SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.11335 · ECCV · 2024

LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction

Yan Wang, Yu Wang, Errui Ding, Jingdong Wang, Yue Liao, Si Liu, Yifan Sun, Gang Zhang, Luting Wang, Penghui Du

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 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; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 1 failure 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/2407.11335.json · badge: /api/ran/2407.11335.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
eternaldolphin/LaMI-DETR community
Papers with Code archive
3 ran · 1 no recorded run
FunctionStatusWhere it lives
MultiScaleDeformableAttention Ran
ran on a synthesized input · constructed an object; no result was computed
eternaldolphin/LaMI-DETR/lami_detr/modeling/lami_transformer.py
code served (permissive licence) · get_code("d89ee7ec1de8fb4e")
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, MultiScaleDeformableAttention: 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
import torch.nn as nn

torch.manual_seed(0)

# MultiScaleDeformableAttention constructor parameters
ARGS = {
    'embed_dim': 256,
    'num_heads': 8,
    'num_levels': 4,
    'num_points': 4,
    'img2col_step': 64,
    'dropout': 0.1,
    'batch_first': 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 MultiScaleDeformableAttention — 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 MultiScaleDeformableAttention 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
MultiScaleDeformableAttnFunction Ran
ran on a synthesized input · constructed an object; no result was computed
eternaldolphin/LaMI-DETR/lami_detr/modeling/lami_transformer.py
code served (permissive licence) · get_code("835cb830bd2670e2")
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, MultiScaleDeformableAttnFunction: 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
import torch.nn as nn

torch.manual_seed(0)

# Create tiny tensors for multi-scale deformable attention
# value: [batch, seq_len, num_heads, head_dim]
value = torch.randn(2, 16, 4, 4, requires_grad=True)

# value_spatial_shapes: [num_levels, 2] - spatial dimensions per level (H, W)
value_spatial_shapes = torch.tensor([[8, 8], [4, 4]], dtype=torch.long)

# value_level_start_index: [num_levels] - cumulative index for each level
value_level_start_index = torch.tensor([0, 64], dtype=torch.long)

# sampling_locations: [batch, seq_len, num_heads, num_levels, num_points, 2]
# Coordinates normalized to [0, 1]
sampling_locations = torch.rand(2, 16, 4, 2, 4, 2, requires_grad=True)

# attention_weights: [batch, seq_len, num_heads, num_levels, num_points]
attention_weights = torch.rand(2, 16, 4, 2, 4, requires_grad=True)

# im2col_step: integer parameter for column-to-image conversion step size
im2col_step = 128

ARGS = {
    'value': value,
    'value_spatial_shapes': value_spatial_shapes,
    'value_level_start_index': value_level_start_index,
    'sampling_locations': sampling_locations,
    'attention_weights': attention_weights,
    'im2col_step': im2col_step
}
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 MultiScaleDeformableAttnFunction — 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 MultiScaleDeformableAttnFunction 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
multi_scale_deformable_attn_pytorch Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
eternaldolphin/LaMI-DETR/lami_detr/modeling/lami_transformer.py
code served (permissive licence) · get_code("be924e54ed79731b")
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 (2, 3, 32) 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)

# Parameters
bs = 2  # batch size
num_heads = 4
embed_dims = 8
num_queries = 3
num_levels = 2
num_points = 5

# value_spatial_shapes: tensor of shape (num_levels, 2) with [H, W] for each level
value_spatial_shapes = torch.tensor([
    [4, 5],  # level 0: 4x5
    [2, 3]   # level 1: 2x3
])

# Compute total spatial positions
total_spatial_positions = (value_spatial_shapes[:, 0] * value_spatial_shapes[:, 1]).sum().item()

# value: shape (bs, total_spatial_positions, num_heads, embed_dims)
value = torch.randn(bs, total_spatial_positions, num_heads, embed_dims)

# sampling_locations: shape (bs, num_queries, num_heads, num_levels, num_points, 2)
# Each location is normalized to [-1, 1]
sampling_locations = torch.rand(bs, num_queries, num_heads, num_levels, num_points, 2) * 2 - 1

# attention_weights: shape (bs, num_queries, num_heads, num_levels, num_points)
attention_weights = torch.rand(bs, num_queries, num_heads, num_levels, num_points)

ARGS = {
    "value": value,
    "value_spatial_shapes": value_spatial_shapes,
    "sampling_locations": sampling_locations,
    "attention_weights": attention_weights
}
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 (2, 3, 32) — shape (2, 3, 32) 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
LaMITransformer Not yet run
unverified (harvested, no recorded run)
eternaldolphin/LaMI-DETR/lami_detr/modeling/lami_transformer.py
code served (permissive licence) · get_code("6d47f8899ba27e4a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 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
multi_scale_deformable_attn_pytorch
brick of this paper's assembly: helper
6 other papers
2 dated earlier, 3 later, 1 same year, order not determinable
2312.02153 (2023-12)
Aligning and Prompting Everything All at Once for Universal Visual Perception
Shen_Aligning_and_Prompting_Everything_All_at_Once_for_Universal_Visual_CVPR_2024_paper (2024 (year only))
Aligning and Prompting Everything All at Once for Universal Visual Perception
2401.14159 (2024-01)
Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks
and 3 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

Existing methods enhance open-vocabulary object detection by leveraging the robust open-vocabulary recognition capabilities of Vision-Language Models (VLMs), such as CLIP. However, two main challenges emerge: (1) A deficiency in concept representation, where the category names in CLIP's text space lack textual and visual knowledge. (2) An overfitting tendency towards base categories, with the open vocabulary knowledge biased towards base categories during the transfer from VLMs to detectors. To address these challenges, we propose the Language Model Instruction (LaMI) strategy, which leverages the relationships between visual concepts and applies them within a simple yet effective DETR-like detector, termed LaMI-DETR. LaMI utilizes GPT to construct visual concepts and employs T5 to investigate visual similarities across categories. These inter-category relationships refine concept representation and avoid overfitting to base categories. Comprehensive experiments validate our approach's superior performance over existing methods in the same rigorous setting without reliance on external training resources. LaMI-DETR achieves a rare box AP of 43.4 on OV-LVIS, surpassing the previous best by 7.8 rare box AP.

For agents

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

get_harvested_code_for_paper("2407.11335")
get_code_for_paper("2407.11335")
have("2407.11335")

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

curl https://syntology.ai/api/ran/2407.11335.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.11335.svg)](https://syntology.ai/paper/2407.11335)

Connect an agent — have() is free.