SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2109.06387 · EMNLP · 2021

Rationales for Sequential Predictions

Yuntian Deng, Alexander Rush, David Blei, Keyon Vafa

arXiv · PDF · Open in the Atlas

Code that ran

4 of 6 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 0 have no contract verdict; 4 failures 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/2109.06387.json · badge: /api/ran/2109.06387.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
keyonvafa/career-code community
Papers with Code archive
2 ran · 1 no recorded run
keyonvafa/sequential-rationales marked official in Papers with Code
Papers with Code archive
2 ran · 1 no recorded run
FunctionStatusWhere it lives
decode_sequence Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
keyonvafa/career-code/fairseq/rationalization.py
code served (permissive licence) · get_code("6e11afaf3878d5ad")
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 torch.nn as nn

torch.manual_seed(0)

class MockDictionary:
    def __init__(self, num_symbols=10):
        self.symbols = [f"word_{i}" for i in range(num_symbols)]
    
    def string(self, indices):
        return "".join([self.symbols[i] for i in indices])

class MockBPE:
    def decode(self, s):
        return s

class MockTask:
    def __init__(self):
        self.source_dictionary = MockDictionary(num_symbols=10)
        self.target_dictionary = MockDictionary(num_symbols=10)

class MockModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.task = MockTask()
        self.bpe = None

model = MockModel()
tokens = [4, 5, 2, 6]

ARGS = {
    "model": model,
    "tokens": tokens,
    "source": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
decode_sequence Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
keyonvafa/sequential-rationales/fairseq/rationalization.py
code served (permissive licence) · get_code("81307176d91be885")
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 torch.nn as nn

torch.manual_seed(0)

class StandInDictionary:
    def __init__(self, num_symbols=10):
        self.symbols = [f"word_{i}" for i in range(num_symbols)]
    
    def string(self, indices):
        return "".join([self.symbols[i] for i in indices])

class StandInBPE:
    def decode(self, string):
        return string

class StandInTask:
    def __init__(self):
        self.source_dictionary = StandInDictionary(num_symbols=10)
        self.target_dictionary = StandInDictionary(num_symbols=10)

class StandInModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.task = StandInTask()
        self.bpe = StandInBPE()

model = StandInModel()
tokens = [0, 1, 2, 3]

ARGS = {
    "model": model,
    "tokens": tokens,
    "source": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
decode_single_word Ran
ran; our contract draft was wrong, not the code
keyonvafa/career-code/fairseq/rationalization.py
code served (permissive licence) · get_code("67860fa68ffc1866")
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
import torch.nn as nn

torch.manual_seed(0)

class StandInDictionary:
    def __init__(self):
        self.symbols = ['<unk>', '<eos>', '<bos>', '<pad>', 'hello', 'world', 'test']
    
    def string(self, indices):
        return ''.join([self.symbols[i] for i in indices])

class StandInBPE:
    def decode(self, string):
        return string.replace('@@', '')

class StandInTask:
    def __init__(self):
        self.source_dictionary = StandInDictionary()
        self.target_dictionary = StandInDictionary()

class StandInModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.task = StandInTask()
        self.bpe = None

class StandInModelWithBPE(nn.Module):
    def __init__(self):
        super().__init__()
        self.task = StandInTask()
        self.bpe = StandInBPE()

model = StandInModel()
word_index = 4

ARGS = {
    'model': model,
    'word_index': word_index,
    'source': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
decode_single_word Ran
ran; our contract draft was wrong, not the code
keyonvafa/sequential-rationales/fairseq/rationalization.py
code served (permissive licence) · get_code("f4b35aa2c181e0e7")
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
import torch.nn as nn

# Create a minimal Dictionary class
class Dictionary:
    def __init__(self, symbols):
        self.symbols = symbols
    
    def string(self, indices):
        return "".join([self.symbols[i] for i in indices])

# Create a minimal BPE class
class BPE:
    def decode(self, text):
        return text.replace("@@", "")

# Create a minimal Task class
class Task:
    def __init__(self):
        self.source_dictionary = Dictionary(["<pad>", "<eos>", "hello", "world"])
        self.target_dictionary = Dictionary(["<pad>", "<eos>", "bonjour", "monde"])

# Create a minimal Model class
class MinimalModel(nn.Module):
    def __init__(self, use_bpe=False):
        super().__init__()
        self.task = Task()
        if use_bpe:
            self.bpe = BPE()
        else:
            self.bpe = None

torch.manual_seed(0)
model = MinimalModel(use_bpe=False)

ARGS = {
    "model": model,
    "word_index": 2,
    "source": True
}
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
rationalize_conditional_model Not yet run
unverified (harvested, no recorded run)
keyonvafa/sequential-rationales/fairseq/rationalization.py
code served (permissive licence) · get_code("f49763cc0d941343")
How it ran

no invocation recorded.

rationalize_occupation_model Not yet run
unverified (harvested, no recorded run)
keyonvafa/career-code/fairseq/rationalization.py
code served (permissive licence) · get_code("63536eed85fd01b3")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 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.

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

Sequence models are a critical component of modern NLP systems, but their predictions are difficult to explain. We consider model explanations though rationales, subsets of context that can explain individual model predictions. We find sequential rationales by solving a combinatorial optimization: the best rationale is the smallest subset of input tokens that would predict the same output as the full sequence. Enumerating all subsets is intractable, so we propose an efficient greedy algorithm to approximate this objective. The algorithm, which is called greedy rationalization, applies to any model. For this approach to be effective, the model should form compatible conditional distributions when making predictions on incomplete subsets of the context. This condition can be enforced with a short finetuning step. We study greedy rationalization on language modeling and machine translation. Compared to existing baselines, greedy rationalization is best at optimizing the sequential objective and provides the most faithful rationales. On a new dataset of annotated sequential rationales, greedy rationales are most similar to human rationales.

For agents

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

get_harvested_code_for_paper("2109.06387")
get_code_for_paper("2109.06387")
have("2109.06387")

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

curl https://syntology.ai/api/ran/2109.06387.json

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

[![Syntology run record](https://syntology.ai/api/ran/2109.06387.svg)](https://syntology.ai/paper/2109.06387)

Connect an agent — have() is free.