SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.05640 · EMNLP · 2023

FinGPT: Large Generative Models for a Small Language

Teven Le Scao, Aleksandra Piktus, Thomas Wolf, Thomas Wang, Niklas Muennighoff, Sampo Pyysalo, Jenna Kanerva, Veronika Laippala, Filip Ginter, Nouamane Tazi, Risto Luukkonen, Ville Komulainen, and 9 more

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 6 failures of our instrument; 1 no recorded run), measured 2026-11-22; 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/2311.05640.json · badge: /api/ran/2311.05640.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
TurkuNLP/FIN-bench canonical (classifier label)
classifier basis: census
6 ran · 1 no recorded run
FunctionStatusWhere it lives
deduplicate_analogies Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/benchmark_tasks/analogies/make_task.py
code served (permissive licence) · get_code("e00c9e9c6bfb22b7")
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 defaultdict. 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.

from collections import defaultdict

# Construct valid arguments for deduplicate_analogies
# The function expects:
# - analogies: a dict mapping category names to lists of quadruples (tuples of 4 items)

# Create sample analogies data
# Each quadruple is a 4-tuple representing an analogy (e.g., (a, b, c, d))
# where a:b :: c:d
analogies = {
    "capital": [
        ("Paris", "France", "Berlin", "Germany"),
        ("London", "England", "Madrid", "Spain"),
        ("Rome", "Italy", "Athens", "Greece"),
    ],
    "animal": [
        ("dog", "puppy", "cat", "kitten"),
        ("lion", "cub", "bear", "cub"),
        ("horse", "foal", "sheep", "lamb"),
    ],
}

ARGS = {
    "analogies": analogies,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
get_divisible_pair_generator Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/benchmark_tasks/arithmetic/generate_examples.py
code served (permissive licence) · get_code("dbcb85405bcad52b")
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 function. 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 random

# Set seed for reproducibility
random.seed(42)

# Define the arguments for get_divisible_pair_generator
ARGS = {
    "digits": 3,
    "num_examples": 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
get_random_pair_generator Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/benchmark_tasks/arithmetic/generate_examples.py
code served (permissive licence) · get_code("c7f8c1d96e05cacf")
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 function. 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 random

def generate_number(digits):
    """Generate an integer with the given number of digits."""
    low = 10 ** (digits - 1)
    high = 10 ** digits - 1
    return random.randint(low, high)

def get_random_pair_generator(digits, num_examples):
    """Generate random number pairs with the given number of digits.

    The generated sequence does not contain duplicates.
    """

    def pair_generator():
        history = []
        for _ in range(num_examples):
            while True:
                pair = (generate_number(digits), generate_number(digits))
                if pair not in history:
                    break
            history.append(pair)
            yield pair
    return pair_generator


# Set random seed for reproducibility
random.seed(0)

ARGS = {
    "digits": 3,
    "num_examples": 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
load_analogy_data Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/benchmark_tasks/analogies/make_task.py
code served (permissive licence) · get_code("65f8331340e9f1ea")
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 defaultdict. 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 tempfile
import os

# Create a temporary file with analogy data in the expected format
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write(": capital-common-countries,gram1: city-and-country\n")
temp_file.write("Athens Greece Baghdad Iraq\n")
temp_file.write("Bangkok Thailand Beijing China\n")
temp_file.write(": capital-world\n")
temp_file.write("Abuja Nigeria Algiers Algeria\n")
temp_file.write("Amman Jordan Ankara Turkey\n")
temp_file.close()

fn = temp_file.name

ARGS = {
    "fn": fn
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
load_name_counts Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/gender-occupation-bias/combine_names.py
code served (permissive licence) · get_code("af2e1cb4c59627c9")
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 defaultdict. 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 tempfile
import os
from collections import defaultdict

# Create a temporary file with valid Finnish name data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write('Etunimi\tLukumäärä\n')
temp_file.write('Matti\t1234\n')
temp_file.write('Maria\t5678\n')
temp_file.write('Pekka\t910\n')
temp_file.close()

ARGS = {
    'fn': temp_file.name
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
load_word_data Ran
ran; our contract draft was wrong, not the code
TurkuNLP/FIN-bench/benchmark_tasks/analogies/make_task.py
code served (permissive licence) · get_code("d0c9746eed432c8c")
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 json
import tempfile
import os

# Create a temporary file with JSON lines data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.jsonl')
temp_filename = temp_file.name

# Write sample JSON lines to the file
sample_data = [
    {'word': 'hello', 'value': 1},
    {'word': 'world', 'value': 2},
    {'word': 'test', 'value': 3},
    {'word': 'python', 'value': 4},
]

for item in sample_data:
    temp_file.write(json.dumps(item) + '\n')

temp_file.close()

# Define the words to filter by
words_set = {'hello', 'test'}

ARGS = {
    'fn': temp_filename,
    'words': words_set,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
generate_number Not yet run
unverified (harvested, no recorded run)
TurkuNLP/FIN-bench/benchmark_tasks/arithmetic/generate_examples.py
code served (permissive licence) · get_code("16fccce675cdf16d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 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 7 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

Large language models (LLMs) excel in many tasks in NLP and beyond, but most open models have very limited coverage of smaller languages and LLM work tends to focus on languages where nearly unlimited data is available for pretraining. In this work, we study the challenges of creating LLMs for Finnish, a language spoken by less than 0.1% of the world population. We compile an extensive dataset of Finnish combining web crawls, news, social media and eBooks. We pursue two approaches to pretrain models: 1) we train seven monolingual models from scratch (186M to 13B parameters) dubbed FinGPT, 2) we continue the pretraining of the multilingual BLOOM model on a mix of its original training data and Finnish, resulting in a 176 billion parameter model we call BLUUMI. For model evaluation, we introduce FIN-bench, a version of BIG-bench with Finnish tasks. We also assess other model qualities such as toxicity and bias. Our models and tools are openly available at https://turkunlp.org/gpt3-finnish.

For agents

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

get_harvested_code_for_paper("2311.05640")
get_code_for_paper("2311.05640")
have("2311.05640")

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

curl https://syntology.ai/api/ran/2311.05640.json

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

[![Syntology run record](https://syntology.ai/api/ran/2311.05640.svg)](https://syntology.ai/paper/2311.05640)

Connect an agent — have() is free.