SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.05856 · 2024

POV: Prompt-Oriented View-Agnostic Learning for Egocentric Hand-Object Interaction in the Multi-View World

arXiv · PDF · Open in the Atlas

Code that ran

6 of 10 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 0 have no contract verdict; 5 failures of our instrument; 4 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/2403.05856.json · badge: /api/ran/2403.05856.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
xuboshen/pov_acmmm2023 canonical (classifier label)
classifier basis: census
6 ran · 4 no recorded run
FunctionStatusWhere it lives
attention_pool Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xuboshen/pov_acmmm2023/slowfast/models/attention_mprompt.py
code served (permissive licence) · get_code("e60a7360a40621b1")
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. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple pooling module that takes (B, C, T, H, W) and outputs pooled tensor
class SimplePool(nn.Module):
    def forward(self, x):
        # x shape: (B, C, T, H, W)
        # Simple average pooling: reduce spatial dimensions
        return x[:, :, ::2, ::2, ::2]  # Downsample by 2 in each dimension

# Create a simple normalization module
class SimpleNorm(nn.Module):
    def forward(self, x):
        # x shape: (B, N, L, C)
        return x

# Construct arguments
pool = SimplePool()
norm = SimpleNorm()

# Create a 3D tensor (will be unsqueezed to 4D in function)
# Shape: (B, L, C) where B=2, L=magic_number+spatial_tokens, C=4
magic_number = 5
B = 2
C = 4
T, H, W = 3, 2, 2  # Temporal, Height, Width dimensions
L = magic_number + (T * H * W)  # 1 cls token + 4 prompt tokens + spatial tokens

tensor = torch.randn(B, L, C)

# Pooling object
pool = SimplePool()

# Spatio-temporal shape (T, H, W)
thw_shape = [T, H, W]

# Magic number for prompt tokens
magic_number = 5

# Whether to include class embedding
has_cls_embed = True

# Normalization
norm = SimpleNorm()

ARGS = {
    "tensor": tensor,
    "pool": pool,
    "thw_shape": thw_shape,
    "magic_number": magic_number,
    "has_cls_embed": has_cls_embed,
    "norm": norm,
}
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 — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cal_rel_pos_spatial Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xuboshen/pov_acmmm2023/slowfast/models/attention_mprompt.py
code served (permissive licence) · get_code("d39331c5b5e58672")
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 4, dtype family float, shape (2, 4, 19, 19) 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.functional as F

torch.manual_seed(0)

# Set up dimensions
B = 2  # batch size
n_head = 4  # number of heads
dim = 8  # embedding dimension
q_t, q_h, q_w = 2, 3, 3  # query spatial dimensions
k_t, k_h, k_w = 2, 3, 3  # key spatial dimensions

# Create attention tensor
# Shape: [B, n_head, q_N, k_N] where q_N = 1 (cls) + q_t*q_h*q_w, k_N = 1 (cls) + k_t*k_h*k_w
q_N = 1 + q_t * q_h * q_w
k_N = 1 + k_t * k_h * k_w
attn = torch.randn(B, n_head, q_N, k_N)

# Create query and key tensors
# q shape: [B, n_head, q_N, dim]
q = torch.randn(B, n_head, q_N, dim)
k = torch.randn(B, n_head, k_N, dim)

# Create relative position embeddings
# For height: shape [ori_d, dim] where ori_d = 2*max(q_h, k_h) - 1
dh_ori = 2 * max(q_h, k_h) - 1
rel_pos_h = torch.randn(dh_ori, dim)

# For width: shape [ori_d, dim] where ori_d = 2*max(q_w, k_w) - 1
dw_ori = 2 * max(q_w, k_w) - 1
rel_pos_w = torch.randn(dw_ori, dim)

# Parameters
has_cls_embed = True
q_shape = (q_t, q_h, q_w)
k_shape = (k_t, k_h, k_w)
magic_number = 1  # sp_idx when has_cls_embed is True and prompt_enable is True

ARGS = {
    "attn": attn,
    "q": q,
    "k": k,
    "has_cls_embed": has_cls_embed,
    "q_shape": q_shape,
    "k_shape": k_shape,
    "rel_pos_h": rel_pos_h,
    "rel_pos_w": rel_pos_w,
    "magic_number": magic_number,
}
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 4, dtype family float, shape (2, 4, 19, 19) — shape (2, 4, 19, 19) 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_path Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xuboshen/pov_acmmm2023/slowfast/models/common.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from rajatmodi62/OccludedActionBenchmark, is pointer only (licence: NONE) · get_code("bcc1cdae3bb3212c")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from rajatmodi62/OccludedActionBenchmark, 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.

find_most_h_w Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xuboshen/pov_acmmm2023/slowfast/models/MultiscalePromptVit_stage2.py
code served (permissive licence) · get_code("fa1b9bbe882fd839")
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 1, dtype family int; 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)

# The function expects all_layer_matrices to be an iterable of tensors
# It accesses .shape[-2] and .shape[-1] for each tensor
# So we need a list of tensors with at least 2 dimensions

all_layer_matrices = [
    torch.randn(2, 4, 3, 5),  # shape[-2]=3, shape[-1]=5
    torch.randn(2, 4, 3, 5),  # shape[-2]=3, shape[-1]=5
    torch.randn(2, 4, 3, 5),  # shape[-2]=3, shape[-1]=5
]

ARGS = {
    "all_layer_matrices": all_layer_matrices
}
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 1, dtype family int — 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
get_gkern Ran
ran, honoured the contract we drafted
xuboshen/pov_acmmm2023/slowfast/models/utils.py
code served (permissive licence) · get_code("025b885a8f9114be")
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 (5, 5) 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 get_gkern takes two parameters:
# - kernlen: an integer specifying the kernel length
# - std: a float specifying the standard deviation for the Gaussian

ARGS = {
    "kernlen": 5,
    "std": 1.5
}
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 (5, 5) — shape (5, 5) 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_rel_pos Ran
ran; our contract draft was wrong, not the code
xuboshen/pov_acmmm2023/slowfast/models/attention_mprompt.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from rajatmodi62/OccludedActionBenchmark, is pointer only (licence: NONE) · get_code("5fc6c9c916e0ebc7")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from rajatmodi62/OccludedActionBenchmark, 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.

align_scale Not yet run
unverified (harvested, no recorded run)
xuboshen/pov_acmmm2023/slowfast/models/MultiscalePromptVit_stage2.py
code served (permissive licence) · get_code("4038555957e35d15")
How it ran

no invocation recorded.

get_3d_sincos_pos_embed Not yet run
unverified (harvested, no recorded run)
xuboshen/pov_acmmm2023/slowfast/models/utils.py
code served (permissive licence) · get_code("14f6b1125cb2fc10")
How it ran

no invocation recorded.

get_stem_func Not yet run
unverified (harvested, no recorded run)
xuboshen/pov_acmmm2023/slowfast/models/stem_helper.py
code served (permissive licence) · get_code("d296e9fcbb9a1e04")
How it ran

no invocation recorded.

resize_last_dim_to_most Not yet run
unverified (harvested, no recorded run)
xuboshen/pov_acmmm2023/slowfast/models/MultiscalePromptVit_stage2.py
code served (permissive licence) · get_code("a6b157934033010d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 7 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 3 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
drop_path 11 other papers
10 dated earlier, 1 later
2106.09703 (2021-06)
MaCLR: Motion-aware Contrastive Learning of Representations for Videos
2110.06915 (2021-10)
Object-Region Video Transformers
2110.07058 (2021-10)
Ego4D: Around the World in 3,000 Hours of Egocentric Video
and 8 more not named here
get_rel_pos 3 other papers
2 dated earlier, 1 later
2302.04869 (2023-02)
Reversible Vision Transformers
2401.11174 (2024-01)
Pixel-Wise Recognition for Holistic Surgical Scene Understanding
2410.19553 (2024-10)
On Occlusions in Video Action Detection: Benchmark Datasets And Training Recipes
get_3d_sincos_pos_embed 2 other papers
2 later
2406.18070 (2024-06)
EgoVideo: Exploring Egocentric Foundation Model and Downstream Adaptation
2503.14237 (2025-03)
Make Your Training Flexible: Towards Deployment-Efficient Video Models
align_scale 1 other paper
1 dated earlier
2104.11227 (2021-04)
Multiscale Vision Transformers
find_most_h_w 1 other paper
1 dated earlier
2104.11227 (2021-04)
Multiscale Vision Transformers
get_stem_func 1 other paper
1 dated earlier
2401.11174 (2024-01)
Pixel-Wise Recognition for Holistic Surgical Scene Understanding
resize_last_dim_to_most 1 other paper
1 dated earlier
2104.11227 (2021-04)
Multiscale Vision Transformers

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

We humans are good at translating third-person observations of hand-object interactions (HOI) into an egocentric view. However, current methods struggle to replicate this ability of view adaptation from third-person to first-person. Although some approaches attempt to learn view-agnostic representation from large-scale video datasets, they ignore the relationships among multiple third-person views. To this end, we propose a Prompt-Oriented View-agnostic learning (POV) framework in this paper, which enables this view adaptation with few egocentric videos. Specifically, We introduce interactive masking prompts at the frame level to capture fine-grained action information, and view-aware prompts at the token level to learn view-agnostic representation. To verify our method, we establish two benchmarks for transferring from multiple third-person views to the egocentric view. Our extensive experiments on these benchmarks demonstrate the efficiency and effectiveness of our POV framework and prompt tuning techniques in terms of view adaptation and view generalization. Our code is available at \url{https://github.com/xuboshen/pov_acmmm2023}.

For agents

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

get_harvested_code_for_paper("2403.05856")
get_code_for_paper("2403.05856")
have("2403.05856")

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

curl https://syntology.ai/api/ran/2403.05856.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.05856.svg)](https://syntology.ai/paper/2403.05856)

Connect an agent — have() is free.