SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2205.01068 · 2022

OPT: Open Pre-trained Transformer Language Models

Myle Ott, Naman Goyal, Luke Zettlemoyer, Tianlu Wang, Mikel Artetxe, Xian Li, Todor Mihaylov, Susan Zhang, Moya Chen, Kurt Shuster, Daniel Simig, Stephen Roller, and 7 more

arXiv · PDF · Open in the Atlas

Code that ran

14 of 24 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 13 have no contract verdict; 1 failure of our instrument; 10 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/2205.01068.json · badge: /api/ran/2205.01068.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
facebookresearch/metaseq canonical (classifier label)
classifier basis not recorded on the link
9 ran · 3 no recorded run
mthcom/hscore-dataset-pruning community (classifier label)
classifier basis not recorded on the link
2 ran · 6 no recorded run
ecolab-postech/owq community
Papers with Code archive
2 ran · 0 no recorded run
znhy1024/protoco community
Papers with Code archive
1 ran · 0 no recorded run
xvyaward/owq community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
LoRALinear Ran
ran on a synthesized input
znhy1024/protoco/models/lora.py
pointer only (licence: NONE) · get_code("e32e446737bb12b7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, znhy1024/protoco 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.

QuantLinear Ran
ran on a synthesized input · constructed an object; no result was computed
ecolab-postech/owq/owq/quant.py
pointer only (licence: NONE) · get_code("be9074477838726e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ecolab-postech/owq 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.

QuantMatMul Ran
ran on a synthesized input · constructed an object; no result was computed
ecolab-postech/owq/owq/quant.py
pointer only (licence: NONE) · get_code("f5d0291b559e1a6a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ecolab-postech/owq 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.

checkpoint_paths Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/checkpoint_utils.py
pointer only (licence: MIT) · get_code("c570c23ad25ca1ab")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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_offsets Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/file_chunker_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from leililab/enzygen, is pointer only (licence: NONE) · get_code("840d9070e4fd8ba6")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

flatten_outputs Ran
ran on a synthesized input
mthcom/hscore-dataset-pruning/create_scores.py
code served (permissive licence) · get_code("8ce19774aeb3fda0")
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, 2, 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

torch.manual_seed(0)

# Create a simple output object with logits attribute
class MockOutput:
    def __init__(self, logits):
        self.logits = logits

# Parameters:
# - outputs: list of objects with .logits attribute (or iterable of tensors)
# - nb_labels: number of labels (int)
# - nb_epochs: number of epochs (int)

nb_labels = 3
nb_epochs = 2

# Create mock outputs: 2 epochs, each with logits of shape (batch_size, nb_labels)
# Epoch 1: 2 samples
logits_epoch1 = torch.randn(2, nb_labels)
output1 = MockOutput(logits_epoch1)

# Epoch 2: 2 samples
logits_epoch2 = torch.randn(2, nb_labels)
output2 = MockOutput(logits_epoch2)

outputs = [output1, output2]

ARGS = {
    'outputs': outputs,
    'nb_labels': nb_labels,
    'nb_epochs': nb_epochs
}
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, 2, 3) — shape (2, 2, 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: none by the harness; any seed is in the setup block
flatten_outputs Ran
ran on a synthesized input
mthcom/hscore-dataset-pruning/create_scores_squad.py
code served (permissive licence) · get_code("be1104875f8a36ce")
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 (3, 2, 1, 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
import torch.nn as nn

torch.manual_seed(0)

# The function expects:
# - outputs: an iterable of tuples, where each tuple has two elements (o[0] and o[1])
#   that can be flattened and converted to lists
# - nb_epochs: an integer representing the number of epochs
# - nb_labels: an integer representing the number of labels (default 384)

# Looking at the function body:
# - outputs is iterated: for i, o in enumerate(outputs)
# - each o is indexed: o[0] and o[1]
# - each o[i] has .flatten() and .tolist() called on it
# - This means o[0] and o[1] should be tensors

# The reshape at the end is: view(2, nb_epochs, -1, nb_labels).permute(1, 0, 2, 3)
# This means the total flattened length must equal 2 * nb_epochs * something * nb_labels
# If we have nb_epochs output tuples, and each contributes tensors that flatten to nb_labels elements,
# then: total_elements = 2 * nb_epochs * nb_labels (where the middle dimension becomes 1)

# Let's create outputs: a list of tuples, one per epoch
# Each tuple contains two tensors that flatten to nb_labels elements

nb_epochs = 3
nb_labels = 4

# Create outputs as a list of tuples
# Each tuple has (tensor1, tensor2) where each tensor flattens to nb_labels
outputs = []
for epoch in range(nb_epochs):
    # Create two tensors that flatten to nb_labels elements each
    tensor1 = torch.randn(1, nb_labels)  # shape (1, 4) -> flattens to 4 elements
    tensor2 = torch.randn(1, nb_labels)  # shape (1, 4) -> flattens to 4 elements
    outputs.append((tensor1, tensor2))

ARGS = {
    "outputs": outputs,
    "nb_epochs": nb_epochs,
    "nb_labels": nb_labels,
}
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 (3, 2, 1, 4) — shape (3, 2, 1, 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
gelu Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/modules/activation_functions.py
pointer only (licence: MIT) · get_code("04b65716e8591ed3")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

gelu_back Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/modules/activation_functions.py
pointer only (licence: MIT) · get_code("81272627e6280106")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

get_next_token Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/hub_utils.py
pointer only (licence: MIT) · get_code("70652fe4ded4564b")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

relu_squared Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/modules/activation_functions.py
pointer only (licence: MIT) · get_code("684b43f1cd6f7e56")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

tensorize_input Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/hub_utils.py
pointer only (licence: MIT) · get_code("e7743da916a7857a")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

url_to_filename Ran
ran; our contract draft was wrong, not the code
facebookresearch/metaseq/metaseq/file_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from agi-content-generation/EKD, is pointer only (licence: NOASSERTION) · get_code("9418b43048e4e4ec")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

with_incremental_state Ran
ran on a synthesized input
facebookresearch/metaseq/metaseq/incremental_decoding_utils.py
pointer only (licence: MIT) · get_code("1dcaa0202f989ada")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

Quantizer Not yet run
unverified (harvested, no recorded run)
xvyaward/owq/owq/quant.py
pointer only (licence: NONE) · get_code("352d10b844458470")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, xvyaward/owq 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. No invocation recorded.

add_to_rewards Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/create_scores.py
code served (permissive licence) · get_code("00c32fbe5c2af279")
How it ran

no invocation recorded.

cached_path Not yet run
unverified (harvested, no recorded run)
facebookresearch/metaseq/metaseq/file_utils.py
pointer only (licence: MIT) · get_code("fb80a6deb3bee2c5")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

flatten_outputs Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/create_scores_race.py
code served (permissive licence) · get_code("84c5f4182ed8c57d")
How it ran

no invocation recorded.

load_archive_file Not yet run
unverified (harvested, no recorded run)
facebookresearch/metaseq/metaseq/file_utils.py
pointer only (licence: MIT) · get_code("0d372aa1e290a613")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

load_labels Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/create_scores.py
code served (permissive licence) · get_code("559cff9b7463d528")
How it ran

no invocation recorded.

load_labels Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/create_scores_race.py
code served (permissive licence) · get_code("9387fa891879e614")
How it ran

no invocation recorded.

postprocess_qa_predictions Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/utils_qa.py
code served (permissive licence) · get_code("6eff2c131107723d")
How it ran

no invocation recorded.

postprocess_qa_predictions_with_beam_search Not yet run
unverified (harvested, no recorded run)
mthcom/hscore-dataset-pruning/utils_qa.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from misonsky/chunk, is pointer only (licence: NONE) · get_code("8c51f5aaf2f659e2")
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 misonsky/chunk, 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. No invocation recorded.

setup_vocab_and_merges Not yet run
unverified (harvested, no recorded run)
facebookresearch/metaseq/metaseq/hub_utils.py
pointer only (licence: MIT) · get_code("54309a919eca36b7")
How it ran

The body of this sample is withheld: MIT, as GitHub detects it on this paper's repository, facebookresearch/metaseq, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with it. 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.

Also present in other papers' code

Of 24 samples lifted from this paper's linked repositories, 24 carry a function identity and 0 do not. Of the 24 distinct functions with an identity, 5 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 1 the same month, 5 later) and 19 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
url_to_filename 48 other papers
33 dated earlier, 1 the same month, 13 later, 1 same year, order not determinable · 8 only as an assembly brick
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
1809.11096 (2018-09)
Large Scale GAN Training for High Fidelity Natural Image Synthesis
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
and 45 more not named here
load_archive_file 12 other papers
6 dated earlier, 5 later, 1 same year, order not determinable
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
2021.findings-emnlp.366 (2021 (year only))
Rethinking Zero-shot Neural Machine Translation: From a Perspective of Latent Variables
2104.08771 (2021-04)
Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translation
and 9 more not named here
find_offsets 5 other papers
2 dated earlier, 3 later
1610.02424 (2016-10)
Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models
2203.10426 (2022-03)
STEMM: Self-learning with Speech-text Manifold Mixup for Speech Translation
2210.15379 (2022-10)
MorphTE: Injecting Morphology in Tensorized Embeddings
and 2 more not named here
postprocess_qa_predictions_with_beam_search 5 other papers
1 dated earlier, 4 later
2106.09685 (2021-06)
LoRA: Low-Rank Adaptation of Large Language Models
2023.acl-short.31 (2023 (year only))
PLUE: Language Understanding Evaluation Benchmark for Privacy Policies in English
2310.14709 (2023-10)
Once Upon a $\textit{Time}$ in $\textit{Graph}$: Relative-Time Pretraining for Complex Temporal Reasoning
and 2 more not named here
with_incremental_state 1 other paper
1 later
2401.10440 (2024-01)
Breaking the Curse of Multilinguality with Cross-lingual Expert Language Models

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

Large language models, which are often trained for hundreds of thousands of compute days, have shown remarkable capabilities for zero-and few-shot learning. Given their computational cost, these models are difficult to replicate without significant capital. For the few that are available through APIs, no access is granted to the full model weights, making them difficult to study. We present Open Pre-trained Transformers (OPT), a suite of decoder-only pre-trained transformers ranging from 125M to 175B parameters, which we aim to fully and responsibly share with interested researchers. We show that OPT-175B is comparable to GPT-3, 1 while requiring only 1/7th the carbon footprint to develop. We are also releasing our logbook detailing the infrastructure challenges we faced, along with code for experimenting with all of the released models. * Equal contribution. † Work done while at Meta AI. 1 Following Brown et al. (2020), we use GPT-3 to refer to both the 175B model and the smaller scale models as well. 2 Exceptions include work by EleutherAI, who released dense models up to 20B in size (Black et al., 2022), Salesforce (Nijkamp et al., 2022), and Meta AI, who released dense models up to 13B and sparse models up to 1. 1T (Artetxe et al., 2021). There is also ongoing work from the BigScience workshop (https://bigscience. huggingface.co/), which aims to open source very large multilingual language models and datasets.

For agents

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

get_harvested_code_for_paper("2205.01068")
get_code_for_paper("2205.01068")
have("2205.01068")

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

curl https://syntology.ai/api/ran/2205.01068.json

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

[![Syntology run record](https://syntology.ai/api/ran/2205.01068.svg)](https://syntology.ai/paper/2205.01068)

Connect an agent — have() is free.