Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
Job has been terminated due to a temporary spike in resource usage and may be restarted later.
Error code:   JobManagerCrashedError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

text
string
category
string
source
string
a combination of one or more elementary reaction steps which start with the appropriate reactants and end with the appropriate product(s) a description of the path, or sequence of steps, by which a reaction occurs a description of the path that a reaction takes a detailed description of how a chemical reaction occurs a...
educational
FineWeb-Edu sample-10BT
Recursion with the Y Combinator In this article, we’ll introduce a higher-order function called the Y combinator. It’s immediately recognizable thanks to the famous startup incubator of the same name, but what is this strange sounding term all about? In most languages, recursion is supported directly for named function...
educational
FineWeb-Edu sample-10BT
About Noise Simulation Laboratory Realistic Audiovisual Recording, Simulation, and Reproduction of Traffic Noise and Noise Abatement Measures When it comes to planning transport infrastructure and noise abatement measures different complex physical phenomena are an issue. Yet only a relatively small group of experts fu...
educational
FineWeb-Edu sample-10BT
The Badrouzi Festival or “Kajin” was held in ancient Iran in honor of the Avestan “Akhshij” Bad (wind) or “Vata” and its guardian Izad (god), who was considered one of the greatest Iranian deities in the “Zarwani” beliefs, annually on Bahman 22 (February 11). Other names for this feast are “Badra” or “Bazoreh. Backgrou...
educational
FineWeb-Edu sample-10BT
Remember that Wikipedia wormhole you fell down the other night? Maybe it started with you looking at a list of every single episode of Glee and somehow ended with reading the biographical entry on John Wayne Gacy. You might not remember, and you might not want anyone else to know about it. But your history might still ...
educational
FineWeb-Edu sample-10BT
Business Taxation is a broad time period that refers back to the earnings tax. The Income Tax could be divided into two elements: Company Revenue Tax and private earnings tax. Corporate Income Tax is charged to the company entities and personal Earnings Tax is levied on people. Recent legislation passed by the US gover...
educational
FineWeb-Edu sample-10BT
Send to a friend Half the world will be online by the end of 2018, according to our estimates at the Web Foundation. This is an incredible milestone — but it also means that nearly four billion people across the globe are still offline, shut out from the digital revolution and the opportunities that many take for grant...
educational
FineWeb-Edu sample-10BT
Ongonig Projects - towards large-scale cellular agriculture Supporting one person’s demand for beef is known to be 40 times more resource intensive than producing the equivalent amount of crops required. Even though meat production already utilizes 70% of world’s agricultural land, the demand for meat is rapidly increa...
educational
FineWeb-Edu sample-10BT
Automation is something businesses in almost every sector are familiar with, as part of their efforts to make systems more efficient. It’s something that the cybersecurity industry is increasingly using, with automated data collection and processing playing an ever-growing role in protecting against data breaches and c...
educational
FineWeb-Edu sample-10BT
Workplace sexual harassment is a very big deal. If ignored or addressed improperly, it can impact everything from worker productivity and a company’s bottom line to individual employees’ job satisfaction and self esteem. Its effects cascade down through entire teams, and those around them. Ultimately, it is an illegal ...
educational
FineWeb-Edu sample-10BT
8 edition of The Jungle Book found in the catalog. The Jungle Book March 31, 2006 by Townsend Pr Written in English |The Physical Object| Apr 30, · The Jungle Book Kids Animation Story | Fairy Tales & Bedtime Story For Kids - Duration: T-Series Kids Hut Recommended for you. The Jungle Book, collection of stories by Rud...
educational
FineWeb-Edu sample-10BT
Discourse Analysis, Safety Alerts and Safety Boards There’s nothing quite like the tokenistic display of a safety alert. Go into any organization and see what’s on the safety notice board, observe the wall paper and then contemplate why people don’t take safety seriously. It’s like the safety industry has taken in all ...
educational
FineWeb-Edu sample-10BT
Actions and Detail Panel Let's Talk Science Saturdays: Gr. 1 (Structures & Materials) Sat, 8 April 2017, 2:30 PM – 3:30 PM EDT Do you wish you had more opportunities to learn with your child, or do you love finding unique and fun learning opportunities for them? Would you like to get your children more interested in sc...
educational
FineWeb-Edu sample-10BT
Watch: 1929 Stock Market Crash and the Great Depression – Documentary [58:35 minutes] After watching the above video and the relevant discussion in Special Topic 6 in the textbook, discuss why might business cycle occur? Summarize the causes and consequences of the economic downturn. What lessons should the government ...
educational
FineWeb-Edu sample-10BT
Early Christian tradition identifies John Mark as the author of the Gospel that bears his name. Medieval literature named him Mark the Evangelist. The name Mark is from the Latin Marcus, and is the surname of the writer (Acts 12:12, 25). The book of Mark probably dates from AD 66–70. It is one of the four canonical gos...
educational
FineWeb-Edu sample-10BT
Every season bring its own joy and pleasure. Being blessed with four seasons and the changes we experience are impossible to explain in words. Spring is full of colors, flowers, fruits, birds and festivals. Summer is the warmest of all the seasons, yet it has a lot of attractions with long days and short nights. Winter...
educational
FineWeb-Edu sample-10BT
There was no question that the monarch was in charge. Elizabeth I (1558-1603) and James I (1603-25) both made it very clear that they ruled the country. They made the laws, they fought the wars, they appointed the top ministers and so on. However, the monarchy worked on the basis of cooperation between the monarch and ...
educational
FineWeb-Edu sample-10BT
Humans may soon have to look to their laurels as the planet’s dominant species. Turkeys, heretofore harmless, have been exploding in size, swelling from an average 13.2lb (6kg) in 1929 to over 30lb today. On the fairly scientific assumption that present trends will persist, The Economist estimates that turkeys will be ...
educational
FineWeb-Edu sample-10BT
The Suez Canal The Suez Canal connects the Mediterranean Sea (foreground left) with the Gulf of Suez and thereby the Red Sea and the Indian Ocean and this way shortens the sea passage from Europe to the Middle and Far East. The northern end of the canal is in the green Nile Delta but it then crosses directly through th...
educational
FineWeb-Edu sample-10BT
some of these are INSANE… Here is a fun look at the largest black holes in the universe, as well as a perspective in size to our own galaxy and solar system: Our Milky Way may harbor millions of black holes… the ultra dense remnants of dead stars. But now, in the universe far beyond our galaxy, there’s evidence of some...
educational
FineWeb-Edu sample-10BT
Just what is Personalized Learning? Personalized learning is a partnership between students and teachers in the design of learning that emerges from students’ interests, questions, needs, and preferences, towards an aim of self-directed learning (Bray & McClaskey, 2014). The best personalization is both personal and so...
educational
FineWeb-Edu sample-10BT
The thirteenth section, consists of one hundred pages and concentrates on pain. This chapter, 'Descriptions of Pain,' is the major portion of the book, and is of great value. Pain, admittedly one of the most frequent complaints homeopaths hear about, is examined in all the repertory sections. This should certainly help...
educational
FineWeb-Edu sample-10BT
New research in the journal Animal Behaviour confirms what dog people have always known – dogs and people have a lot in common. Researchers at the University of Arizona looked at how 2 year olds, dogs and chimpanzees performed in tests designed to measure how subjects acquire knowledge and understanding, and found that...
educational
FineWeb-Edu sample-10BT
by Diana Yates Weed scientists in at least two Midwestern states have been reporting for years that a conservation program meant to provide habitat for pollinating insects is sowing bad seeds – including seeds of the potentially devastating agricultural weed Palmer amaranth – along with the good. Now, researchers at th...
educational
FineWeb-Edu sample-10BT
Living in America means that you have infinite opportunity to be whatever you want. All you have to do is follow the law in America you are allowed the resources you need to reach your goals in life. That is why many people come immigrate to the U.S and refer to it as a land of opportunity. One reason there is so much ...
educational
FineWeb-Edu sample-10BT
Raw coffee beans are actually green seeds that contain carbohydrates, water, proteins, lipids, acids, and alkaloids. Green coffee tastes and smells like peas and features a hard covering. Green beans should be roasted. The high-temperature processing allows us to alter the bean structure and its biochemistry and makes ...
educational
FineWeb-Edu sample-10BT
By Leila Ugincius University Public Affairs What do you do if you love engineering but you’re too young for a program that lets high school students research engineering at a college level? If you’re Alexandra Wright, you go straight to the top. Wright, a 14-year-old high school sophomore, wanted to further her studies...
educational
FineWeb-Edu sample-10BT
What is nutrition? One of the most important aspects of our health and development is whether we are overweight, nutrition. This is the process where we consume and utilize healthy food and nutrition and nutrition. Healthy nutrition that is part of nutrition is vital to ensuring a productive and enjoyable life without ...
educational
FineWeb-Edu sample-10BT
Drawing Using A Grid Open whiteboard/fullscreen version Sorry your device is unable to run Adobe Flash Activities You can play any activities that do not have the Flash Symbol while on this device. help on making these flash activites work try our help page. Please Note: This activity needs Adobe Flash Player to run. S...
educational
FineWeb-Edu sample-10BT
The hectic pace of daily life and the stresses that accompany it may make you want to tune out. A healthier approach may be to tune in. I know that sounds counterintuitive. But paying more attention to what is going on around you, not less, is the first step toward cultivating mindfulness, an excellent technique to hel...
educational
FineWeb-Edu sample-10BT
Sign up for our newsletter Gardening in the Rainy Zone. Pronounced: AL-lee-um ka-ra-tah-vee-EN-see Sunset zones 1-24. USDA zones: 5-9. Heat zones: 9-5. Height: 4-10 inches (10-25 cm). Two to three-inch diameter spherical umbel of 50 or more, flowers on a 6-inch stem. Just above ground level the thick, leathery, pleated...
educational
FineWeb-Edu sample-10BT
ID || first name || last name | 123 || Peter || Miller | 345 || Mary || Smith | 456 || John || Brown section nr || school || course nr | 45 || informatics || Q200 | 73 || comp. sci || C360 | 84 || informatics || Q205 ID || section nr | 123 || 73 | 123 || 84 | 345 || 73 | 456 || 45 entity: a table. (called class in obje...
educational
FineWeb-Edu sample-10BT
Nicolaus Copernicus born Mikolaj Kopernik was a Polish astronomer and mathematician who was a proponent of the view of an Earth in daily motion about its axis and in yearly motion around a stationary sun. This theory profoundly altered later workers’ view of the universe, but was rejected by the Catholic church. - Age:...
educational
FineWeb-Edu sample-10BT
“In war, there are no unwounded soldiers” is a quote often attributed to Argentinian writer and aphorist José Narosky. It rings true; veterans put their lives on the line to protect others, often at great personal sacrifice. Mental health for veterans is a fundamental piece of overall veteran health care that must be u...
educational
FineWeb-Edu sample-10BT
Bat diversity and abundance are highest in old deciduous forest stands on the river banks in Eastern Ukraine European forest-dwelling bats require complex woodland structures at both the micro-habitat and the landscape level for successful breeding in summer. Particularly, the results from Kharkiv region (Eastern Ukrai...
educational
FineWeb-Edu sample-10BT
Deep in a desiccated, Utah desert, surrounded by mountains and fringed with scorched sage and saltbush, stand the surreal remains of German Village. Out of bounds, out of place, out of time and 90 miles from Salt Lake City, it is surely the most bizarre feature of Dugway Proving Ground, a test site created by the Allie...
educational
FineWeb-Edu sample-10BT
Diploid cells (WI-38, MRC-5) vaccines have their origin in induced abortions. Among these vaccines we find the following: rubella, measles, mumps, rabies, polio, smallpox, hepatitis A, chickenpox, and herpes zoster. Nowadays, other abortion tainted vaccines cultivated on transformed cells (293, PER.C6) are in the pipel...
educational
FineWeb-Edu sample-10BT
He turned and made his way through the doors he suddenly remembered traversing many times the past few days. He was still above the timberline, devoid of any trees that would impair visibility so it was clear enough to follow the road with its many switchbacks and curves traversing the mountain below him, a black line ...
educational
FineWeb-Edu sample-10BT
What Do My Test Results Mean? Having a hard time understanding the lab tests ordered by your health care practitioner? It's easy to get confused by all the terminology you see when looking at your test results, so let us help you better understand the meaning of some of the most common tests ordered by your functional ...
educational
FineWeb-Edu sample-10BT
What fun it is to build a snowman each year! This Shared Reading book is similar to the familar "Buckle My Shoe" but with original text. It shows each step necessary to make a snowman. It is a perfect way to develop the concept of sequencing with your students. ?This full colored printable comes with matching pictures ...
educational
FineWeb-Edu sample-10BT
dians. He reported that the settlements in Tamaulipas on both sides of the Río Grande, known as the "villas del Norte," had generously offered to provide seven hundred mounted men, armed and provisioned for one month in the field. Three weeks later he again referred to the opportunity afforded by peace at home to carry...
educational
FineWeb-Edu sample-10BT
Cordelia is the youngest of Lear’s three daughters in Shakespeare’s play, King Lear. The play opens with Lear, the king of Britain, an old man now, gathering his family and his court around him to make a momentous announcement. He has decided to retire. He is going to divide his kingdom into three parts and award the s...
educational
FineWeb-Edu sample-10BT
The fight against the scab on the apple tree Let’s find out what is the disease scab on the apple (see photo) and where is it come from? The scab appears on the apples after a bad cleaning of leaves from the last year. On these last year’s leaves the scab spends the winter, and when the tree starts to dissolve the kidn...
educational
FineWeb-Edu sample-10BT
The US Navy Department intends to subsidize the development of an advanced Russian aircraft, which will eventually be extinguishing forest fires on US territory. This UFO-shaped flying machine is currently being designed by the Saratov aviation concern. The Saratov aircraft enterprise, which is located in Russia's Volg...
educational
FineWeb-Edu sample-10BT
Experienced project managers and developers understand the value of translating software requirements into robust designs and rational project plans. These steps are necessary whether the next release represents one percent or one hundred percent of the final product. This article explores some approaches for bridging ...
educational
FineWeb-Edu sample-10BT
What is Cap and Trade? In case you haven’t been following it, “Cap and Trade” is a generic shorthand term describing a proposed system for reducing pollution, or at least keeping a lid on pollution. WiseGeek has one of the clearest definitions of Cap and Trade: A cap and trade system is a method for managing pollution,...
educational
FineWeb-Edu sample-10BT
Most computer users are familiar with the concept of spyware — at least, they understand that it is something that can end up on their computers and cause some sort of harm, similar to viruses. The average computer user arguably has no need to understand the intricacies of spyware, or any other form of malware. But at ...
educational
FineWeb-Edu sample-10BT
Let’s Talk Fatty Liver Disease When we think about essential bodily organs, there are two primary candidates: our brain and our heart. However, every single organ in the body has a role in something, which makes every organ essential. Logically, we know this. It’s just that, if we were to develop a condition such as fa...
educational
FineWeb-Edu sample-10BT
TED ED: The Science of Baking a Chocolate Chip Cookie Ever baked chocolate chip cookies? Ever had your child help? Then you’ll both want to watch this adorable, informative TED ED animated video about what’s happening inside a cookie-in-the-making. (Ages eight and up.) It’s quite the drama of chemical reactions. After ...
educational
FineWeb-Edu sample-10BT
Since the 10 central European countries joined the EU on May 1st 2004, their average income converged towards the EU average. Several aspects of EU membership could have contributed to this process, such as improved market mechanisms, institutions and business environment, access to the single market, the involvement o...
educational
FineWeb-Edu sample-10BT
End of preview.

TinyBrain Pretrain Corpus 2B — Pretraining data for small language models

TinyBrain Pretrain Corpus 2B

A mixed-source English pretraining corpus for training small language models.

TinyBrain Pretrain Corpus 2B is a mixed-source dataset built for pretraining small causal language models, especially the TinyBrain-100M Base model.

The dataset combines educational text, factual/wiki-style text, math reasoning data, Python code-summary data, clean web text, and conversation-style data. It is designed to give small models a useful general foundation before supervised fine-tuning.

The uploaded train split contains 3,013,308 rows. A full scan estimates about 7.77B characters, 1.25B words, and roughly 1.81B tokens using a tokenizer-independent estimate. The exact token count may differ depending on the tokenizer used during training.

Most large pretraining corpora are built for much bigger models. TinyBrain Pretrain Corpus 2B is intentionally compact and focused on data that is useful for small models around 100M parameters.

Quick Start

Load the dataset with Hugging Face Datasets:

from datasets import load_dataset

ds = load_dataset("exnivo/tinybrain-pretrain-corpus-2b", split="train")

print(ds)
print(ds[0])

At a Glance

Item Details
Dataset type Base language model pretraining corpus
Rows 3,013,308
Approx. characters 7,767,447,861
Approx. words 1,249,832,587
Approx. tokens ~1.81B estimated tokens
Language Mostly English
Format Text rows with metadata
Main fields text, category, source
Best for Small causal language models around 100M parameters
Main use Training a base model before instruction tuning
Related base model exnivo/tinybrain-100m-base
Related SFT dataset exnivo/tinybrain-instruct-sft-200k
Related instruct model exnivo/tinybrain-100m-instruct

Why Use This Dataset?

TinyBrain Pretrain Corpus 2B is made for people training small language models from scratch.

Use it if you want to:

  • train a small causal language model
  • pretrain a model around 100M parameters
  • experiment with compact pretraining data
  • train a base model before SFT/instruction tuning
  • study how small models learn from educational, factual, math, code, web, and conversation data
  • reproduce or extend the TinyBrain-100M training pipeline
  • build a lightweight local base model

This dataset was used as the base pretraining corpus for exnivo/tinybrain-100m-base.

Dataset Summary

TinyBrain Pretrain Corpus 2B is a mixed-source English pretraining dataset.

The dataset was built to balance useful general knowledge with small-model-friendly sources. It includes educational web text, factual reference text, math reasoning data, code-related data, clean web text, and conversation-style examples.

The goal is not to create the largest possible web scrape. The goal is to create a compact, useful, and varied pretraining corpus that gives a small model enough language, factual, math, coding, and dialogue exposure before instruction tuning.

Each row contains text plus basic metadata:

{
  "text": "...",
  "category": "...",
  "source": "..."
}

Real Dataset Stats

Source Breakdown

Source Rows Percent
FineWeb-Edu sample-10BT 473,127 15.70%
CodeSearchNet / Python code-summary data 326,019 10.82%
TinyFacts generated QA from Wikipedia intros + seed facts 321,568 10.67%
Wikipedia English 290,353 9.64%
SmolLM-Corpus / Cosmopedia v2 279,498 9.28%
FineMath-4+ 261,414 8.68%
FineWeb sample-10BT 231,103 7.67%
Simple Wikipedia 161,571 5.36%
OpenMathInstruct-2 147,786 4.90%
smol-smoltalk / filtered SmolTalk 129,503 4.30%
UltraChat 200k 114,614 3.80%
MathInstruct 112,629 3.74%
OpenWebMath 111,512 3.70%
OpenAssistant OASST1 52,611 1.75%

Category Breakdown

Category Rows Percent
factual 773,492 25.67%
educational 752,625 24.98%
math_reasoning 633,341 21.02%
code 326,019 10.82%
conversation 296,728 9.85%
clean_web 231,103 7.67%

Source × Category Breakdown

Source / Category Rows Percent
FineWeb-Edu sample-10BT / educational 473,127 15.70%
CodeSearchNet / Python code-summary data / code 326,019 10.82%
TinyFacts generated QA from Wikipedia intros + seed facts / factual 321,568 10.67%
Wikipedia English / factual 290,353 9.64%
SmolLM-Corpus / Cosmopedia v2 / educational 279,498 9.28%
FineMath-4+ / math_reasoning 261,414 8.68%
FineWeb sample-10BT / clean_web 231,103 7.67%
Simple Wikipedia / factual 161,571 5.36%
OpenMathInstruct-2 / math_reasoning 147,786 4.90%
smol-smoltalk / filtered SmolTalk / conversation 129,503 4.30%
UltraChat 200k / conversation 114,614 3.80%
MathInstruct / math_reasoning 112,629 3.74%
OpenWebMath / math_reasoning 111,512 3.70%
OpenAssistant OASST1 / conversation 52,611 1.75%

Text Length and Token Estimate

Metric Value
Total rows scanned 3,013,308
Total characters 7,767,447,861
Total words 1,249,832,587
Approx. tokens from sample average 1,813,283,114
Average characters per row 2,577.7
Average words per row 414.8
Min characters per row 120
Max characters per row 24,000
Min words per row 2
Max words per row 7,605
Sample chars median 1,537
Sample chars p95 8,190
Sample approx tokens median 357
Sample approx tokens p95 1,896

Token counts are approximate and estimated without a tokenizer. For exact token counts, tokenize the dataset with the same tokenizer used during training.

Dataset Structure

Each row contains one text document or text chunk.

Main fields:

Field Description
text The training text
category Broad data category
source Source dataset/name used for the row

Example row:

{
  "text": "Photosynthesis is the process by which plants use sunlight, water, and carbon dioxide to make food...",
  "category": "educational",
  "source": "FineWeb-Edu sample-10BT"
}

Data Composition

TinyBrain Pretrain Corpus 2B is built from several broad data types.

Area Purpose
Factual text Helps with entities, places, history, reference-style knowledge, and factual completion
Educational text Helps the model learn explanations, school-style text, and general knowledge
Math reasoning Gives exposure to arithmetic, formulas, worked solutions, and mathematical language
Code Adds Python/code-summary exposure
Conversation Gives the base model early exposure to dialogue-like text
Clean web Adds general language variety

Format

The dataset is stored as text data with metadata.

A typical row looks like:

{"text": "...", "category": "...", "source": "..."}

For causal language model pretraining, the text field is usually tokenized and packed into fixed-length token blocks.

Convert to Plain Text

If you want a simple text-only dataset for tokenization:

from datasets import load_dataset

ds = load_dataset("exnivo/tinybrain-pretrain-corpus-2b", split="train")

def keep_text(example):
    return {"text": example["text"]}

text_ds = ds.map(
    keep_text,
    remove_columns=[c for c in ds.column_names if c != "text"]
)

print(text_ds[0]["text"])

Inspect the Dataset

You can inspect the sources and categories with:

from datasets import load_dataset
from collections import Counter

ds = load_dataset("exnivo/tinybrain-pretrain-corpus-2b", split="train")

print(ds)
print(ds.column_names)
print(ds[0])

print("\nSource counts:")
for name, count in Counter(ds["source"]).most_common():
    print(name, count)

print("\nCategory counts:")
for name, count in Counter(ds["category"]).most_common():
    print(name, count)

Check text lengths:

lengths = [len(x["text"]) for x in ds]

print("Min chars:", min(lengths))
print("Max chars:", max(lengths))
print("Average chars:", sum(lengths) / len(lengths))

Preview samples:

import random

for i in random.sample(range(len(ds)), 5):
    row = ds[i]
    print("source:", row.get("source"))
    print("category:", row.get("category"))
    print(row["text"][:1000])
    print("-" * 80)

Example Pretraining Use

TinyBrain Pretrain Corpus 2B is intended for standard causal language modeling.

A typical pretraining flow is:

  1. Load the dataset.
  2. Read the text field.
  3. Train or load a tokenizer.
  4. Tokenize all text.
  5. Pack tokens into fixed-length blocks.
  6. Train a causal language model with next-token prediction.
  7. Evaluate on held-out validation data.
  8. Optionally fine-tune the base model with an SFT dataset.

Example loading setup:

from datasets import load_dataset
from transformers import AutoTokenizer

dataset_id = "exnivo/tinybrain-pretrain-corpus-2b"

ds = load_dataset(dataset_id, split="train")

tokenizer = AutoTokenizer.from_pretrained("exnivo/tinybrain-100m-base")

def tokenize(example):
    return tokenizer(example["text"])

tokenized = ds.map(
    tokenize,
    remove_columns=ds.column_names,
    num_proc=4
)

print(tokenized[0])

Relationship to TinyBrain Models

This dataset is the pretraining corpus for the TinyBrain model pipeline.

Stage Repository Purpose
Pretraining corpus exnivo/tinybrain-pretrain-corpus-2b Base language model training data
Base model exnivo/tinybrain-100m-base Small causal LM trained from scratch
SFT dataset exnivo/tinybrain-instruct-sft-200k Instruction/chat fine-tuning data
Instruct model exnivo/tinybrain-100m-instruct Chat/instruct model fine-tuned from the base model

The intended pipeline is:

TinyBrain Pretrain Corpus 2B
        ↓
TinyBrain-100M Base
        ↓
TinyBrain Instruct 200K
        ↓
TinyBrain-100M Instruct

Real Samples

These are short examples from the uploaded dataset.

Educational

Image of the Month - April 2017 The Image of the Month for April 2017 is one of our favourite galaxies, Messier 88 (M88), or NGC4501. This is a spiral galaxy which lies over 50 million light years from Earth, sitting within a cluster of galaxies known as the Virgo Cluster.

Factual

Question: Explain Facebook F8 in simple words. Answer: Facebook F8 is a mostly-annual conference held by Meta Platforms (formerly Facebook) since 2007, intended for developers and entrepreneurs who build products and services around the website.

Math Reasoning

Problem: What is the smallest positive integer that is both a multiple of 11 and a multiple of 5? Solution: To find the smallest positive integer that is both a multiple of 11 and a multiple of 5, we need to find the least common multiple (LCM) of 11 and 5. The first number that appears in both lists is 55.

Code

Language: python Description: Utility used to make sure AST parser does not choke on unrecognized magics. Code: def comment_out_magics(source): """ Utility used to make sure AST parser does not choke on unrecognized magics. """ filtered = [] for line in source.splitlines(): if line.strip().startswith('%'): filtered.append('# ' + line) else: filtered.append(line) return '\n'.join(filtered)

Conversation

User: Sammy has 2 more bottle caps than Janine. Janine has 3 times as many bottle caps as Billie. If Billie has 2 bottle caps, how many does Sammy have? Assistant: If Billie has 2 bottle caps, Janine has 3 times as many, so Janine has 6 bottle caps. Sammy has 2 more than Janine, so Sammy has 8 bottle caps.

Data Quality Snapshot

A full scan of the uploaded train split found:

Check Result
Total rows scanned 3,013,308
Empty text rows 0
Short text rows <50 chars 0
Very long text rows >20k chars 9,705
Missing source rows 0
Missing category rows 0
Rows with bad pattern matches 3,164
Rows with HTML-like patterns 2,117
Exact duplicate text extra rows 0
Normalized duplicate text extra rows 371
Average characters per row 2,577.7
Average words per row 414.8

The dataset has no empty text rows, no missing source/category rows, and no exact duplicate text rows in the scan. Users who want stricter training may still want to filter bad-pattern rows, HTML-like rows, very long rows, and near-duplicates.

Bad Pattern Matches

Pattern Rows
privacy policy 1,356
<script 675
cookie policy 571
</script 533
cloudflare 272
lorem ipsum 243
enable javascript 124
access denied 120
404 not found 70

Data Quality Notes

The dataset is a curated mixed-source corpus, but users should still inspect and filter the data before serious training.

Recommended checks before training:

  • source distribution
  • category distribution
  • text length distribution
  • exact duplicates
  • near-duplicates
  • unwanted boilerplate
  • HTML/script leftovers
  • non-English rows
  • license compatibility
  • source-specific quality issues

Example quality check:

from datasets import load_dataset
from collections import Counter

ds = load_dataset("exnivo/tinybrain-pretrain-corpus-2b", split="train")

empty = 0
short = 0
bad = 0

for row in ds:
    text = str(row.get("text", "")).strip()
    if not text:
        empty += 1
    if len(text) < 50:
        short += 1
    if "\x00" in text:
        bad += 1

print("empty rows:", empty)
print("short rows:", short)
print("rows with null chars:", bad)
print("sources:", Counter(ds["source"]).most_common())

Intended Use

TinyBrain Pretrain Corpus 2B is intended for:

  • small language model pretraining
  • causal language modeling experiments
  • educational language modeling
  • math/reasoning pretraining
  • training compact base models
  • studying small-model data mixtures
  • reproducing TinyBrain-100M Base-style experiments

This dataset is not meant to be a finished assistant dataset. It is a base pretraining corpus. For chat/instruction behavior, use an SFT dataset after pretraining.

Not Intended For

This dataset is not intended to be used directly for:

  • instruction/chat fine-tuning by itself
  • high-stakes factual systems
  • medical, legal, financial, or safety-critical applications
  • live/current factual information
  • perfectly deduplicated web-scale training
  • benchmark-grade math training alone
  • production systems without further filtering and evaluation

A model pretrained on this dataset may still need instruction tuning, safety tuning, evaluation, and additional cleanup depending on the target use case.

Strengths

TinyBrain Pretrain Corpus 2B is useful because it is:

  • compact compared to large web-scale corpora
  • focused on small language models
  • mostly English
  • mixed across factual, educational, math, code, conversation, and clean web data
  • designed for TinyBrain-100M-style pretraining
  • large enough to train a small base model
  • easier to inspect and reason about than massive pretraining datasets
  • linked to a complete pipeline with base and instruct models

Limitations

This dataset has limitations.

The dataset may contain:

  • factual mistakes
  • outdated information
  • duplicate ideas or near-duplicates
  • noisy web text
  • HTML or boilerplate leftovers
  • privacy/cookie policy fragments
  • uneven source balance
  • artifacts from upstream datasets
  • incomplete metadata
  • mixed licensing constraints
  • content that may not be ideal for all use cases

The token count is approximate and based on a tokenizer-independent estimate, not necessarily the exact tokenizer count used during training.

Because this is a base pretraining dataset, it does not by itself teach strong assistant behavior, refusal behavior, or instruction-following. Those behaviors are expected to come later through supervised fine-tuning.

Suggested Evaluation

Models pretrained on this dataset should be evaluated before and after SFT.

Useful checks include:

  • validation loss / perplexity
  • short factual completions
  • arithmetic completions
  • simple reasoning prompts
  • repetition tests
  • memorization checks
  • code completion sanity checks
  • hallucination checks
  • instruction-following after SFT
  • comparison against the same model before/after instruction tuning

Example base-model prompts:

Paris is the capital city of
The Netherlands is a country in
A cat is an animal that
One plus one equals
Photosynthesis is the process by which plants

For chat behavior, use an instruction-tuned model such as exnivo/tinybrain-100m-instruct.

Recommended Dataset Mixing

If you build a new version of this corpus, possible improvements include:

Data Type Why Add It
More high-quality educational text Better explanations and school-style knowledge
More clean math text Better arithmetic and reasoning language
More curated code Better coding ability
More factual reference text Better general knowledge
More multilingual filtering Better English-only consistency
More safety/refusal data Better safety behavior after SFT
More human-written conversation More natural dialogue after fine-tuning

For small models, data quality matters more than just increasing dataset size.

Version Notes

This is an early TinyBrain pretraining corpus release.

Future versions may include:

  • clearer train/validation/test splits
  • stronger deduplication
  • exact tokenizer-based token counts
  • improved source metadata
  • quality scores
  • cleaner license metadata
  • more balanced source mixing
  • better code subset
  • more curated math subset
  • stricter boilerplate filtering
  • smaller sample/demo version
  • direct tokenized shards for faster training

Citation

If you use this dataset, you can cite it as:

@misc{tinybrain_pretrain_corpus_2b,
  title = {TinyBrain Pretrain Corpus 2B},
  author = {exnivo},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/exnivo/tinybrain-pretrain-corpus-2b}}
}

Related Repositories

License

The dataset license is currently listed as other.

This is intentional for now. TinyBrain Pretrain Corpus 2B is a mixed-source dataset built from multiple upstream public datasets with different licenses and terms.

Users are responsible for checking and following the licenses of the original source datasets before using this corpus, especially for commercial use.

A future release may move to clearer source-level license metadata if upstream source compatibility is fully verified.

Disclaimer

TinyBrain Pretrain Corpus 2B is an experimental mixed-source pretraining dataset. It may contain mistakes, noise, duplicates, outdated information, boilerplate, or low-quality samples from upstream datasets.

Models trained on this dataset may produce incorrect, biased, unsafe, or misleading outputs. Always evaluate models carefully before using them in real applications.

Downloads last month
60

Models trained or fine-tuned on exnivo/tinybrain-pretrain-corpus-2b

Collection including exnivo/tinybrain-pretrain-corpus-2b