SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.00670 · ICML · 2024

Cascade-CLIP: Cascaded Vision-Language Embeddings Alignment for Zero-Shot Semantic Segmentation

Ming-Ming Cheng, Qibin Hou, Yunheng Li, Zhong-Yu Li, Quansheng Zeng

arXiv · PDF · Open in the Atlas

Code that ran

7 of 7 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; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 0 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/2406.00670.json · badge: /api/ran/2406.00670.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
hvision-nku/cascade-clip canonical (classifier label)
classifier basis: census
7 ran · 0 no recorded run
FunctionStatusWhere it lives
cosine_margin_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
hvision-nku/cascade-clip/models/losses/criterion.py
code served (permissive licence) · get_code("864ff812c98727df")
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 0, dtype family float, shape () 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)

# Analyze the function signature and body:
# - q: shape (bs, n_cls, n_dim) - query tensor
# - e: shape (n_cls + 1, n_dim) - embedding tensor
# - labels: shape (bs * n_cls,) - long integer labels for indexing e
# - tau: float, temperature parameter
# - m: float, margin parameter

bs = 2
n_cls = 3
n_dim = 4

# q: query tensor of shape (bs, n_cls, n_dim)
q = torch.randn(bs, n_cls, n_dim)

# e: embedding tensor of shape (n_cls + 1, n_dim)
# The assertion checks that q.shape[1] + 1 == e.shape[0]
# So e.shape[0] must be n_cls + 1 = 4
e = torch.randn(n_cls + 1, n_dim)

# labels: long tensor used to index e
# After reshape, q becomes (bs * n_cls, n_dim) = (6, 4)
# labels[labels.long()] is used to index e, so labels should have shape (bs * n_cls,)
# and values should be valid indices in range [0, n_cls)
labels = torch.tensor([0, 1, 2, 0, 1, 2], dtype=torch.long)

# tau and m are float parameters
tau = 1.0
m = 0.5

ARGS = {
    "q": q,
    "e": e,
    "labels": labels,
    "tau": tau,
    "m": m
}
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 0, dtype family float, shape () — shape () 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
dice_loss Ran
ran; our contract draft was wrong, not the code
hvision-nku/cascade-clip/models/losses/criterion.py
code served (permissive licence) · get_code("016dba4163da28df")
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 0, dtype family float, shape () 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 small tensors for inputs and targets
# inputs: arbitrary shape float tensor (predictions)
# targets: same shape as inputs, binary labels (0 or 1)
batch_size = 2
num_elements = 4

inputs = torch.randn(batch_size, num_elements, dtype=torch.float32)
targets = torch.randint(0, 2, (batch_size, num_elements), dtype=torch.float32)

# num_masks: integer representing number of masks
num_masks = 2

ARGS = {
    "inputs": inputs,
    "targets": targets,
    "num_masks": num_masks,
}
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 0, dtype family float, shape () — shape () 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
gaussian Ran
ran; our contract draft was wrong, not the code
hvision-nku/cascade-clip/models/decode_heads/decode_seg.py
code served (permissive licence) · get_code("39f08d24e0c8725c")
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 (2, 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)

# Construct arguments for gaussian(x, mu, sigma)
x = torch.randn(2, 4)
mu = torch.randn(2, 4)
sigma = torch.randn(2, 4)

ARGS = {
    "x": x,
    "mu": mu,
    "sigma": sigma,
}
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 (2, 4) — shape (2, 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_angles Ran
ran, honoured the contract we drafted
hvision-nku/cascade-clip/models/decode_heads/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from chihchenghsieh/eventlogdice, is pointer only (licence: NONE) · get_code("7816c6d65ab55d8f")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from chihchenghsieh/eventlogdice, 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.

nested_tensor_from_tensor_list Ran
ran; our contract draft was wrong, not the code
hvision-nku/cascade-clip/models/losses/misc.py
code served (permissive licence) · get_code("58cc9ff3bf75e753")
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 NestedTensor. 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.functional as F
from typing import List

torch.manual_seed(0)

# Create a list of 3D tensors with varying shapes
# Each tensor has shape (C, H, W) where C is channels, H is height, W is width
tensor1 = torch.randn(3, 4, 5)  # 3 channels, 4 height, 5 width
tensor2 = torch.randn(3, 6, 7)  # 3 channels, 6 height, 7 width
tensor3 = torch.randn(3, 5, 8)  # 3 channels, 5 height, 8 width

tensor_list = [tensor1, tensor2, tensor3]

ARGS = {
    "tensor_list": tensor_list
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NestedTensor — a NestedTensor output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NestedTensor 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
positional_encoding Ran
ran, honoured the contract we drafted
hvision-nku/cascade-clip/models/decode_heads/utils.py
code served (permissive licence) · get_code("da3acf7173bd84f0")
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 (1, 8, 16) 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 = {
    "position": 8,
    "d_model": 16
}
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 (1, 8, 16) — shape (1, 8, 16) 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
sigmoid_focal_loss Ran
ran; our contract draft was wrong, not the code
hvision-nku/cascade-clip/models/losses/criterion.py
code served (permissive licence) · get_code("2675e89c94ae08bf")
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 0, dtype family float, shape () 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 small tensors for inputs and targets
# inputs: predictions (logits) - shape [batch_size, num_elements]
# targets: binary labels - same shape as inputs
batch_size = 2
num_elements = 4

inputs = torch.randn(batch_size, num_elements)
targets = torch.randint(0, 2, (batch_size, num_elements)).float()

# num_masks: used as divisor in the loss computation
num_masks = 2

# alpha and gamma are already specified with defaults in the function signature
alpha = 0.25
gamma = 2.0

ARGS = {
    'inputs': inputs,
    'targets': targets,
    'num_masks': num_masks,
    'alpha': alpha,
    'gamma': gamma
}
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 0, dtype family float, shape () — shape () 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

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 1 later) and 1 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
nested_tensor_from_tensor_list 55 other papers
43 dated earlier, 10 later, 2 same year, order not determinable · 17 only as an assembly brick
1608.03983 (2016-08)
SGDR: Stochastic Gradient Descent with Warm Restarts
2005.12872 (2020-05)
End-to-End Object Detection with Transformers
2101.01909 (2021-01)
Line Segment Detection Using Transformers without Edges
and 52 more not named here
get_angles 14 other papers
14 dated earlier · 2 only as an assembly brick
1704.04368 (2017-04)
Get To The Point: Summarization with Pointer-Generator Networks
1706.03762 (2017-06)
Attention Is All You Need
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
and 11 more not named here
dice_loss 5 other papers
5 dated earlier
2211.16799 (2022-11)
NOPE-SAC: Neural One-Plane RANSAC for Sparse-View Planar 3D Reconstruction
2212.03588 (2022-12)
ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentation
2308.01045 (2023-08)
Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation
and 2 more not named here
positional_encoding 3 other papers
3 dated earlier
2212.03588 (2022-12)
ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentation
2312.12754 (2023-12)
Spectral Prompt Tuning:Unveiling Unseen Classes for Zero-Shot Semantic Segmentation
2403.14183 (2024-03)
OTSeg: Multi-prompt Sinkhorn Attention for Zero-Shot Semantic Segmentation
cosine_margin_loss 2 other papers
2 dated earlier
2212.03588 (2022-12)
ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentation
2403.14183 (2024-03)
OTSeg: Multi-prompt Sinkhorn Attention for Zero-Shot Semantic Segmentation
sigmoid_focal_loss 2 other papers
2 dated earlier
2212.03588 (2022-12)
ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentation
2403.14183 (2024-03)
OTSeg: Multi-prompt Sinkhorn Attention for Zero-Shot Semantic Segmentation

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

Pre-trained vision-language models, e.g., CLIP, have been successfully applied to zero-shot semantic segmentation. Existing CLIP-based approaches primarily utilize visual features from the last layer to align with text embeddings, while they neglect the crucial information in intermediate layers that contain rich object details. However, we find that directly aggregating the multilevel visual features weakens the zero-shot ability for novel classes. The large differences between the visual features from different layers make these features hard to align well with the text embeddings. We resolve this problem by introducing a series of independent decoders to align the multi-level visual features with the text embeddings in a cascaded way, forming a novel but simple framework named Cascade-CLIP. Our Cascade-CLIP is flexible and can be easily applied to existing zero-shot semantic segmentation methods. Experimental results show that our simple Cascade-CLIP achieves superior zero-shot performance on segmentation benchmarks, like COCO-Stuff, Pascal-VOC, and Pascal-Context. Our code is available at https://github. com/HVision-NKU/Cascade-CLIP.

For agents

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

get_harvested_code_for_paper("2406.00670")
get_code_for_paper("2406.00670")
have("2406.00670")

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

curl https://syntology.ai/api/ran/2406.00670.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.00670.svg)](https://syntology.ai/paper/2406.00670)

Connect an agent — have() is free.