recorded.

MultilaterationHead Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("ef5dbe43a9efc444")
How it ran

no invocation recorded.

ResidualFlow_DiffEmbTransformer Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("8908af7a125f3b6b")
How it ran

no invocation recorded.

ResidualMLPHead Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("fe5f0050772bc142")
How it ran

no invocation recorded.

VNArgs Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("e77d7759eb725806")
How it ran

no invocation recorded.

VN_DGCNN Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("9c5f4a44ab78a987")
How it ran

no invocation recorded.

create_embedding_network Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("97b9e03315e765bc")
How it ran

no invocation recorded.

estimate_p Not yet run
unverified (harvested, no recorded run)
r-pad/taxpose/taxpose/nets/transformer_flow.py
code served (permissive licence) · get_code("92c2f28b7b961ce2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 22 samples lifted from this paper's linked repositories, 22 carry a function identity and 0 do not. Of the 22 distinct functions with an identity, 11 are also present in other papers whose code we hold (11 with at least one paper dated earlier, 0 the same month, 0 later) and 11 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
knn
brick of this paper's assembly: helper
5 other papers
5 dated earlier · 1 only as an assembly brick
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
2109.06619 (2021-09)
Sampling Network Guided Cross-Entropy Method for Unsupervised Point Cloud Registration
2112.12053 (2021-12)
Multi-View Partial (MVP) Point Cloud Challenge 2021 on Completion and Registration: Methods and Results
and 2 more not named here
EncoderLayer
brick of this paper's assembly: helper
3 other papers
3 dated earlier · 1 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
2203.07628 (2022-03)
P-STMO: Pre-Trained Spatial Temporal Many-to-One Model for 3D Human Pose Estimation
attention
brick of this paper's assembly: helper
3 other papers
3 dated earlier
1905.03304 (2019-05)
Deep Closest Point: Learning Representations for Point Cloud Registration
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
2110.02210 (2021-10)
Mix3D: Out-of-Context Data Augmentation for 3D Scenes
Decoder
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
DecoderLayer
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
Encoder
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
EncoderDecoder
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
MultiHeadedAttention
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
PositionwiseFeedForward
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
SublayerConnection
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
get_graph_feature
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2109.06619 (2021-09)
Sampling Network Guided Cross-Entropy Method for Unsupervised Point Cloud Registration

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

Many robot manipulation tasks can be framed as geometric reasoning tasks, where an agent must be able to precisely manipulate an object into a position that satisfies the task from a set of initial conditions. Often, task success is defined based on the relationship between two objects -for instance, hanging a mug on a rack. In such cases, the solution should be equivariant to the initial position of the objects as well as the agent, and invariant to the pose of the camera. This poses a challenge for learning systems which attempt to solve this task by learning directly from high-dimensional demonstrations: the agent must learn to be both equivariant as well as precise, which can be challenging without any inductive biases about the problem. In this work, we propose a method for precise relative pose prediction which is provably SE(3)-equivariant, can be learned from only a few demonstrations, and can generalize across variations in a class of objects. We accomplish this by factoring the problem into learning an SE(3) invariant task-specific representation of the scene and then interpreting this representation with novel geometric reasoning layers which are provably SE(3) equivariant. We demonstrate that our method can yield substantially more precise predictions in simulated placement tasks than previous methods trained with the same amount of data, and can accurately represent relative placement relationships data collected from real-world demonstrations. Supplementary information and videos can be found at this URL.

For agents

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

get_harvested_code_for_paper("2404.13478")
get_code_for_paper("2404.13478")
have("2404.13478")

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

curl https://syntology.ai/api/ran/2404.13478.json

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

[![Syntology run record](https://syntology.ai/api/ran/2404.13478.svg)](https://syntology.ai/paper/2404.13478)

Connect an agent — have() is free.