SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2501.07256 · CVPR · 2025

EdgeTAM: On-Device Track Anything Model

Bo Dai, Chen Change Loy, Vikas Chandra, Yunyang Xiong, Raghuraman Krishnamoorthi, Bilge Soran, Saksham Suri, Chong Zhou, Fanyi Xiao, Lemeng Wu, Chenchen Zhu

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 1 violated it and 2 have no contract verdict; 3 failures of our instrument; 1 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/2501.07256.json · badge: /api/ran/2501.07256.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
facebookresearch/sam2 community
Papers with Code archive
7 ran · 1 no recorded run
FunctionStatusWhere it lives
Attention Ran
ran on a synthesized input · constructed an object; no result was computed
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("c140570681a00c3c")
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 an instance of its own class, Attention: the run constructed the object and did not call it, so no result was computed. 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 import nn

torch.manual_seed(0)

ARGS = {
    "embedding_dim": 8,
    "num_heads": 2,
    "downsample_rate": 1,
    "dropout": 0.1,
    "kv_in_dim": 8,
}
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 Attention — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Attention 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
RoPEAttention Ran
ran on a synthesized input · constructed an object; no result was computed
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("4f347f5c9d9e7597")
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 an instance of its own class, RoPEAttention: the run constructed the object and did not call it, so no result was computed. 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 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

torch.manual_seed(0)

ARGS = {
    'embedding_dim': 8,
    'num_heads': 2,
    'downsample_rate': 1,
    'dropout': 0.1,
    'kv_in_dim': 8,
    'rope_theta': 10000.0,
    'rope_k_repeat': False,
    'feat_sizes': (4, 4),
}
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 RoPEAttention — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RoPEAttention 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
apply_rotary_enc Ran
ran; our contract draft was wrong, not the code
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("844d836b99dc1aa7")
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 tuple, ndim 5, 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

torch.manual_seed(0)

# xq and xk are query and key tensors
# They need to have shape (..., seq_len, dim) where dim is even (for complex view)
# freqs_cis has shape (seq_len, dim//2) representing complex frequencies

# Create small tensors
batch_size = 2
seq_len_q = 4
seq_len_k = 4
dim = 8  # must be even for view_as_complex

# xq: (batch, seq_len_q, dim)
xq = torch.randn(batch_size, seq_len_q, dim)

# xk: (batch, seq_len_k, dim)
xk = torch.randn(batch_size, seq_len_k, dim)

# freqs_cis: (seq_len_q, dim//2) - represents rotary frequencies as complex numbers
# Shape should be (x.shape[-2], x.shape[-1]) where x is reshaped to (..., -1, 2)
# After reshape xq to (..., dim//2, 2), we need freqs_cis to be (seq_len_q, dim//2)
freqs_cis = torch.randn(seq_len_q, dim // 2, dtype=torch.complex64)

ARGS = {
    "xq": xq,
    "xk": xk,
    "freqs_cis": freqs_cis,
    "repeat_freqs_k": False,
}
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 tuple, ndim 5, dtype family float — a tuple output has no array shape of its own; none is stated
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
compute_axial_cis Ran
ran, violated the contract we drafted
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("7ceaad42da42edd5")
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 2, dtype family float, shape (12, 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

torch.manual_seed(0)

ARGS = {
    "dim": 8,
    "end_x": 4,
    "end_y": 3,
    "theta": 10000.0,
}
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 2, dtype family float, shape (12, 4) — shape (12, 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: none by the harness; any seed is in the setup block
get_clones Ran
ran; our contract draft was wrong, not the code
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("3ef3feb872840557")
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 ModuleList. 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 import nn
import copy

torch.manual_seed(0)

# Create a simple neural network module
class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Instantiate the module
module = SimpleModule()

# Define number of clones to create
N = 5

# Construct ARGS dict
ARGS = {
    "module": module,
    "N": N
}
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 ModuleList — a ModuleList output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ModuleList output the recorded ndim 1 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_t_xy Ran
ran, honoured the contract we drafted
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("bb060d29525f90eb")
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 tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    'end_x': 4,
    'end_y': 3,
}
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 tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
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
reshape_for_broadcast Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("f8f09f78b474ee7c")
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 3, dtype family float, shape (1, 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

torch.manual_seed(0)

# The function expects:
# - freqs_cis: a 2D tensor with shape (x.shape[-2], x.shape[-1])
# - x: a tensor with ndim >= 2

# Let's create x with shape (batch, seq_len, hidden_dim) = (2, 3, 4)
# This means x.shape[-2] = 3 and x.shape[-1] = 4
x = torch.randn(2, 3, 4)

# freqs_cis should have shape (3, 4) to match (x.shape[-2], x.shape[-1])
freqs_cis = torch.randn(3, 4)

ARGS = {
    "freqs_cis": freqs_cis,
    "x": x,
}
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 3, dtype family float, shape (1, 3, 4) — shape (1, 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: none by the harness; any seed is in the setup block
MemoryAttention Not yet run
unverified (harvested, no recorded run)
facebookresearch/sam2/sam2/modeling/memory_attention.py
code served (permissive licence) · get_code("dcd56b84a90a2922")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 5 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 3 the same month, 4 later) and 3 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
get_clones
brick of this paper's assembly: helper
40 other papers
36 dated earlier, 4 later · 8 only as an assembly brick
1905.12794 (2019-05)
Fashion IQ: A New Dataset Towards Retrieving Images by Natural Language Feedback
2002.07033 (2020-02)
Towards an Appropriate Query, Key, and Value Computation for Knowledge Tracing
2010.12042 (2020-10)
SAINT+: Integrating Temporal Features for EdNet Correctness Prediction
and 37 more not named here
compute_axial_cis
brick of this paper's assembly: helper
13 other papers
9 dated earlier, 1 the same month, 3 later
2302.01872 (2023-02)
MOSE: A New Dataset for Video Object Segmentation in Complex Scenes
2306.01567 (2023-06)
Segment Anything in High Quality
2408.00874 (2024-08)
Medical SAM 2: Segment medical images as video via Segment Anything Model 2
and 10 more not named here
init_t_xy
brick of this paper's assembly: helper
13 other papers
9 dated earlier, 1 the same month, 3 later
2302.01872 (2023-02)
MOSE: A New Dataset for Video Object Segmentation in Complex Scenes
2306.01567 (2023-06)
Segment Anything in High Quality
2408.00874 (2024-08)
Medical SAM 2: Segment medical images as video via Segment Anything Model 2
and 10 more not named here
reshape_for_broadcast
brick of this paper's assembly: helper
13 other papers
9 dated earlier, 1 the same month, 3 later
2302.01872 (2023-02)
MOSE: A New Dataset for Video Object Segmentation in Complex Scenes
2306.01567 (2023-06)
Segment Anything in High Quality
2408.00874 (2024-08)
Medical SAM 2: Segment medical images as video via Segment Anything Model 2
and 10 more not named here
Attention
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2408.00714 (2024-08)
SAM 2: Segment Anything in Images and Videos

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

On top of Segment Anything Model (SAM), SAM 2 further extends its capability from image to video inputs through a memory bank mechanism and obtains a remarkable performance compared with previous methods, making it a foundation model for video segmentation task. In this paper, we aim at making SAM 2 much more efficient so that it even runs on mobile devices while maintaining a comparable performance. Despite several works optimizing SAM for better efficiency, we find they are not sufficient for SAM 2 because they all focus on compressing the image encoder, while our benchmark shows that the newly introduced memory attention blocks are also the latency bottleneck. Given this observation, we propose EdgeTAM, which leverages a novel 2D Spatial Perceiver to reduce the computational cost. In particular, the proposed 2D Spatial Perceiver encodes the densely stored frame-level memories with a lightweight Transformer that contains a fixed set of learnable queries. Given that video segmentation is a dense prediction task, we find preserving the spatial structure of the memories is essential so that the queries are split into global-level and patch-level groups. We also propose a distillation pipeline that further improves the performance without inference overhead. As a result, EdgeTAM achieves 87.7, 70.0, 72.3, and 71.7 J &F on DAVIS 2017, MOSE, SA-V val, and SA-V test, while running at 16 FPS on iPhone 15 Pro Max. The code and models are available here.

For agents

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

get_harvested_code_for_paper("2501.07256")
get_code_for_paper("2501.07256")
have("2501.07256")

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

curl https://syntology.ai/api/ran/2501.07256.json

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

[![Syntology run record](https://syntology.ai/api/ran/2501.07256.svg)](https://syntology.ai/paper/2501.07256)

Connect an agent — have() is free.