SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2301.12959 · CVPR · 2023

GALIP: Generative Adversarial CLIPs for Text-to-Image Synthesis

Hao Tang, Changsheng Xu, Ming Tao, Bing-Kun Bao

arXiv · PDF · Open in the Atlas

Code that ran

7 of 11 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; 0 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 1 failure 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/2301.12959.json · badge: /api/ran/2301.12959.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
tobran/GALIP community
Papers with Code archive
6 ran · 3 no recorded run
tobran/DF-GAN community
Papers with Code archive
1 ran · 1 no recorded run
FunctionStatusWhere it lives
Affine Ran
ran on a synthesized input
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("dcdfeb32d64cbd7d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, dtype family float, shape (2, 4, 3, 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
import torch.nn as nn
from collections import OrderedDict

class Affine(nn.Module):

    def __init__(self, cond_dim, num_features):
        super(Affine, self).__init__()
        self.fc_gamma = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self.fc_beta = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self._initialize()

    def _initialize(self):
        nn.init.zeros_(self.fc_gamma.linear2.weight.data)
        nn.init.ones_(self.fc_gamma.linear2.bias.data)
        nn.init.zeros_(self.fc_beta.linear2.weight.data)
        nn.init.zeros_(self.fc_beta.linear2.bias.data)

    def forward(self, x, y=None):
        weight = self.fc_gamma(y)
        bias = self.fc_beta(y)
        if weight.dim() == 1:
            weight = weight.unsqueeze(0)
        if bias.dim() == 1:
            bias = bias.unsqueeze(0)
        size = x.size()
        weight = weight.unsqueeze(-1).unsqueeze(-1).expand(size)
        bias = bias.unsqueeze(-1).unsqueeze(-1).expand(size)
        return weight * x + bias


INIT_ARGS = {
    'cond_dim': 5,
    'num_features': 4,
}

ARGS = {
    'x': torch.randn(2, 4, 3, 3),
    'y': torch.randn(2, 5),
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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, 3, 3) — shape (2, 4, 3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
CLIP_Adapter Ran
ran on a synthesized input
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("fa55421f21962920")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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
import torch.nn as nn
from collections import OrderedDict

# Create a minimal mock CLIP object with a visual attribute
class MockCLIPVisual(nn.Module):
    def __init__(self):
        super(MockCLIPVisual, self).__init__()
        # CLIP visual expects specific architecture
        self.conv1 = nn.Conv2d(3, 768, kernel_size=32, stride=32, padding=0)
        self.class_embedding = nn.Parameter(torch.randn(768))
        # For a 224x224 image with 32x32 patches, we get 7x7 = 49 patches
        # Plus 1 class token = 50 positions
        self.positional_embedding = nn.Parameter(torch.randn(50, 768))
        self.ln_pre = nn.LayerNorm(768)
        
        # Create transformer with 12 residual blocks
        self.transformer = nn.Module()
        self.transformer.resblocks = nn.ModuleList([
            nn.TransformerEncoderLayer(d_model=768, nhead=8, dim_feedforward=2048, batch_first=False)
            for _ in range(12)
        ])

class MockCLIP(nn.Module):
    def __init__(self):
        super(MockCLIP, self).__init__()
        self.visual = MockCLIPVisual()

# Create mock CLIP instance
mock_clip = MockCLIP()

INIT_ARGS = {
    'in_ch': 64,
    'mid_ch': 48,
    'out_ch': 32,
    'G_ch': 128,
    'CLIP_ch': 768,
    'cond_dim': 256,
    'k': 3,
    's': 1,
    'p': 1,
    'map_num': 2,
    'CLIP': mock_clip
}

# For forward pass:
# out: feature map with in_ch channels
# c: conditioning vector with cond_dim features
# The CLIP_ViT expects img to be processed through FBlocks and conv_fuse
# which outputs CLIP_ch channels, and the spatial size should match CLIP expectations
# CLIP processes 224x224 images into 7x7 grids (with 32x32 patches)
batch_size = 2
spatial_size = 7  # To match CLIP's expected grid size after patch embedding

ARGS = {
    'out': torch.randn(batch_size, 64, spatial_size, spatial_size),
    'c': torch.randn(batch_size, 256)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
DFBLK Ran
ran on a synthesized input
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("95f9b8a9bd0d3592")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, dtype family float, shape (2, 4, 3, 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
import torch.nn as nn
from collections import OrderedDict

class Affine(nn.Module):

    def __init__(self, cond_dim, num_features):
        super(Affine, self).__init__()
        self.fc_gamma = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self.fc_beta = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self._initialize()

    def _initialize(self):
        nn.init.zeros_(self.fc_gamma.linear2.weight.data)
        nn.init.ones_(self.fc_gamma.linear2.bias.data)
        nn.init.zeros_(self.fc_beta.linear2.weight.data)
        nn.init.zeros_(self.fc_beta.linear2.bias.data)

    def forward(self, x, y=None):
        weight = self.fc_gamma(y)
        bias = self.fc_beta(y)
        if weight.dim() == 1:
            weight = weight.unsqueeze(0)
        if bias.dim() == 1:
            bias = bias.unsqueeze(0)
        size = x.size()
        weight = weight.unsqueeze(-1).unsqueeze(-1).expand(size)
        bias = bias.unsqueeze(-1).unsqueeze(-1).expand(size)
        return weight * x + bias

class DFBLK(nn.Module):

    def __init__(self, cond_dim, in_ch):
        super(DFBLK, self).__init__()
        self.affine0 = Affine(cond_dim, in_ch)
        self.affine1 = Affine(cond_dim, in_ch)

    def forward(self, x, y=None):
        h = self.affine0(x, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        h = self.affine1(h, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        return h

INIT_ARGS = {
    'cond_dim': 5,
    'in_ch': 4,
}

ARGS = {
    'x': torch.randn(2, 4, 3, 3),
    'y': torch.randn(2, 5),
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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, 3, 3) — shape (2, 4, 3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
G_Block Ran
ran on a synthesized input
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("50f859c074a3bb1e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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
import torch.nn as nn
import torch.nn.functional as F
from collections import OrderedDict

class Affine(nn.Module):

    def __init__(self, cond_dim, num_features):
        super(Affine, self).__init__()
        self.fc_gamma = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self.fc_beta = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self._initialize()

    def _initialize(self):
        nn.init.zeros_(self.fc_gamma.linear2.weight.data)
        nn.init.ones_(self.fc_gamma.linear2.bias.data)
        nn.init.zeros_(self.fc_beta.linear2.weight.data)
        nn.init.zeros_(self.fc_beta.linear2.bias.data)

    def forward(self, x, y=None):
        weight = self.fc_gamma(y)
        bias = self.fc_beta(y)
        if weight.dim() == 1:
            weight = weight.unsqueeze(0)
        if bias.dim() == 1:
            bias = bias.unsqueeze(0)
        size = x.size()
        weight = weight.unsqueeze(-1).unsqueeze(-1).expand(size)
        bias = bias.unsqueeze(-1).unsqueeze(-1).expand(size)
        return weight * x + bias

class DFBLK(nn.Module):

    def __init__(self, cond_dim, in_ch):
        super(DFBLK, self).__init__()
        self.affine0 = Affine(cond_dim, in_ch)
        self.affine1 = Affine(cond_dim, in_ch)

    def forward(self, x, y=None):
        h = self.affine0(x, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        h = self.affine1(h, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        return h

class G_Block(nn.Module):

    def __init__(self, cond_dim, in_ch, out_ch, imsize):
        super(G_Block, self).__init__()
        self.imsize = imsize
        self.learnable_sc = in_ch != out_ch
        self.c1 = nn.Conv2d(in_ch, out_ch, 3, 1, 1)
        self.c2 = nn.Conv2d(out_ch, out_ch, 3, 1, 1)
        self.fuse1 = DFBLK(cond_dim, in_ch)
        self.fuse2 = DFBLK(cond_dim, out_ch)
        if self.learnable_sc:
            self.c_sc = nn.Conv2d(in_ch, out_ch, 1, stride=1, padding=0)

    def shortcut(self, x):
        if self.learnable_sc:
            x = self.c_sc(x)
        return x

    def residual(self, h, y):
        h = self.fuse1(h, y)
        h = self.c1(h)
        h = self.fuse2(h, y)
        h = self.c2(h)
        return h

    def forward(self, h, y):
        h = F.interpolate(h, size=(self.imsize, self.imsize))
        return self.shortcut(h) + self.residual(h, y)


INIT_ARGS = {
    'cond_dim': 5,
    'in_ch': 3,
    'out_ch': 4,
    'imsize': 8
}

ARGS = {
    'h': torch.randn(2, 3, 4, 4),
    'y': torch.randn(2, 5)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
G_Block Ran
ran on a synthesized input
tobran/DF-GAN/code/models/GAN.py
pointer only (licence: NOASSERTION) · get_code("5a3683f208ea7376")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, tobran/DF-GAN (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.

M_Block Ran
ran on a synthesized input
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("700aca9a6a4eac99")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, dtype family float, shape (2, 5, 8, 8) 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 as nn
from collections import OrderedDict

class Affine(nn.Module):

    def __init__(self, cond_dim, num_features):
        super(Affine, self).__init__()
        self.fc_gamma = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self.fc_beta = nn.Sequential(OrderedDict([('linear1', nn.Linear(cond_dim, num_features)), ('relu1', nn.ReLU(inplace=True)), ('linear2', nn.Linear(num_features, num_features))]))
        self._initialize()

    def _initialize(self):
        nn.init.zeros_(self.fc_gamma.linear2.weight.data)
        nn.init.ones_(self.fc_gamma.linear2.bias.data)
        nn.init.zeros_(self.fc_beta.linear2.weight.data)
        nn.init.zeros_(self.fc_beta.linear2.bias.data)

    def forward(self, x, y=None):
        weight = self.fc_gamma(y)
        bias = self.fc_beta(y)
        if weight.dim() == 1:
            weight = weight.unsqueeze(0)
        if bias.dim() == 1:
            bias = bias.unsqueeze(0)
        size = x.size()
        weight = weight.unsqueeze(-1).unsqueeze(-1).expand(size)
        bias = bias.unsqueeze(-1).unsqueeze(-1).expand(size)
        return weight * x + bias

class DFBLK(nn.Module):

    def __init__(self, cond_dim, in_ch):
        super(DFBLK, self).__init__()
        self.affine0 = Affine(cond_dim, in_ch)
        self.affine1 = Affine(cond_dim, in_ch)

    def forward(self, x, y=None):
        h = self.affine0(x, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        h = self.affine1(h, y)
        h = nn.LeakyReLU(0.2, inplace=True)(h)
        return h

class M_Block(nn.Module):

    def __init__(self, in_ch, mid_ch, out_ch, cond_dim, k, s, p):
        super(M_Block, self).__init__()
        self.conv1 = nn.Conv2d(in_ch, mid_ch, k, s, p)
        self.fuse1 = DFBLK(cond_dim, mid_ch)
        self.conv2 = nn.Conv2d(mid_ch, out_ch, k, s, p)
        self.fuse2 = DFBLK(cond_dim, out_ch)
        self.learnable_sc = in_ch != out_ch
        if self.learnable_sc:
            self.c_sc = nn.Conv2d(in_ch, out_ch, 1, stride=1, padding=0)

    def shortcut(self, x):
        if self.learnable_sc:
            x = self.c_sc(x)
        return x

    def residual(self, h, text):
        h = self.conv1(h)
        h = self.fuse1(h, text)
        h = self.conv2(h)
        h = self.fuse2(h, text)
        return h

    def forward(self, h, c):
        return self.shortcut(h) + self.residual(h, c)


INIT_ARGS = {
    'in_ch': 3,
    'mid_ch': 4,
    'out_ch': 5,
    'cond_dim': 6,
    'k': 3,
    's': 1,
    'p': 1
}

ARGS = {
    'h': torch.randn(2, 3, 8, 8),
    'c': torch.randn(2, 6)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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, 5, 8, 8) — shape (2, 5, 8, 8) 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
get_G_in_out_chs Ran
ran; our contract draft was wrong, not the code
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("7b1f9c201441fa1c")
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) and returned zip. 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 numpy as np

ARGS = {
    "nf": 64,
    "imsize": 256,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type zip — a zip output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a zip 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
CLIP_Mapper Not yet run
unverified (harvested, no recorded run)
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("4d38c586ad9ce3fc")
How it ran

no invocation recorded.

NetG Not yet run
unverified (harvested, no recorded run)
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("98f6bc75d1387ee2")
How it ran

no invocation recorded.

NetG Not yet run
unverified (harvested, no recorded run)
tobran/DF-GAN/code/models/GAN.py
pointer only (licence: NOASSERTION) · get_code("74fe6925fe919a9d")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, tobran/DF-GAN (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. No invocation recorded.

dummy_context_mgr Not yet run
unverified (harvested, no recorded run)
tobran/GALIP/code/models/GALIP.py
code served (permissive licence) · get_code("95d80ac15e73e11c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 9 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
dummy_context_mgr
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 2 only as an assembly brick
2107.10060 (2021-07)
Conditional GANs with Auxiliary Discriminative Classifier
2204.14249 (2022-04)
OSSGAN: Open-Set Semi-Supervised Image Generation
get_G_in_out_chs
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2111.13792 (2021-11)
LAFITE: Towards Language-Free Training for Text-to-Image Generation

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

Synthesizing high-fidelity complex images from text is challenging. Based on large pretraining, the autoregressive and diffusion models can synthesize photo-realistic images. Although these large models have shown notable progress, there remain three flaws. 1) These models require tremendous training data and parameters to achieve good performance. 2) The multi-step generation design slows the image synthesis process heavily. 3) The synthesized visual features are challenging to control and require delicately designed prompts. To enable high-quality, efficient, fast, and controllable text-to-image synthesis, we propose Generative Adversarial CLIPs, namely GALIP. GALIP leverages the powerful pretrained CLIP model both in the discriminator and generator. Specifically, we propose a CLIP-based discriminator. The complex scene understanding ability of CLIP enables the discriminator to accurately assess the image quality. Furthermore, we propose a CLIPempowered generator that induces the visual concepts from CLIP through bridge features and prompts. The CLIPintegrated generator and discriminator boost training efficiency, and as a result, our model only requires about 3% training data and 6% learnable parameters, achieving comparable results to large pretrained autoregressive and diffusion models. Moreover, our model achieves ∼120×faster synthesis speed and inherits the smooth latent space from GAN. The extensive experimental results demonstrate the excellent performance of our GALIP. Code is available at https://github.com/tobran/GALIP.

For agents

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

get_harvested_code_for_paper("2301.12959")
get_code_for_paper("2301.12959")
have("2301.12959")

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

curl https://syntology.ai/api/ran/2301.12959.json

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

[![Syntology run record](https://syntology.ai/api/ran/2301.12959.svg)](https://syntology.ai/paper/2301.12959)

Connect an agent — have() is free.