SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.20009 · 2023

Protein Design with Guided Discrete Diffusion

arXiv · PDF · Open in the Atlas

Code that ran

14 of 16 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 12 have no contract verdict; 1 failure of our instrument; 2 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/2305.20009.json · badge: /api/ran/2305.20009.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
ngruver/nos canonical (classifier label)
classifier basis: census
14 ran · 2 no recorded run
FunctionStatusWhere it lives
betas_for_alpha_bar Ran
ran, honoured the contract we drafted
ngruver/nos/seq_models/schedule/noise_schedule.py
pointer only (licence: MIT); the stored copy get_code serves, from NVlabs/DiffPure, is pointer only (licence: NOASSERTION) · get_code("2ab2316ac6fdd869")
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 NVlabs/DiffPure, and GitHub could not classify the licence of that repository (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.

betas_for_alpha_bar2 Ran
ran on a synthesized input
ngruver/nos/seq_models/schedule/noise_schedule.py
code served (permissive licence) · get_code("fd26f763d09f5953")
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 1, dtype family float, shape (10,) 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

def alpha_bar_fn(t):
    """
    A simple alpha_bar function that decreases from 1 to near 0 as t goes from 0 to 1.
    This simulates a cumulative product schedule commonly used in diffusion models.
    """
    return np.cos((t + 0.008) / 1.008 * np.pi / 2) ** 2

ARGS = {
    "num_diffusion_timesteps": 10,
    "alpha_bar": alpha_bar_fn,
    "max_beta": 0.999
}
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 1, dtype family float, shape (10,) — shape (10,) 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
chains_to_formatted Ran
ran on a synthesized input
ngruver/nos/seq_models/sample.py
code served (permissive licence) · get_code("ba68ea0d339e00f9")
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 str. 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

def chains_to_formatted(vh, vl):
    return ' '.join(['[AgHC]'] + vh + ['[AgLC]'] + vl + ['[Ag]'])

vh = ['EVQL', 'GVQPD', 'RASM']
vl = ['DIQM', 'TPYT', 'SSVK', 'GAFT']

ARGS = {
    'vh': vh,
    'vl': vl
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
formatted_to_chains Ran
ran on a synthesized input
ngruver/nos/seq_models/sample.py
code served (permissive licence) · get_code("c7011cc258dff6b6")
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 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 numpy as np

torch.manual_seed(0)

# Analyze the function:
# - x: a string that gets split by spaces, then indices [1:150] and [151:-1] are joined
# - aligned_masks: a dict with keys 'vh' and 'vl', values are strings/iterables
# - The function processes chains and masks by removing '-' characters

# Create a valid x string:
# After split by ' ', we need at least 151 elements (for index 150)
# and elements beyond 151 (for index 151:-1)
# Create a string with space-separated elements
x_parts = ['start'] + ['aa'] * 150 + ['bb'] * 50  # enough elements
x = ' '.join(x_parts)

# Create aligned_masks dict with 'vh' and 'vl' keys
# Each mask should be a string of characters where some positions have '-'
# The mask length should match the chain length after joining x[1:150] and x[151:-1]
# x[1:150] gives us 149 'aa's joined = 149*2 = 298 characters
# x[151:-1] gives us some 'bb's joined

vh_chain_length = 149 * 2  # 298 chars from joining 'aa' * 149
vl_chain_length = 49 * 2   # 98 chars from joining 'bb' * 49

# Create masks with same length as chains, some with '-' some without
vh_mask = 'M' * (vh_chain_length // 2) + '-' * (vh_chain_length // 2)
vl_mask = 'M' * (vl_chain_length // 2) + '-' * (vl_chain_length // 2)

aligned_masks = {
    'vh': vh_mask,
    'vl': vl_mask
}

ARGS = {
    'x': x,
    'aligned_masks': aligned_masks
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, 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 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
get_named_beta_schedule Ran
ran on a synthesized input
ngruver/nos/seq_models/schedule/noise_schedule.py
code served (permissive licence) · get_code("14cec0043dedf345")
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 1, dtype family float, shape (10,) 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 math
import numpy as np

# Test setup for get_named_beta_schedule function
ARGS = {
    'schedule_name': 'linear',
    'num_diffusion_timesteps': 10
}
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 1, dtype family float, shape (10,) — shape (10,) 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_ranges Ran
ran on a synthesized input
ngruver/nos/seq_models/compare_samples.py
code served (permissive licence) · get_code("58f8c46fa4402e38")
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 list, ndim 2, 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 get_ranges takes a mask parameter which is used with:
# - mask.find('1', end) - string method find
# - mask.find('0', start) - string method find
# - len(mask) - length of string
# So mask should be a string

# Create a simple mask string with some ranges of '1's and '0's
mask = "11010110111"

ARGS = {
    "mask": mask
}
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 list, ndim 2, dtype family int — a list 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
iglm_perplexity Ran
ran on a synthesized input
ngruver/nos/seq_models/compare_samples.py
code served (permissive licence) · get_code("000d647ece3c1093")
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 as nn

torch.manual_seed(0)

class MockIGLMModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def log_likelihood(self, chain, chain_token, species_token, infill_range=None):
        # Return a scalar tensor representing log likelihood
        # Negative value so perplexity accumulates as positive
        return torch.tensor(-2.5)

# Create model instance
iglm_model = MockIGLMModel()

# chain: sequence data, can be a list or tensor
chain = torch.randn(2, 4)

# ranges: list of tuples or ranges specifying infill regions
ranges = [(0, 1), (1, 2)]

# chain_token: token identifier for the chain
chain_token = "chain_1"

# species_token: token identifier for the species
species_token = "species_1"

ARGS = {
    "iglm_model": iglm_model,
    "chain": chain,
    "ranges": ranges,
    "chain_token": chain_token,
    "species_token": species_token,
}
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
infill_metrics_sample Ran
ran on a synthesized input
ngruver/nos/seq_models/compare_samples.py
code served (permissive licence) · get_code("8ad4d05cd69ac220")
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 dict. 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

class MockIGLMModel:
    def log_likelihood(self, chain, chain_token, species_token, infill_range=None):
        # Return a mock log likelihood value
        return -5.0

# Create a mock model
iglm_model = MockIGLMModel()

# Create sample sequences and masks
vh_seed = 'ACDEFGHIKLMNPQRSTVWY'
vl_seed = 'ACDEFGH'
vh_sample = 'ACDEFGHIKLMNPQRSTVWY'
vl_sample = 'ACDEFGH'

# Masks: '0' for seed positions (not infilled), '1' for infilled positions
vh_mask = '00000000000000000000'  # All seed, no infill
vl_mask = '0000000'  # All seed, no infill

species_token = '[HUMAN]'

ARGS = {
    'iglm_model': iglm_model,
    'vh_seed': vh_seed,
    'vl_seed': vl_seed,
    'vh_sample': vh_sample,
    'vl_sample': vl_sample,
    'vh_mask': vh_mask,
    'vl_mask': vl_mask,
    'species_token': species_token,
    'sample_tag': 'test_sample',
    'model_tag': 'test_model'
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
make_tags Ran
ran on a synthesized input
ngruver/nos/seq_models/sample.py
code served (permissive licence) · get_code("a0447d09ab7442d2")
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 list, ndim 1, 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 numpy as np

def make_tags(numbering_schemes, cdr_combos):
    tags = []
    for num_scheme in numbering_schemes:
        tags += [num_scheme + ':' + '/'.join(cdr_ids) for cdr_ids in cdr_combos]
    return tags

ARGS = {
    'numbering_schemes': ['scheme_a', 'scheme_b', 'scheme_c'],
    'cdr_combos': [
        ['cdr1', 'cdr2'],
        ['cdr3', 'cdr4', 'cdr5'],
        ['cdr6']
    ]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
make_wandb_log Ran
ran on a synthesized input
ngruver/nos/seq_models/metrics.py
code served (permissive licence) · get_code("c6234f423e26b09b")
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 dict. 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 numpy as np
import wandb

torch.manual_seed(0)

# Construct the log dict following the structure expected by make_wandb_log
log = {
    'seq_samples': [
        'ACGTACGT',
        'TGCATGCA',
        'AAATTTGGG'
    ],
    'biopython': {
        'property_A': {
            'metric_1': np.array([1.0, 2.0, 3.0, 4.0]),
            'metric_2': np.array([0.5, 1.5, 2.5]),
        },
        'property_B': {
            'metric_3': np.array([10.0, 20.0, 30.0, 40.0, 50.0]),
        }
    },
    'reg_label': {
        'loss': np.array([0.1, 0.2, 0.15, 0.25]),
        'accuracy': np.array([0.9, 0.92, 0.88, 0.95, 0.93]),
    }
}

log_prefix = 'training/epoch_1'

ARGS = {
    'log': log,
    'log_prefix': log_prefix,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, wandb — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
pad_to_length Ran
ran on a synthesized input
ngruver/nos/seq_models/data.py
code served (permissive licence) · get_code("8c3f4d642b502689")
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 1, dtype family float, shape (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)

# Create a small 1D tensor to pad
arr = torch.tensor([1.0, 2.0, 3.0])

# Define max_length for padding
max_length = 5

ARGS = {
    "arr": arr,
    "max_length": max_length
}
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 1, dtype family float, shape (5,) — shape (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
random_sequences Ran
ran on a synthesized input
ngruver/nos/seq_models/data.py
code served (permissive licence) · get_code("a063d5b22365eced")
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 list, ndim 1, 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 numpy as np
import transformers
import tempfile
import os

# Create a minimal vocab file for BERT tokenizer
vocab_content = """[PAD]
[unused0]
[unused1]
A
C
D
E
F
G
H
I
K
L
M
N
P
Q
R
S
T
V
W
Y
[CLS]
[SEP]
[MASK]
a
c
d
e
f
g
h
i
k
l
m
n
p
q
r
s
t
v
w
y
"""

# Write vocab to a temporary file
with tempfile.NamedTemporaryFile(mode='w', suffix='.txt', delete=False) as f:
    f.write(vocab_content)
    vocab_file_path = f.name

try:
    # Verify the tokenizer can be created with this vocab
    test_tokenizer = transformers.BertTokenizerFast(vocab_file=vocab_file_path, do_lower_case=False)
    
    ARGS = {
        'num_samples': 3,
        'lengths': [5, 8, 10],
        'vocab_file': vocab_file_path
    }
finally:
    # Clean up temp file after verification
    if os.path.exists(vocab_file_path):
        os.unlink(vocab_file_path)
    
    # Recreate the file since we deleted it
    with tempfile.NamedTemporaryFile(mode='w', suffix='.txt', delete=False) as f:
        f.write(vocab_content)
        vocab_file_path = f.name
    
    ARGS = {
        'num_samples': 3,
        'lengths': [5, 8, 10],
        'vocab_file': vocab_file_path
    }
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
reformat Ran
ran on a synthesized input
ngruver/nos/seq_models/data.py
code served (permissive licence) · get_code("26fe4feb6d44885d")
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 str. 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

torch.manual_seed(0)

def reformat(seq):
    if len(seq.split(' ')) <= 1:
        h, l = seq.split('[AbLC]')
        h = h.replace('[AbHC]', '')
        l = l.replace('[Ag]', '')
        return '[AbHC] ' + ' '.join(h) + ' [AbLC] ' + ' '.join(l) + ' [Ag]'
    else:
        return seq

# Test case 1: seq with no spaces (single token case)
seq_1 = '[AbHC]hello[AbLC]world[Ag]'

# Test case 2: seq with spaces (multi-token case)
seq_2 = '[AbHC] hello world [AbLC] foo bar [Ag]'

ARGS = {
    'seq': seq_1
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
timestep_embedding Ran
ran; our contract draft was wrong, not the code
ngruver/nos/seq_models/nets/util.py
code served (permissive licence) · get_code("6c617b25817f8107")
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 (4, 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 math
import torch

torch.manual_seed(0)

# Create valid arguments for timestep_embedding function
timesteps = torch.tensor([0, 5, 10, 15], dtype=torch.long)
dim = 8
max_period = 10000

ARGS = {
    "timesteps": timesteps,
    "dim": dim,
    "max_period": max_period,
}
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 (4, 8) — shape (4, 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: none by the harness; any seed is in the setup block
pplm_step Not yet run
unverified (harvested, no recorded run)
ngruver/nos/seq_models/model/autoregressive.py
code served (permissive licence) · get_code("02575eab5609b8d7")
How it ran

no invocation recorded.

regression_labels Not yet run
unverified (harvested, no recorded run)
ngruver/nos/seq_models/metrics.py
code served (permissive licence) · get_code("1854c648c2dc84eb")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 1 the same month, 2 later) and 14 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
betas_for_alpha_bar 92 other papers
39 dated earlier, 1 the same month, 50 later, 2 same year, order not determinable · 3 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2102.09672 (2021-02)
Improved Denoising Diffusion Probabilistic Models
and 89 more not named here
timestep_embedding 29 other papers
9 dated earlier, 20 later · 9 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
2103.11161 (2021-03)
MonteFloor: Extending MCTS for Reconstructing Accurate Large-Scale Floor Plans
and 26 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

A popular approach to protein design is to combine a generative model with a discriminative model for conditional sampling. The generative model samples plausible sequences while the discriminative model guides a search for sequences with high fitness. Given its broad success in conditional sampling, classifier-guided diffusion modeling is a promising foundation for protein design, leading many to develop guided diffusion models for structure with inverse folding to recover sequences. In this work, we propose diffusioN Optimized Sampling (NOS), a guidance method for discrete diffusion models that follows gradients in the hidden states of the denoising network. NOS makes it possible to perform design directly in sequence space, circumventing significant limitations of structure-based methods, including scarce data and challenging inverse design. Moreover, we use NOS to generalize LaMBO, a Bayesian optimization procedure for sequence design that facilitates multiple objectives and edit-based constraints. The resulting method, LaMBO-2, enables discrete diffusions and stronger performance with limited edits through a novel application of saliency maps. We apply LaMBO-2 to a real-world protein design task, optimizing antibodies for higher expression yield and binding affinity to several therapeutic targets under locality and developability constraints, attaining a 99% expression rate and 40% binding rate in exploratory in vitro experiments.

For agents

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

get_harvested_code_for_paper("2305.20009")
get_code_for_paper("2305.20009")
have("2305.20009")

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

curl https://syntology.ai/api/ran/2305.20009.json

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

[![Syntology run record](https://syntology.ai/api/ran/2305.20009.svg)](https://syntology.ai/paper/2305.20009)

Connect an agent — have() is free.