SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2104.00563 · 2021

Latent Variable Sequential Set Transformers For Joint Multi-Agent Motion Prediction

arXiv · PDF · Open in the Atlas

Code that ran

11 of 17 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; 3 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 7 failures of our instrument; 6 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/2104.00563.json · badge: /api/ran/2104.00563.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
roggirg/AutoBots canonical (classifier label)
classifier basis: census
11 ran · 6 no recorded run
FunctionStatusWhere it lives
angle_of_rotation Ran
ran, honoured the contract we drafted
roggirg/AutoBots/datasets/argoverse/create_h5_argo.py
code served (permissive licence) · get_code("db4fd23bb7142d4b")
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, through the batch API) and returned float64, 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 numpy as np

# The function takes a single yaw parameter (a float representing an angle in radians)
# and returns a float representing the rotation angle needed

ARGS = {
    "yaw": 0.5  # A reasonable yaw angle in radians (about 28.6 degrees)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
convert_global_coords_to_local Ran
ran, violated the contract we drafted
roggirg/AutoBots/datasets/argoverse/create_h5_argo.py
code served (permissive licence) · get_code("4efe4a6acb3051fb")
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, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (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 numpy as np

ARGS = {
    "coordinates": np.array([
        [1.0, 2.0],
        [3.0, 4.0],
        [5.0, 6.0]
    ]),
    "translation": (0.5, 1.5),
    "yaw": 0.785398  # approximately pi/4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (3, 2) — shape (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
create_results_folder Ran
ran; our contract draft was wrong, not the code
roggirg/AutoBots/process_args.py
code served (permissive licence) · get_code("b61e662d9f8a3be9")
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, through the batch API) and returned str. 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 os
import tempfile

class Args:
    def __init__(self):
        self.model_type = 'Joint'
        self.num_modes = 6
        self.hidden_size = 128
        self.num_encoder_layers = 2
        self.num_decoder_layers = 2
        self.tx_hidden_size = 256
        self.tx_num_heads = 4
        self.entropy_weight = 0.1
        self.kl_weight = 0.5
        self.use_FDEADE_aux_loss = True
        self.use_map_image = True
        self.use_map_lanes = False
        self.exp_id = 'test_exp'
        self.seed = 42
        self.save_dir = tempfile.gettempdir()
        self.dataset = 'argoverse'

ARGS = {'args': Args()}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block
drop_distant Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
roggirg/AutoBots/datasets/trajnetpp/create_data_npys.py
code served (permissive licence) · get_code("3dba06ecf04eff9a")
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, through the batch API) and returned ndarray, ndim 3, dtype family float, shape (4, 5, 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 numpy as np

# Construct a small example for drop_distant
# The function expects xy to be a 3D array with shape (time_steps, num_pedestrians, 2)
# where the last dimension contains (x, y) coordinates

# Create a small sample with shape (4, 6, 2):
# - 4 time steps
# - 6 pedestrians (we'll keep 5 closest)
# - 2 coordinates (x, y)
np.random.seed(0)
xy = np.random.randn(4, 6, 2).astype(np.float32)

# Add some NaN values to test nanmin behavior
xy[1, 2, :] = np.nan

ARGS = {
    "xy": xy,
    "max_num_peds": 5,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (4, 5, 2) — shape (4, 5, 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
drop_inactive Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
roggirg/AutoBots/datasets/trajnetpp/create_data_npys.py
code served (permissive licence) · get_code("577c3685905bbabc")
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, through the batch API) and returned ndarray, ndim 3, dtype family float, shape (10, 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 numpy as np

# Create a sample xy array
# Shape: (time_steps, num_agents, features)
# obs_horizon=9, so we need at least 9 timesteps
# Let's use 10 timesteps, 4 agents, 2 features (x, y)

xy = np.random.randn(10, 4, 2).astype(np.float32)

# Make some agents inactive (NaN values) at different timesteps
# but keep at least some agents active at timestep obs_horizon-1 (timestep 8)
xy[8, 0, 0] = np.nan  # Agent 0 inactive at timestep 8
xy[8, 1, 0] = 1.0     # Agent 1 active at timestep 8
xy[8, 2, 0] = np.nan  # Agent 2 inactive at timestep 8
xy[8, 3, 0] = 2.0     # Agent 3 active at timestep 8

ARGS = {
    "xy": xy,
    "obs_horizon": 9
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (10, 2, 2) — shape (10, 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: none by the harness; any seed is in the setup block
get_BVG_distributions Ran
ran; our contract draft was wrong, not the code
roggirg/AutoBots/utils/train_helpers.py
code served (permissive licence) · get_code("519ae373e4e4d7cd")
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, through the batch API) and returned MultivariateNormal. 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
from torch.distributions import MultivariateNormal

torch.manual_seed(0)

# The function expects pred to be a tensor with shape (B, T, 5)
# where the last dimension contains: [mu_x, mu_y, sigma_x, sigma_y, rho]
# B = batch size, T = time steps

B = 2
T = 3
pred = torch.randn(B, T, 5)

# Ensure sigma_x and sigma_y are positive (they're squared in covariance)
pred[:, :, 2] = torch.abs(pred[:, :, 2]) + 0.1
pred[:, :, 3] = torch.abs(pred[:, :, 3]) + 0.1

# Ensure rho is in valid range [-1, 1] for correlation
pred[:, :, 4] = torch.tanh(pred[:, :, 4])

ARGS = {
    "pred": pred
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MultivariateNormal — a MultivariateNormal output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MultivariateNormal output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block
init Ran
ran; our contract draft was wrong, not the code
roggirg/AutoBots/models/autobot_joint.py
pointer only (licence: BSD-3-Clause); the stored copy get_code serves, from ArnaudFickinger/adversarial-surprise, is pointer only (licence: NONE) · get_code("9de355e93051e4ad")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (BSD-3-Clause), but get_code serves the stored copy, first harvested from ArnaudFickinger/adversarial-surprise, and GitHub detected no licence in that repository when it was checked. 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.

init Ran
ran; our contract draft was wrong, not the code
roggirg/AutoBots/models/autobot_ego.py
pointer only (licence: BSD-3-Clause); the stored copy get_code serves, from vita-epfl/UniTraj, is pointer only (licence: NOASSERTION) · get_code("d7a51202f8456490")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (BSD-3-Clause), but get_code serves the stored copy, first harvested from vita-epfl/UniTraj, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

interpolate_trajectories Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
roggirg/AutoBots/utils/metric_helpers.py
code served (permissive licence) · get_code("4e0c5fccd80c3976")
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, through the batch API) and returned Tensor, ndim 5, dtype family float, shape (2, 6, 2, 4, 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)

# Create a small tensor with shape (K, T_in, B, N, out)
# K=2 (num modes/predictions)
# T_in=3 (input timesteps)
# B=2 (batch size)
# N=4 (num agents)
# out=3 (output dim: x, y, yaw)
preds = torch.randn(2, 3, 2, 4, 3)

ARGS = {
    "preds": preds
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 5, dtype family float, shape (2, 6, 2, 4, 3) — shape (2, 6, 2, 4, 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
make_2d_rotation_matrix Ran
ran, honoured the contract we drafted
roggirg/AutoBots/datasets/argoverse/create_h5_argo.py
code served (permissive licence) · get_code("cd4282007f2ad836")
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, through the batch API) and returned ndarray, 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 numpy as np

ARGS = {
    "angle_in_radians": 0.5,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, 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: none by the harness; any seed is in the setup block
shift Ran
ran, honoured the contract we drafted
roggirg/AutoBots/datasets/trajnetpp/create_data_npys.py
code served (permissive licence) · get_code("1761d9ef84caa582")
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, through the batch API) and returned ndarray, ndim 4, dtype family float, shape (2, 3, 4, 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 numpy as np

# The function expects:
# - xy: a 3D numpy array (batch, height, width, channels)
# - center: a 1D numpy array or array-like with shape (channels,)

# Create a small xy array with shape (batch=2, height=3, width=4, channels=2)
xy = np.random.randn(2, 3, 4, 2).astype(np.float32)

# Create a center array with shape (2,) to match the last dimension of xy
center = np.array([0.5, 1.0], dtype=np.float32)

ARGS = {
    "xy": xy,
    "center": center,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family float, shape (2, 3, 4, 2) — shape (2, 3, 4, 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
get_Laplace_dist Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/utils/train_helpers.py
code served (permissive licence) · get_code("5112337a26b088f3")
How it ran

no invocation recorded.

load_config Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/process_args.py
code served (permissive licence) · get_code("3eb3d4a61fbb224d")
How it ran

no invocation recorded.

min_xde_K Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/utils/metric_helpers.py
code served (permissive licence) · get_code("b10ff2bd7a695f93")
How it ran

no invocation recorded.

nll_pytorch_dist Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/utils/train_helpers.py
code served (permissive licence) · get_code("2e9627477244b6e9")
How it ran

no invocation recorded.

recompute_probs Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/useful_scripts/generate_nuscene_results.py
code served (permissive licence) · get_code("29e4fbe72133e718")
How it ran

no invocation recorded.

yaw_from_predictions Not yet run
unverified (harvested, no recorded run)
roggirg/AutoBots/utils/metric_helpers.py
code served (permissive licence) · get_code("21100d7d06393031")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 1 later) and 15 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
init 34 other papers
12 dated earlier, 1 the same month, 21 later · 5 only as an assembly brick
1705.05363 (2017-05)
Curiosity-driven Exploration by Self-supervised Prediction
1706.02275 (2017-06)
Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments
1810.12894 (2018-10)
Exploration by Random Network Distillation
and 31 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

Robust multi-agent trajectory prediction is essential for the safe control of robotic systems. A major challenge is to efficiently learn a representation that approximates the true joint distribution of contextual, social, and temporal information to enable planning. We propose Latent Variable Sequential Set Transformers which are encoder-decoder architectures that generate scene-consistent multi-agent trajectories. We refer to these architectures as "AutoBots". The encoder is a stack of interleaved temporal and social multi-head self-attention (MHSA) modules which alternately perform equivariant processing across the temporal and social dimensions. The decoder employs learnable seed parameters in combination with temporal and social MHSA modules allowing it to perform inference over the entire future scene in a single forward pass efficiently. AutoBots can produce either the trajectory of one ego-agent or a distribution over the future trajectories for all agents in the scene. For the single-agent prediction case, our model achieves top results on the global nuScenes vehicle motion prediction leaderboard, and produces strong results on the Argoverse vehicle prediction challenge. In the multi-agent setting, we evaluate on the synthetic partition of TrajNet++ dataset to showcase the model's socially-consistent predictions. We also demonstrate our model on general sequences of sets and provide illustrative experiments modelling the sequential structure of the multiple strokes that make up symbols in the Omniglot data. A distinguishing feature of AutoBots is that all models are trainable on a single desktop GPU (1080 Ti) in under 48h.

For agents

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

get_harvested_code_for_paper("2104.00563")
get_code_for_paper("2104.00563")
have("2104.00563")

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

curl https://syntology.ai/api/ran/2104.00563.json

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

[![Syntology run record](https://syntology.ai/api/ran/2104.00563.svg)](https://syntology.ai/paper/2104.00563)

Connect an agent — have() is free.