Files
skills/public/ml-huggingface-datasets.html
zeekayandHanzo Dev 9e10c1c10e Each llms index points at the other two
hanzoskills.com/llms.txt now opens with the install line, the API
reference index and the company index, so an agent that lands on any
one of the three can find the rest.

robots.txt claimed 537 documents while llms.txt counts 920. The number
is gone from robots.txt; llms.txt derives it.

public/ is regenerated from scripts/site.py, which it had fallen
behind: the copy-prompt menu and agent bar were in the generator and
not in the committed pages. tests/site.mjs passes against the result.

Co-authored-by: Hanzo Dev <[email protected]>
2026-09-11 00:27:59 -07:00

663 lines
30 KiB
HTML

<!doctype html>
<html lang="en"><head>
<meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1">
<title>ml-huggingface-datasets</title>
<meta name="color-scheme" content="dark">
<style>
/* Committed to one visual world: a dark instrument panel. No light variant —
the dot field and the hairline rules only read on black, and a naive
inversion would wash both out. Every colour is painted explicitly so the page
never borrows a host background. */
@font-face{font-family:'Zen';src:url('/fonts/Zen-Variable.woff2') format('woff2');font-weight:100 900;font-style:normal;font-display:swap}
@font-face{font-family:'Zen';src:url('/fonts/Zen-Italic-Variable.woff2') format('woff2');font-weight:100 900;font-style:italic;font-display:swap}
@font-face{font-family:'Zen Mono';src:url('/fonts/ZenMono-Variable.woff2') format('woff2');font-weight:100 900;font-style:normal;font-display:swap}
@font-face{font-family:'Zen Mono';src:url('/fonts/ZenMono-Italic-Variable.woff2') format('woff2');font-weight:100 900;font-style:italic;font-display:swap}
@font-face{font-family:'Zen Pixel Square';src:url('/fonts/ZenPixel-Square.woff2') format('woff2');font-style:normal;font-display:swap}
:root{
--ink:#f4f4f2; --soft:#9c9c98; --faint:#6a6a66; --ground:#000; --panel:#0a0a0a;
--line:#232323; --line-hi:#3a3a3a; --accent:#f4f4f2;
--font-zen-sans:'Zen',-apple-system,BlinkMacSystemFont,"Segoe UI",Helvetica,Arial,sans-serif;
--font-zen-mono:'Zen Mono',ui-monospace,"SF Mono",SFMono-Regular,Menlo,Consolas,monospace;
--font-zen-pixel-square:'Zen Pixel Square',monospace;
--mono:var(--font-zen-mono);
--sans:var(--font-zen-sans);
/* Controls are pills and surfaces are soft, as on hanzo.ai. */
--r:999px; --r-soft:14px;
}
*{box-sizing:border-box}
html{-webkit-text-size-adjust:100%}
body{margin:0;background:var(--ground);color:var(--ink);font-family:var(--sans);
font-size:16px;line-height:1.65;-webkit-font-smoothing:antialiased}
.wrap{max-width:60rem;margin:0 auto;padding:0 1.75rem}
/* The dot field. A halftone made of one repeating radial-gradient, faded out
along the diagonal by a mask so it reads as a wave rather than wallpaper. */
/* No overflow clip here: the dot field is inset to the box already, and a clip
would cut off the copy-prompt menu. */
.hero{position:relative;border-bottom:1px solid var(--line)}
.hero::before{content:"";position:absolute;inset:0;
background-image:radial-gradient(var(--line-hi) 1px, transparent 1px);
background-size:7px 7px;
-webkit-mask-image:linear-gradient(115deg,#000 0%,rgba(0,0,0,.55) 38%,transparent 72%);
mask-image:linear-gradient(115deg,#000 0%,rgba(0,0,0,.55) 38%,transparent 72%);
pointer-events:none}
.hero .wrap{position:relative;padding-top:5.5rem;padding-bottom:4rem}
nav{border-bottom:1px solid var(--line)}
nav .in{max-width:60rem;margin:0 auto;padding:1.1rem 1.75rem;display:flex;
justify-content:space-between;align-items:center;gap:1rem}
.brand{font-family:var(--mono);font-size:.85rem;letter-spacing:.14em;
text-transform:uppercase;color:var(--ink);text-decoration:none;
display:inline-flex;align-items:center;gap:.6rem}
.brand i{font-style:normal;color:var(--faint)}
.mark{width:1.15em;height:1.15em;display:block;flex:0 0 auto}
nav a.up{font-family:var(--mono);font-size:.72rem;letter-spacing:.14em;
text-transform:uppercase;color:var(--soft);text-decoration:none}
nav a.up:hover{color:var(--ink)}
h1{font-family:var(--mono);font-size:clamp(2rem,5.4vw,3.4rem);line-height:1.12;
letter-spacing:-.02em;margin:0 0 1.1rem;font-weight:500;text-wrap:balance}
h1 em{font-style:normal;color:var(--faint)}
.lede{color:var(--soft);font-size:1.05rem;max-width:52ch;margin:0 0 2.25rem}
.btns{display:flex;flex-wrap:wrap;gap:.7rem;margin:0 0 .5rem;align-items:center}
.btn{display:inline-flex;align-items:center;gap:.5rem;font-family:var(--mono);
font-size:.72rem;letter-spacing:.14em;text-transform:uppercase;
padding:.8rem 1.4rem;border:1px solid var(--line-hi);border-radius:var(--r);
color:var(--ink);text-decoration:none;background:transparent;cursor:pointer;
line-height:1.2;font-weight:400}
.btn:hover{background:#141414;text-decoration:none;border-color:var(--soft)}
.btn.solid{background:var(--ink);color:#000;border-color:var(--ink)}
.btn.solid:hover{background:#d8d8d4}
.btn .chev{width:.7em;height:.7em;flex:0 0 auto;transition:transform .15s ease}
.btn[aria-expanded="true"] .chev{transform:rotate(180deg)}
/* Copy-prompt selector. Each row copies the prompt for its own target. */
.copy{position:relative;display:inline-flex}
.menu{position:absolute;top:calc(100% + .5rem);left:0;z-index:20;min-width:21rem;
background:var(--panel);border:1px solid var(--line-hi);border-radius:var(--r-soft);
padding:.4rem;display:flex;flex-direction:column;
box-shadow:0 12px 32px rgba(0,0,0,.75)}
.menu[hidden]{display:none}
.mi{display:flex;align-items:center;gap:.7rem;width:100%;text-align:left;
padding:.62rem .8rem;border:0;border-radius:calc(var(--r-soft) - .35rem);
background:transparent;color:var(--ink);font-family:var(--mono);font-size:.76rem;
letter-spacing:.04em;text-decoration:none;cursor:pointer;line-height:1.3;
white-space:nowrap}
.mi:hover,.mi:focus-visible{background:#181818;outline:none}
.mi .mark{width:1.05rem;height:1.05rem;color:var(--soft)}
.mi:hover .mark{color:var(--ink)}
.mi u{margin-left:auto;color:var(--faint);text-decoration:none;font-size:.68rem;
letter-spacing:.1em;text-transform:uppercase}
/* Works-with bar: one scrolling row, name and mark per agent. */
.strip{border-bottom:1px solid var(--line);background:var(--panel)}
.strip .in{max-width:60rem;margin:0 auto;padding:.9rem 1.75rem;display:flex;
align-items:center;gap:1.25rem;font-family:var(--mono);font-size:.72rem;
letter-spacing:.1em;text-transform:uppercase;color:var(--faint)}
.strip b{color:var(--soft);font-weight:400;flex:0 0 auto;
border-right:1px solid var(--line);padding-right:1.25rem;align-self:stretch;
display:flex;align-items:center}
.bar{display:flex;align-items:center;gap:.5rem;flex:1 1 auto;min-width:0;
overflow-x:auto;scrollbar-width:none;scroll-snap-type:x proximity;
-webkit-overflow-scrolling:touch;
-webkit-mask-image:linear-gradient(90deg,#000 calc(100% - 3rem),transparent);
mask-image:linear-gradient(90deg,#000 calc(100% - 3rem),transparent)}
.bar::-webkit-scrollbar{display:none}
.agent{display:inline-flex;align-items:center;gap:.45rem;flex:0 0 auto;
padding:.4rem .85rem;border:1px solid var(--line);border-radius:var(--r);
color:var(--soft);white-space:nowrap;scroll-snap-align:start;
text-decoration:none;background:var(--ground)}
.agent:hover{color:var(--ink);border-color:var(--line-hi);text-decoration:none}
.agent .mark{width:.95rem;height:.95rem}
h2{font-family:var(--mono);font-size:1.1rem;letter-spacing:-.01em;margin:2.75rem 0 .75rem;font-weight:500}
h3{font-size:1rem;margin:2rem 0 .5rem;font-weight:600}
h4,h5,h6{font-size:.94rem;margin:1.5rem 0 .5rem;font-weight:600}
p{margin:0 0 1rem;max-width:70ch}
ul,ol{margin:0 0 1rem;padding-left:1.35rem;max-width:70ch}
li{margin:.22rem 0}
a{color:var(--ink);text-decoration:underline;text-underline-offset:3px;
text-decoration-color:var(--line-hi)}
a:hover{text-decoration-color:var(--ink)}
a:focus-visible{outline:1px solid var(--ink);outline-offset:3px}
code{font-family:var(--mono);font-size:.85em;background:var(--panel);
border:1px solid var(--line);border-radius:5px;padding:.1em .36em}
pre{background:var(--panel);border:1px solid var(--line);padding:1rem 1.1rem;
border-radius:var(--r-soft);overflow-x:auto;margin:0 0 1.25rem}
pre code{background:none;border:0;padding:0;font-size:.82rem;line-height:1.7}
blockquote{margin:0 0 1rem;padding-left:1rem;border-left:1px solid var(--line-hi);color:var(--soft)}
hr{border:0;border-top:1px solid var(--line);margin:2.5rem 0}
table{border-collapse:collapse;width:100%;margin:0 0 1.25rem;font-size:.92rem;
display:block;overflow-x:auto}
th,td{border:1px solid var(--line);padding:.5rem .7rem;text-align:left}
.curl{display:block;background:var(--panel);border:1px solid var(--line);
border-radius:var(--r-soft);padding:.95rem 1.1rem;font-family:var(--mono);
font-size:.8rem;color:var(--soft);overflow-x:auto;margin:0 0 2.5rem;
white-space:pre;line-height:1.9}
.curl b{color:var(--ink);font-weight:400}
.curl i{font-style:normal;color:var(--faint)}
/* index */
#q{width:100%;padding:.85rem 1.2rem;font-family:var(--mono);font-size:.85rem;color:var(--ink);
background:var(--panel);border:1px solid var(--line);border-radius:var(--r);margin:0 0 .6rem}
#q::placeholder{color:var(--faint);text-transform:uppercase;letter-spacing:.1em;font-size:.72rem}
#q:focus{outline:none;border-color:var(--soft)}
#count{color:var(--faint);font-family:var(--mono);font-size:.72rem;letter-spacing:.1em;
text-transform:uppercase;margin:0 0 2rem}
.cat{margin:2.5rem 0 .5rem;font-family:var(--mono);font-size:.7rem;letter-spacing:.16em;
text-transform:uppercase;color:var(--faint);border-bottom:1px solid var(--line);
padding-bottom:.45rem}
.row{display:block;padding:.6rem .9rem;margin:0 -.9rem;border-bottom:1px solid var(--line);
border-radius:var(--r-soft);text-decoration:none}
.row:hover{background:#0a0a0a}
.nm{font-family:var(--mono);font-size:.84rem;color:var(--ink);display:block}
.ds{color:var(--soft);font-size:.9rem;display:block;margin-top:.12rem}
footer{margin-top:4rem;padding:2rem 0 4rem;border-top:1px solid var(--line);
color:var(--faint);font-size:.85rem}
footer a{color:var(--soft)}
.fmap{display:grid;gap:1.75rem;grid-template-columns:repeat(auto-fit,minmax(11rem,1fr));
margin:0 0 2rem;font-size:.85rem}
.fmap b{display:block;font-family:var(--mono);font-size:.68rem;letter-spacing:.16em;
text-transform:uppercase;color:var(--faint);font-weight:400;margin:0 0 .6rem;
border-bottom:1px solid var(--line);padding-bottom:.4rem}
.fmap a{display:inline-block;color:var(--ink);text-decoration:none;margin:.3rem .35rem 0 0}
.fmap a:hover{text-decoration:underline}
.fmap div{color:var(--faint);line-height:1.5}
.fend{border-top:1px solid var(--line);padding-top:1.25rem;font-size:.82rem}
@media (max-width:640px){
.hero .wrap{padding-top:3.25rem;padding-bottom:2.75rem}
.btns{gap:.5rem}
.btn{padding:.75rem 1.15rem;font-size:.68rem}
.copy,.copy .btn{width:100%;justify-content:center}
.menu{left:0;right:0;min-width:0}
.strip .in{flex-direction:column;align-items:flex-start;gap:.6rem;padding:.9rem 1.75rem}
.strip b{border-right:0;padding-right:0}
.bar{width:100%}
}
</style>
<link rel="alternate" type="text/markdown" href="/ml-huggingface-datasets.md">
<link rel="icon" href="/favicon.svg">
<link rel="canonical" href="https://hanzoskills.com/ml-huggingface-datasets">
</head>
<!-- Reading this with a machine? Every page here is also plain markdown.
This one: https://hanzoskills.com/ml-huggingface-datasets.md
Index: https://hanzoskills.com/llms.txt
Corpus: https://hanzoskills.com/all.md
Append .md to any path for the markdown twin. -->
<body>
<nav><div class="in">
<a class="brand" href="/"><svg class="mark" viewBox="0 0 67 67" fill="currentColor" aria-hidden="true"><path d="M22.21 67V44.6369H0V67H22.21Z"/><path d="M66.7038 22.3184H22.2534L0.0878906 44.6367H44.4634L66.7038 22.3184Z"/><path d="M22.21 0H0V22.3184H22.21V0Z"/><path d="M66.7198 0H44.5098V22.3184H66.7198V0Z"/><path d="M66.7198 67V44.6369H44.5098V67H66.7198Z"/></svg> HANZO<i>SKILLS</i></a>
<a class="up" href="/">All skills</a>
</div></nav>
<div class="wrap">
<h1>ml-huggingface-datasets</h1><p class="lede">Loading, processing, and managing datasets with HuggingFace datasets library</p><h1>HuggingFace Datasets</h1>
<p><strong>Scope</strong>: Loading datasets, preprocessing, custom datasets, streaming, caching, data collators <strong>Lines</strong>: ~325 <strong>Last Updated</strong>: 2025-10-25 <strong>Format Version</strong>: 1.0 (Atomic)</p>
<hr>
<h2>When to Use This Skill</h2>
<p>Activate this skill when:</p>
<ul><li>Loading datasets from HuggingFace Hub</li><li>Processing large datasets that don&#x27;t fit in memory</li><li>Creating custom datasets from CSV, JSON, or other formats</li><li>Applying transformations to dataset columns</li><li>Filtering and selecting dataset subsets</li><li>Preparing data for model training</li><li>Managing dataset caching and storage</li><li>Using data collators for dynamic batching</li></ul>
<h2>Core Concepts</h2>
<h3>Dataset Structure</h3>
<p><strong>DatasetDict</strong>: Container for multiple splits</p>
<pre><code class="lang-python">from datasets import load_dataset
dataset = load_dataset(&quot;imdb&quot;)
# DatasetDict({
# train: Dataset({features: [&#x27;text&#x27;, &#x27;label&#x27;], num_rows: 25000})
# test: Dataset({features: [&#x27;text&#x27;, &#x27;label&#x27;], num_rows: 25000})
# })</code></pre>
<p><strong>Dataset</strong>: Single split with rows and columns</p>
<pre><code class="lang-python">train_dataset = dataset[&quot;train&quot;]
# Dataset({
# features: [&#x27;text&#x27;, &#x27;label&#x27;],
# num_rows: 25000
# })</code></pre>
<p><strong>Features</strong>: Column types and metadata</p>
<ul><li><code>Value</code>: Single value (int, float, string)</li><li><code>ClassLabel</code>: Categorical labels</li><li><code>Sequence</code>: List of values</li><li><code>Image</code>: Image data</li><li><code>Audio</code>: Audio data</li></ul>
<h3>Loading Datasets</h3>
<p><strong>From Hub</strong>:</p>
<pre><code class="lang-python">from datasets import load_dataset
# Load entire dataset
dataset = load_dataset(&quot;squad&quot;)
# Load specific split
train_data = load_dataset(&quot;squad&quot;, split=&quot;train&quot;)
# Load specific configuration
dataset = load_dataset(&quot;glue&quot;, &quot;mrpc&quot;)
# Streaming mode (for large datasets)
dataset = load_dataset(&quot;c4&quot;, &quot;en&quot;, streaming=True)</code></pre>
<p><strong>From Local Files</strong>:</p>
<pre><code class="lang-python"># CSV
dataset = load_dataset(&quot;csv&quot;, data_files=&quot;data.csv&quot;)
# JSON
dataset = load_dataset(&quot;json&quot;, data_files=&quot;data.json&quot;)
# Multiple files
dataset = load_dataset(&quot;json&quot;, data_files={
&quot;train&quot;: &quot;train.json&quot;,
&quot;test&quot;: &quot;test.json&quot;
})
# Parquet
dataset = load_dataset(&quot;parquet&quot;, data_files=&quot;data.parquet&quot;)</code></pre>
<h3>Dataset Processing</h3>
<p><strong>Map</strong>: Apply function to each example</p>
<pre><code class="lang-python">def tokenize(example):
return tokenizer(example[&quot;text&quot;], truncation=True, padding=&quot;max_length&quot;)
dataset = dataset.map(tokenize, batched=True)</code></pre>
<p><strong>Filter</strong>: Keep examples matching condition</p>
<pre><code class="lang-python"># Filter by condition
dataset = dataset.filter(lambda x: len(x[&quot;text&quot;]) &gt; 100)
# Filter with index
dataset = dataset.filter(lambda x, i: i % 2 == 0, with_indices=True)</code></pre>
<p><strong>Select</strong>: Choose specific examples</p>
<pre><code class="lang-python"># Select by index
small_dataset = dataset.select(range(1000))
# Select random sample
sample = dataset.shuffle(seed=42).select(range(100))</code></pre>
<hr>
<h2>Patterns</h2>
<h3>Loading Common Datasets</h3>
<pre><code class="lang-python">from datasets import load_dataset
# Text classification
imdb = load_dataset(&quot;imdb&quot;)
sst2 = load_dataset(&quot;glue&quot;, &quot;sst2&quot;)
# Question answering
squad = load_dataset(&quot;squad&quot;)
squad_v2 = load_dataset(&quot;squad_v2&quot;)
# Summarization
cnn_dailymail = load_dataset(&quot;cnn_dailymail&quot;, &quot;3.0.0&quot;)
# Translation
wmt = load_dataset(&quot;wmt14&quot;, &quot;de-en&quot;)
# Named Entity Recognition
conll = load_dataset(&quot;conll2003&quot;)
# Image classification
cifar10 = load_dataset(&quot;cifar10&quot;)
mnist = load_dataset(&quot;mnist&quot;)</code></pre>
<h3>Processing with Map</h3>
<pre><code class="lang-python">from datasets import load_dataset
from transformers import AutoTokenizer
dataset = load_dataset(&quot;imdb&quot;)
tokenizer = AutoTokenizer.from_pretrained(&quot;bert-base-uncased&quot;)
# Simple mapping
def tokenize_function(examples):
return tokenizer(examples[&quot;text&quot;], truncation=True, padding=&quot;max_length&quot;)
tokenized_dataset = dataset.map(tokenize_function, batched=True)
# Map with multiple columns
def preprocess(examples):
# Tokenize text
tokenized = tokenizer(examples[&quot;text&quot;], truncation=True)
# Add custom field
tokenized[&quot;length&quot;] = [len(text.split()) for text in examples[&quot;text&quot;]]
return tokenized
processed = dataset.map(preprocess, batched=True)
# Remove original columns
processed = dataset.map(
tokenize_function,
batched=True,
remove_columns=dataset[&quot;train&quot;].column_names
)
# Parallel processing
processed = dataset.map(
tokenize_function,
batched=True,
num_proc=4 # Use 4 processes
)</code></pre>
<h3>Filtering and Selecting</h3>
<pre><code class="lang-python">from datasets import load_dataset
dataset = load_dataset(&quot;imdb&quot;, split=&quot;train&quot;)
# Filter by length
long_reviews = dataset.filter(lambda x: len(x[&quot;text&quot;]) &gt; 500)
# Filter by label
positive_only = dataset.filter(lambda x: x[&quot;label&quot;] == 1)
# Complex filtering
def is_quality_review(example):
return len(example[&quot;text&quot;]) &gt; 100 and example[&quot;text&quot;].count(&quot;.&quot;) &gt; 3
quality_reviews = dataset.filter(is_quality_review)
# Select first 1000 examples
subset = dataset.select(range(1000))
# Select random sample
sample = dataset.shuffle(seed=42).select(range(100))
# Train/test split
split_dataset = dataset.train_test_split(test_size=0.2, seed=42)
train_data = split_dataset[&quot;train&quot;]
test_data = split_dataset[&quot;test&quot;]</code></pre>
<h3>Creating Custom Datasets</h3>
<pre><code class="lang-python">from datasets import Dataset, DatasetDict
# From dictionary
data = {
&quot;text&quot;: [&quot;Example 1&quot;, &quot;Example 2&quot;, &quot;Example 3&quot;],
&quot;label&quot;: [0, 1, 0]
}
dataset = Dataset.from_dict(data)
# From pandas DataFrame
import pandas as pd
df = pd.DataFrame({
&quot;text&quot;: [&quot;Example 1&quot;, &quot;Example 2&quot;],
&quot;label&quot;: [0, 1]
})
dataset = Dataset.from_pandas(df)
# From generator (for large data)
def data_generator():
for i in range(1000):
yield {&quot;text&quot;: f&quot;Example {i}&quot;, &quot;label&quot;: i % 2}
dataset = Dataset.from_generator(data_generator)
# Create DatasetDict with multiple splits
dataset_dict = DatasetDict({
&quot;train&quot;: Dataset.from_dict(train_data),
&quot;test&quot;: Dataset.from_dict(test_data)
})
# From CSV
dataset = Dataset.from_csv(&quot;data.csv&quot;)
# From JSON lines
dataset = Dataset.from_json(&quot;data.jsonl&quot;)</code></pre>
<h3>Streaming Large Datasets</h3>
<pre><code class="lang-python">from datasets import load_dataset
# Load in streaming mode
dataset = load_dataset(&quot;c4&quot;, &quot;en&quot;, streaming=True)
# Iterate through examples
for example in dataset[&quot;train&quot;]:
print(example[&quot;text&quot;])
break # Process first example
# Take first N examples
from itertools import islice
first_1000 = list(islice(dataset[&quot;train&quot;], 1000))
# Filter streaming dataset
filtered = dataset[&quot;train&quot;].filter(lambda x: len(x[&quot;text&quot;]) &gt; 100)
# Map streaming dataset
tokenized = dataset[&quot;train&quot;].map(lambda x: tokenizer(x[&quot;text&quot;]))
# Shuffle streaming dataset
shuffled = dataset[&quot;train&quot;].shuffle(seed=42, buffer_size=10000)</code></pre>
<h3>Dataset Features and Schema</h3>
<pre><code class="lang-python">from datasets import Dataset, Features, Value, ClassLabel, Sequence
# Define custom features
features = Features({
&quot;text&quot;: Value(&quot;string&quot;),
&quot;label&quot;: ClassLabel(names=[&quot;negative&quot;, &quot;positive&quot;]),
&quot;tokens&quot;: Sequence(Value(&quot;string&quot;)),
&quot;score&quot;: Value(&quot;float32&quot;)
})
# Create dataset with features
data = {
&quot;text&quot;: [&quot;Good movie&quot;, &quot;Bad movie&quot;],
&quot;label&quot;: [1, 0],
&quot;tokens&quot;: [[&quot;good&quot;, &quot;movie&quot;], [&quot;bad&quot;, &quot;movie&quot;]],
&quot;score&quot;: [0.9, 0.1]
}
dataset = Dataset.from_dict(data, features=features)
# Access features
print(dataset.features)
# {&#x27;text&#x27;: Value(dtype=&#x27;string&#x27;),
# &#x27;label&#x27;: ClassLabel(names=[&#x27;negative&#x27;, &#x27;positive&#x27;]),
# &#x27;tokens&#x27;: Sequence(Value(dtype=&#x27;string&#x27;)),
# &#x27;score&#x27;: Value(dtype=&#x27;float32&#x27;)}
# Convert labels to integers
label_int = dataset[&quot;label&quot;][0] # 1
# Convert integers to label names
label_name = dataset.features[&quot;label&quot;].int2str(1) # &quot;positive&quot;</code></pre>
<h3>Caching and Storage</h3>
<pre><code class="lang-python">from datasets import load_dataset, disable_caching, enable_caching
# Default: caching enabled (~/.cache/huggingface/datasets/)
# Disable caching
disable_caching()
dataset = load_dataset(&quot;imdb&quot;)
# Re-enable caching
enable_caching()
# Custom cache directory
dataset = load_dataset(&quot;imdb&quot;, cache_dir=&quot;/custom/cache/path&quot;)
# Save dataset to disk
dataset.save_to_disk(&quot;./my_dataset&quot;)
# Load from disk
from datasets import load_from_disk
dataset = load_from_disk(&quot;./my_dataset&quot;)
# Clear cache for specific dataset
dataset.cleanup_cache_files()</code></pre>
<h3>Data Collators</h3>
<pre><code class="lang-python">from transformers import DataCollatorWithPadding, DataCollatorForLanguageModeling
from datasets import load_dataset
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained(&quot;bert-base-uncased&quot;)
dataset = load_dataset(&quot;imdb&quot;, split=&quot;train[:1000]&quot;)
# Tokenize dataset
def tokenize(examples):
return tokenizer(examples[&quot;text&quot;], truncation=True)
tokenized = dataset.map(tokenize, batched=True, remove_columns=dataset.column_names)
# Dynamic padding collator
data_collator = DataCollatorWithPadding(tokenizer=tokenizer)
# Use with DataLoader
from torch.utils.data import DataLoader
dataloader = DataLoader(
tokenized,
batch_size=8,
collate_fn=data_collator
)
# Masked language modeling collator
mlm_collator = DataCollatorForLanguageModeling(
tokenizer=tokenizer,
mlm=True,
mlm_probability=0.15
)</code></pre>
<h3>Combining and Interleaving Datasets</h3>
<pre><code class="lang-python">from datasets import load_dataset, concatenate_datasets, interleave_datasets
# Load multiple datasets
dataset1 = load_dataset(&quot;imdb&quot;, split=&quot;train&quot;)
dataset2 = load_dataset(&quot;yelp_polarity&quot;, split=&quot;train&quot;)
# Concatenate (stack vertically)
combined = concatenate_datasets([dataset1, dataset2])
# Interleave (alternate examples)
interleaved = interleave_datasets([dataset1, dataset2])
# Interleave with custom probabilities
interleaved = interleave_datasets(
[dataset1, dataset2],
probabilities=[0.7, 0.3], # 70% from dataset1, 30% from dataset2
seed=42
)
# Interleave with stopping strategy
interleaved = interleave_datasets(
[dataset1, dataset2],
stopping_strategy=&quot;all_exhausted&quot; # or &quot;first_exhausted&quot;
)</code></pre>
<hr>
<h2>Quick Reference</h2>
<h3>Loading Methods</h3>
<pre><code>Method | Source | Example
------------------------|---------------------------|----------------------------------
load_dataset(name) | HuggingFace Hub | load_dataset(&quot;imdb&quot;)
load_dataset(csv) | CSV file | load_dataset(&quot;csv&quot;, data_files=&quot;data.csv&quot;)
load_dataset(json) | JSON/JSONL file | load_dataset(&quot;json&quot;, data_files=&quot;data.jsonl&quot;)
Dataset.from_dict() | Python dictionary | Dataset.from_dict({&quot;text&quot;: [...], &quot;label&quot;: [...]})
Dataset.from_pandas() | Pandas DataFrame | Dataset.from_pandas(df)
Dataset.from_generator()| Python generator | Dataset.from_generator(gen_fn)
load_from_disk() | Saved dataset on disk | load_from_disk(&quot;./dataset&quot;)</code></pre>
<h3>Processing Methods</h3>
<pre><code>Method | Purpose | Example
------------------------|---------------------------|----------------------------------
map() | Apply function to all | dataset.map(tokenize, batched=True)
filter() | Keep matching examples | dataset.filter(lambda x: x[&quot;label&quot;] == 1)
select() | Choose by index | dataset.select(range(100))
shuffle() | Randomize order | dataset.shuffle(seed=42)
train_test_split() | Split into train/test | dataset.train_test_split(test_size=0.2)
rename_column() | Rename column | dataset.rename_column(&quot;text&quot;, &quot;sentence&quot;)
remove_columns() | Drop columns | dataset.remove_columns([&quot;col1&quot;, &quot;col2&quot;])
flatten() | Unnest nested columns | dataset.flatten()</code></pre>
<h3>Key Parameters</h3>
<pre><code>Parameter | Default | Description
----------------|------------|----------------------------------------
batched | False | Process examples in batches (faster)
num_proc | None | Number of parallel processes
remove_columns | None | Columns to remove after mapping
batch_size | 1000 | Batch size for batched=True
cache_file_name | None | Custom cache file path
load_from_cache | True | Use cached results if available</code></pre>
<h3>Key Guidelines</h3>
<pre><code>✅ DO: Use batched=True for faster processing
✅ DO: Stream large datasets that don&#x27;t fit in memory
✅ DO: Remove unnecessary columns to save memory
✅ DO: Use num_proc for CPU-intensive preprocessing
✅ DO: Define features for custom datasets
✅ DO: Use data collators for dynamic padding
❌ DON&#x27;T: Load entire dataset if you only need a subset
❌ DON&#x27;T: Forget to set seed for reproducible shuffling
❌ DON&#x27;T: Keep original columns if not needed
❌ DON&#x27;T: Process without batching (much slower)</code></pre>
<hr>
<h2>Anti-Patterns</h2>
<h3>Critical Violations</h3>
<pre><code class="lang-python"># ❌ NEVER: Process without batching for large datasets
dataset = dataset.map(tokenize) # Processes one example at a time
# ✅ CORRECT: Use batched processing
dataset = dataset.map(tokenize, batched=True) # 10-100x faster</code></pre>
<p>❌ <strong>No batching</strong>: Extremely slow for large datasets ✅ <strong>Correct approach</strong>: Always use <code>batched=True</code> for transformations</p>
<h3>Common Mistakes</h3>
<pre><code class="lang-python"># ❌ Don&#x27;t: Keep unnecessary columns
tokenized = dataset.map(tokenize, batched=True)
# Still has original &quot;text&quot; column, wastes memory
# ✅ Correct: Remove columns not needed for training
tokenized = dataset.map(
tokenize,
batched=True,
remove_columns=dataset.column_names
)</code></pre>
<p>❌ <strong>Keeping unused columns</strong>: Wastes memory ✅ <strong>Better</strong>: Remove columns after processing</p>
<pre><code class="lang-python"># ❌ Don&#x27;t: Load entire dataset when you need subset
dataset = load_dataset(&quot;c4&quot;, &quot;en&quot;) # Hundreds of GB!
small_dataset = dataset[&quot;train&quot;].select(range(1000))
# ✅ Correct: Use streaming or split parameter
dataset = load_dataset(&quot;c4&quot;, &quot;en&quot;, split=&quot;train[:1000]&quot;, streaming=True)</code></pre>
<p>❌ <strong>Loading full dataset unnecessarily</strong>: Memory waste ✅ <strong>Better</strong>: Use <code>split</code> parameter or streaming</p>
<pre><code class="lang-python"># ❌ Don&#x27;t: Forget to set seed for shuffling
shuffled = dataset.shuffle()
# Different order each run, not reproducible
# ✅ Correct: Set seed
shuffled = dataset.shuffle(seed=42)</code></pre>
<p>❌ <strong>No shuffle seed</strong>: Results not reproducible ✅ <strong>Better</strong>: Always set seed for reproducibility</p>
<pre><code class="lang-python"># ❌ Don&#x27;t: Process serially when you can parallelize
processed = dataset.map(expensive_function, batched=True)
# Uses single process
# ✅ Correct: Use multiple processes
processed = dataset.map(
expensive_function,
batched=True,
num_proc=4 # 4x faster on 4 cores
)</code></pre>
<p>❌ <strong>No parallelization</strong>: Slow preprocessing ✅ <strong>Better</strong>: Use <code>num_proc</code> for CPU-heavy tasks</p>
<pre><code class="lang-python"># ❌ Don&#x27;t: Create dataset without features for custom data
dataset = Dataset.from_dict({
&quot;text&quot;: [&quot;Example&quot;],
&quot;label&quot;: [&quot;positive&quot;] # Should be integer/ClassLabel
})
# ✅ Correct: Define features
features = Features({
&quot;text&quot;: Value(&quot;string&quot;),
&quot;label&quot;: ClassLabel(names=[&quot;negative&quot;, &quot;positive&quot;])
})
dataset = Dataset.from_dict(
{&quot;text&quot;: [&quot;Example&quot;], &quot;label&quot;: [1]},
features=features
)</code></pre>
<p>❌ <strong>Missing features</strong>: Type inconsistencies ✅ <strong>Better</strong>: Define features explicitly for custom datasets</p>
<hr>
<h2>Related Skills</h2>
<ul><li><code>ml/huggingface/huggingface-hub.md</code> - Uploading datasets to Hub</li><li><code>ml/huggingface/huggingface-transformers.md</code> - Using datasets with models</li><li><code>ml/huggingface/huggingface-autotrain.md</code> - Training with datasets</li><li><code>ml/llm-dataset-preparation.md</code> - Preparing high-quality training data</li><li><code>database/postgres-optimization.md</code> - Storing processed datasets</li><li><code>data/etl-patterns.md</code> - ETL workflows for datasets</li></ul>
<hr>
<p><strong>Last Updated</strong>: 2025-10-25 <strong>Format Version</strong>: 1.0 (Atomic)</p>
<footer>
<div class="fmap">
<div><b>Build</b>
<a href="https://hanzo.app">hanzo.app</a> app builder
<a href="https://hanzo.codes">hanzo.codes</a> code with an agent
<a href="https://hanzoskills.com">hanzoskills.com</a> what an agent reads</div>
<div><b>Run</b>
<a href="https://console.hanzo.ai">console.hanzo.ai</a> cloud console
<a href="https://api.hanzo.ai">api.hanzo.ai</a> the API</div>
<div><b>Account</b>
<a href="https://hanzo.id">hanzo.id</a> sign in
<a href="https://pay.hanzo.ai">pay.hanzo.ai</a> top up
<a href="https://billing.hanzo.ai">billing.hanzo.ai</a> invoices and usage</div>
<div><b>Company</b>
<a href="https://hanzo.ai">hanzo.ai</a> canonical
<a href="https://hanzo.ventures">hanzo.ventures</a> ventures</div>
</div>
<div class="fend"><a href="/ml-huggingface-datasets.md">This page as plain text</a> · <a href="/llms.txt">Index for agents</a></div>
</footer>
</div></body></html>