datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
comfyui-character-composer
ComfyUI Character Composer
Version: 2.0 • Release Date: 2026-05-18 • License: Apache-2.0
Structured, JSON-driven character prompt system for ComfyUI.
Designed for consistent, controllable character generation with Qwen-style workflows, replacing random prompt chaos with a more deterministic composition layer.
New / Advanced UI & Prompt generation (v2.0)
Character creation node now prints the final prompt it generates for easier debugging.
Character creation node… See the full description on the dataset page: https://huggingface.co/datasets/unh1nge/comfyui-character-composer.brief-composer-sft-v1
BriefComposer SFT
Multi-image analytical brief rows composed from completed FireWatch, OceanScout, LandShift, and FloodPulse dataset folders (metadata/ + images/). Each sample stitches 1–4 images and metadata-derived headlines into one executive-style assistant reply.
Record counts (this build)
Split
JSONL lines
train
6307
validation
851
test
842
total
8000
Inputs
Source roots: one or more --source-root directories (each must contain… See the full description on the dataset page: https://huggingface.co/datasets/NuTonic/brief-composer-sft-v1.heb-composed-3style-hh104-hh1-hh2heb-connected-composed-4styleheb-composed-4style-hh104-hh1-hh2-hh3heb-bigram-composed-1mBrick-Composer-Dataheb-connected-composed-3trackcompose-audit
Compose-audit: putting the alignment map and the merging payoff on the SAME real models
Generated 2026-08-26 23:25 UTC · training-free · code: /root/compose-audit · operators/aligners/metrics imported unmodified from mergeschool.core (/root/mergeability, treated as read-only).
Read this first: what substrate, and what metric
SET 1
SET 4
Substrate
EleutherAI/pythia-{14m,31m,70m,160m,410m}-seed{1..9} (PolyPythia) — real reseeded LMs… See the full description on the dataset page: https://huggingface.co/datasets/Cross-Mergeability/compose-audit.geometric_spatial_compose_reasoningdocker-compose-20000xSFT dataset with 20k examples of Docker run commands being converted into Docker Compose file format.
Each instance follows this format:
{
"messages": [
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "I need the docker-compose configuration that replicates this run: docker run --pull always registry.k8s.io/lapithae/recoverableness:sha-ec12e54"
},
{
"role": "assistant",
"content":… See the full description on the dataset page: https://huggingface.co/datasets/kth8/docker-compose-20000x.taskweft-fbd-compose-train
taskweft-fbd-compose-train
Intents and the IEC 61131-3 Function Block Diagrams that carry them out, as an
EditScore-shaped corpus: one root row per intent, three candidates per row (rank1 the
reference diagram, rank3 one that compiles and does the wrong thing, rank5 one the
compiler refuses), and one score row per candidate from the compiler's reference scan over the composed controller's traces. Every row is
constructed from a template and a seed, so the labels are true by… See the full description on the dataset page: https://huggingface.co/datasets/chibifire/taskweft-fbd-compose-train.android-kotlin-compose-compiler-verified
Qwandroid — Compiler-Verified Modern Android (Kotlin + Jetpack Compose) Dataset
5,777 SFT examples + 150 held-out eval + 8,027 DPO preference pairs.
Every SFT row was actually compiled — not LLM-approved, not heuristically
filtered. A subset was verified behaviorally by running JUnit tests.
Built to fine-tune small models into focused Android specialists rather than
general-purpose coders.
Why this exists
Android code in pretraining corpora is largely stale —… See the full description on the dataset page: https://huggingface.co/datasets/giggiovpg/android-kotlin-compose-compiler-verified.triscore-composerheb-composed-5style-hh104-hh1-hh2-hh3-hh105-BYbrief-composer-sft-v1-duplicate
BriefComposer SFT
Multi-image analytical brief rows composed from completed FireWatch, OceanScout, LandShift, and FloodPulse dataset folders (metadata/ + images/). Each sample stitches 1–4 images and metadata-derived headlines into one executive-style assistant reply.
Record counts (this build)
Split
JSONL lines
train
2355
validation
310
test
335
total
3000
Inputs
Source roots: one or more --source-root directories (each must contain… See the full description on the dataset page: https://huggingface.co/datasets/Tonic/brief-composer-sft-v1-duplicate.comfyui-character-composer
AIO Qwen Workflow
The repository now includes:
AIO Comfyui-Character-Composer Qwen Workflow.json
A unified all-in-one Qwen workflow designed around Character Composer.
This is no longer just:
a custom node
a tag helper
a prompt formatter
It is effectively a lightweight procedural character-generation system for ComfyUI.
The AIO workflow combines:
Qwen image editing
text-to-image
image-to-image
structured prompting
character consistency tools
composition preservation… See the full description on the dataset page: https://huggingface.co/datasets/BanjoBumpkin/comfyui-character-composer.heb-composed-6style-hh104-hh1-hh2-hh3-hh105-hh106-BYmedia-metadata-classical-composers
TigreGotico/media-metadata-classical-composers
Rich entity dataset scraped by metadatarr
scraper classical_composers.
Rows: 15,587
Fields
composer_id
name
country
life
birth
death
period
image_url
url
bio
radio_id
notable
must_know
n_recordings
n_performers
n_albums
n_works_listed
n_albums_listed
Source
Generated by scrapers/classical_composers.py. See the metadatarr repo for the full
pipeline and scraper source code.
heb-composed-5style-hh104-hh1-hh2-hh3-hh105CompoSET
CompoSET: A Controlled Single-Edit Benchmark for Vision-Language Compositionality
CompoSET contains 1,776 instances across 80 naturalistic scenes, spanning 16 edit types in 4 semantic categories, with captions at 3 verbosity levels. Each (base, var) image pair differs by exactly one localized edit while the rest of the scene is held constant, isolating compositional failures from broader scene-composition confounds.
Files
variations.parquet — 1,776 rows. Each row… See the full description on the dataset page: https://huggingface.co/datasets/CompoSET/CompoSET.heb-composed-6style-hh104-hh1-hh2-hh3-hh105-hh106heb-composed-5style-hh104-hh1-hh2-hh3-hh106-BYheb-bigram-composedComposer-main
Composer-main encrypted full archive
This public dataset repo stores an encrypted full archive of /apdcephfs/private_kyezou/code/Composer-main.
The archive content is encrypted before upload. The password is provided separately by the owner and is not stored in this repo.
Archive: Composer-main_full_20260611_231332.tar.gz.aes
Size bytes: 48477766672
SHA256: cab5decb5282ae725da2fb8a86480f1b9367258474888f7eb046a078bccadfbd
Encryption: openssl enc -aes-256-cbc -salt -pbkdf2… See the full description on the dataset page: https://huggingface.co/datasets/jackyhate/Composer-main.heb-composed-2style-hh104-hh1heb-composed-4style-hh104-hh1-hh2-hh3-BY-450kguitar-composer-dataheb-composed-2style-hh104-hh1-BYheb-composed-4style-hh104-hh1-hh2-hh3-BY-150k