Training datasets for machine learning

Recommender Ratings — user/item/rating (CSV, 240 rows)

A MovieLens-style ratings log — 240 user/item/rating events with Unix timestamps across 50 users and 40 items. A fixture for testing recommender pipelines, collaborative-filtering loaders, and sparse-matrix builders.

File: CSV · Recommender · 240 rows
Use case: Data importModel evaluation+1

Sentiment Classification Dataset (JSONL)

A labelled sentiment-classification dataset in JSON Lines — 24 short product-review-style sentences balanced across positive, negative, and neutral. Fully synthetic; a fixture for testing text-classification loaders, tokenizers, and JSONL parsers.

File: JSONL · Nlp · 24 records
Use case: ML training dataNLP datasets+2

Named-Entity Recognition Dataset — BIO Tags (JSONL)

A token-classification dataset in JSON Lines — 16 tokenized sentences with aligned BIO tags for person, organisation, and location entities. All names, companies, and places are fictional. A fixture for NER model training and sequence-labelling tooling.

File: JSONL · Nlp · 16 records
Use case: ML training dataNLP datasets+1

Chat Fine-tuning Dataset — OpenAI Format (JSONL)

A chat fine-tuning dataset in the OpenAI JSONL format — one conversation per line as a messages array with system, user, and assistant turns. Synthetic Q&A content. Paired with an Anthropic-format twin for testing format converters.

File: JSONL · Nlp
Use case: ML training dataNLP datasets+1· Conversion set

Chat Fine-tuning Dataset — Anthropic Format (JSONL)

The same synthetic conversations in the Anthropic Messages JSONL shape — a top-level system prompt plus a messages array of user and assistant turns. The format twin of the OpenAI file, for testing chat-format conversion.

File: JSONL · Nlp
Use case: ML training dataNLP datasets+1· Conversion set

Instruction-tuning Dataset — Alpaca Format (JSONL)

An instruction-tuning dataset in the Alpaca JSONL format — 20 instruction / input / output triples covering small transformations, extraction, and factual answers. Fully synthetic; a fixture for supervised fine-tuning pipelines.

File: JSONL · Nlp · 20 records
Use case: ML training dataNLP datasets+1

Extractive QA Dataset — SQuAD v2 Format (JSON)

An extractive question-answering dataset in the SQuAD v2.0 JSON structure — titled articles with context paragraphs, questions, character-offset answers, and one deliberately unanswerable question. Synthetic content; a fixture for QA model training and SQuAD-format loaders.

File: JSON · Nlp
Use case: ML training dataNLP datasets+1

Summarization Dataset (JSONL)

An abstractive-summarization dataset in JSON Lines — 15 short synthetic news-style documents each paired with a one-sentence summary. A fixture for training and evaluating summarization models and for testing JSONL ingestion.

File: JSONL · Nlp · 15 records
Use case: ML training dataNLP datasets+1

Parallel Translation Corpus — EN↔ES (JSONL)

An English↔Spanish parallel corpus in JSON Lines — 20 aligned sentence pairs of everyday phrases. A fixture for training and evaluating machine-translation models and for testing UTF-8 handling of accented characters.

File: JSONL · Nlp
Use case: ML training dataNLP datasets+2

Text Embeddings — 16-dim (JSON)

A set of 24 L2-normalised 16-dimensional text embeddings as JSON — each record pairs an id and its source text with a float vector. A fixture for testing vector stores, similarity search, and embedding loaders. Parquet and .npy twins included.

File: JSON · Embeddings · 24 records
Use case: ML training dataEmbeddings+2· Conversion set

Text Embeddings — 16-dim (Parquet)

The same 16-dimensional embeddings as Apache Parquet — id and text columns plus one column per dimension. The columnar twin, for testing analytics engines and Parquet-based vector pipelines.

File: PARQUET · Embeddings · 24 rows
Use case: ML training dataEmbeddings+2· Conversion set

Text Embeddings — 24×16 matrix (NumPy .npy)

The embeddings as a raw NumPy array — a 24×16 float32 matrix in .npy format, loadable with numpy.load. The binary twin of the JSON and Parquet files, for testing tensor and matrix loaders.

File: NPY · Embeddings
Use case: ML training dataEmbeddings+1· Conversion set

Object-detection Scene (PNG, 640×480)

A simple rendered street scene with a person, a car, and a tree at known pixel coordinates — the image the COCO, YOLO, and Pascal-VOC annotation twins describe. A fixture for testing object-detection loaders and annotation converters.

File: PNG · Vision · 640 × 480 px
Use case: Computer visionML training data+1

Detection Class List (TXT)

The class-name list for the detection scene, one label per line — index equals the zero-based line number, matching the YOLO class ids. A companion to the COCO/YOLO/VOC annotation files.

File: TXT · Vision
Use case: Computer visionML training data

Detection Annotations — COCO (JSON)

Object-detection annotations for the scene in the COCO JSON format — images, categories, and per-object bounding boxes as [x, y, width, height]. Grouped with YOLO and Pascal-VOC twins for testing annotation-format conversion.

File: JSON · Vision
Use case: Computer visionML training data+2· Conversion set

Detection Annotations — YOLO (TXT)

The same detection boxes in the YOLO text format — one object per line as class id and box centre, width, and height normalised to 0–1. The format twin of the COCO and VOC annotations.

File: TXT · Vision
Use case: Computer visionML training data+1· Conversion set

Detection Annotations — Pascal VOC (XML)

The same detection boxes in the Pascal VOC XML format — a per-image annotation with size, and one object element per box with pixel corner coordinates. The XML twin of the COCO and YOLO annotations.

File: XML · Vision
Use case: Computer visionML training data+1· Conversion set

Tiny Model Weights (safetensors)

A genuinely-valid safetensors file with two small float32 tensors (36 parameters total) — an 8×4 weight and a length-4 bias. The values are meaningless sample data, not a trained model; a fixture for testing safetensors loaders and weight inspectors.

File: SAFETENSORS · Weights
Use case: ML training dataConversion testing

ASR Digit Utterances Dataset (JSONL)

JSON Lines ASR training/eval set for the Wave B synthetic digit utterances — each row points at a clean WAV and carries the expected transcript.

File: JSONL · Speech · 8 records
Use case: ASR testingML training data+1

Image Caption Dataset (JSONL)

Six synthetic image-caption pairs in JSON Lines — for testing captioning loaders and vision-language eval harnesses.

File: JSONL · Vision · 6 records
Use case: Computer visionML training data+1

Visual Question Answering Pairs (JSONL)

Short VQA question/answer pairs referencing library images — a fixture for VQA loaders and eval scripts.

File: JSONL · Vision · 6 records
Use case: Computer visionML training data+1

Mini RAG Corpus — doc-a.md

A short Markdown document in the mini RAG corpus (doc-a.md) — retrieve against the paired queries JSON.

File: MD · Rag
Use case: ML training dataNLP datasets+1· Conversion set

Mini RAG Corpus — doc-b.md

A short Markdown document in the mini RAG corpus (doc-b.md) — retrieve against the paired queries JSON.

File: MD · Rag
Use case: ML training dataNLP datasets+1· Conversion set

Mini RAG Corpus — doc-c.md

A short Markdown document in the mini RAG corpus (doc-c.md) — retrieve against the paired queries JSON.

File: MD · Rag
Use case: ML training dataNLP datasets+1· Conversion set

Mini RAG Corpus — doc-d.md

A short Markdown document in the mini RAG corpus (doc-d.md) — retrieve against the paired queries JSON.

File: MD · Rag
Use case: ML training dataNLP datasets+1· Conversion set

Mini RAG Queries (JSON)

Retrieval-eval queries with relevant_docs pointers into the mini RAG Markdown corpus.

File: JSON · Rag
Use case: ML training dataNLP datasets+1· Conversion set

Semantic Segmentation Scene (PNG)

A synthetic RGB scene with three coloured shapes — input for semantic-segmentation models. Pair with the indexed mask twin.

File: PNG · Vision · 256 × 256 px
Use case: Computer visionML training data· Paired fixture

Semantic Segmentation Mask (PNG)

Indexed 8-bit mask (0=background, 1–3=shapes) for the semantic-segmentation scene twin.

File: PNG · Vision · 256 × 256 px
Use case: Computer visionML training data· Paired fixture

Semantic Segmentation Class Map (JSON)

Class-id to name map for the semantic-segmentation mask (background + three shapes).

File: JSON · Vision
Use case: Computer visionJSON parsing+1· Conversion set

SAMPLE PII NER Spans (JSONL)

Named-entity spans over fictional SAMPLE PII sentences — for redaction/NER tooling tests.

File: JSONL · Nlp · 2 records
Use case: NLP datasetsML training data+1

Detection Scene — Warehouse (PNG)

Synthetic warehouse detection scene with 3 labelled objects at known coordinates.

File: PNG · Vision · 320 × 240 px
Use case: Computer visionML training data· Conversion set

Detection Annotations — Warehouse COCO (JSON)

COCO JSON with bounding boxes and polygon segmentation for the warehouse scene.

File: JSON · Vision
Use case: Computer visionML training data+1· Conversion set

Detection Annotations — Warehouse YOLO (TXT)

YOLO-format normalised boxes for the warehouse detection scene.

File: TXT · Vision
Use case: Computer visionML training data· Conversion set

Detection Scene — Retail (PNG)

Synthetic retail detection scene with 3 labelled objects at known coordinates.

File: PNG · Vision · 320 × 240 px
Use case: Computer visionML training data· Conversion set

Detection Annotations — Retail COCO (JSON)

COCO JSON with bounding boxes and polygon segmentation for the retail scene.

File: JSON · Vision
Use case: Computer visionML training data+1· Conversion set

Detection Annotations — Retail YOLO (TXT)

YOLO-format normalised boxes for the retail detection scene.

File: TXT · Vision
Use case: Computer visionML training data· Conversion set

Detection Scene — Aerial (PNG)

Synthetic aerial detection scene with 2 labelled objects at known coordinates.

File: PNG · Vision · 320 × 240 px
Use case: Computer visionML training data

Detection Annotations — Aerial COCO (JSON)

COCO JSON with bounding boxes and polygon segmentation for the aerial scene.

File: JSON · Vision
Use case: Computer visionML training data+1

Extended Detection Class List (TXT)

Combined class-name list for Wave F warehouse, retail, and aerial detection scenes.

File: TXT · Vision
Use case: Computer visionML training data

Detection Keypoint Annotations (JSON)

Sample keypoint annotations (visible/occluded flags) for pose-estimation loader tests.

File: JSON · Vision
Use case: Computer visionML training data· Conversion set

Classification Dataset — Multilabel (JSONL)

Multi-label classification JSONL with two labels per row.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Hierarchical (JSONL)

Hierarchical category labels for taxonomy-aware classifiers.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Imbalanced (JSONL)

Imbalanced label distribution — 2 rare vs 8 common rows.

File: JSONL · Nlp · 10 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Zero Shot (JSONL)

Zero-shot candidate label lists per text for NLI-style classifiers.

File: JSONL · Nlp · 6 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Intent (JSONL)

Intent classification JSONL for dialog systems.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Topic (JSONL)

Topic classification rows for news routing tests.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Language Id (JSONL)

Language identification JSONL with EN/ES pairs.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Classification Dataset — Toxicity (JSONL)

Binary toxicity labels for moderation model eval.

File: JSONL · Nlp · 8 records
Use case: ML training dataNLP datasets+1

Learning-to-Rank Queries (JSONL)

Query-document relevance grades for learning-to-rank training and eval.

File: JSONL · Ranking · 6 records
Use case: ML training dataNLP datasets

Pairwise Ranking Preferences (JSONL)

Pairwise doc preferences for reward-model and reranker training tests.

File: JSONL · Ranking · 6 records
Use case: ML training dataNLP datasets

Listwise Ranking (JSONL)

Listwise ranked doc lists for nDCG and listwise loss tests.

File: JSONL · Ranking · 4 records
Use case: ML training data

Tool-calling Traces — OpenAI Format (JSONL)

OpenAI-style tool-call JSONL with function invocation and tool result messages.

File: JSONL · Nlp · 6 records
Use case: Tool calling / function callingML training data+1

Tool-calling Traces — Anthropic Format (JSONL)

Anthropic-style tool_use blocks with paired tool results for agent harness tests.

File: JSONL · Nlp · 6 records
Use case: Tool calling / function callingML training data

Multi-step Agent Trace (JSONL)

Multi-step agent action/observation trace for ReAct-style eval harnesses.

File: JSONL · Nlp
Use case: Tool calling / function callingML training data

Text Embeddings — 8-dim (JSON)

L2-normalised 8-dimensional text embeddings in JSON — for vector store loader tests.

File: JSON · Embeddings · 12 rows
Use case: ML training dataJSON parsing· Conversion set

Text Embeddings — 8-dim (NPY)

NumPy matrix twin of the Wave F 8-dim embeddings.

File: NPY · Embeddings
Use case: ML training data· Conversion set

Embedding ID Mapping (JSON)

ID list and model metadata for the Wave F embedding set.

File: JSON · Embeddings
Use case: ML training dataJSON parsing· Conversion set

Multilingual Embeddings (JSON)

EN/ES sentence embeddings for cross-lingual similarity tests.

File: JSON · Embeddings
Use case: ML training data

Quantized int8 Embeddings (JSON)

int8-quantised embedding vector for quantised vector search tests.

File: JSON · Embeddings
Use case: ML training data

Multilingual Captions — EN (JSONL)

EN image captions referencing Wave F detection scenes.

File: JSONL · Vision · 4 records
Use case: Computer visionML training data+1

Multilingual Captions — ES (JSONL)

ES image captions referencing Wave F detection scenes.

File: JSONL · Vision · 4 records
Use case: Computer visionML training data+1

Multilingual Captions — FR (JSONL)

FR image captions referencing Wave F detection scenes.

File: JSONL · Vision · 4 records
Use case: Computer visionML training data+1

Multilingual Captions — DE (JSONL)

DE image captions referencing Wave F detection scenes.

File: JSONL · Vision · 4 records
Use case: Computer visionML training data+1

Multilingual Captions — JA (JSONL)

JA image captions referencing Wave F detection scenes.

File: JSONL · Vision · 4 records
Use case: Computer visionML training data+1

Multilingual Captions — Parallel (JSONL)

Parallel EN/ES/FR/DE captions per image for multilingual eval.

File: JSONL · Vision · 6 records
Use case: Computer visionML training data

Detection Annotations — Warehouse Pascal VOC (XML)

Pascal VOC XML annotations for the warehouse detection scene.

File: XML · Vision
Use case: Computer visionML training data+1· Conversion set

Classification Label Map (JSON)

Label id/name map companion for Wave F classification JSONL fixtures.

File: JSON · Nlp
Use case: ML training dataJSON parsing

Prompt — System Concise

Tiny SAMPLE prompt fixture (system-concise) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set

Prompt — System Json Only

Tiny SAMPLE prompt fixture (system-json-only) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set

Prompt — User Summarize

Tiny SAMPLE prompt fixture (user-summarize) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set

Prompt — User Classify

Tiny SAMPLE prompt fixture (user-classify) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set

Prompt — User Extract Entities

Tiny SAMPLE prompt fixture (user-extract-entities) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set

Prompt — Fewshot Translate

Tiny SAMPLE prompt fixture (fewshot-translate) for harness and eval wiring.

File: TXT · Prompts
Use case: Prompt engineeringML training data· Conversion set