datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Unified-FeedbackCollections of pairwise feedback datasets.
openai/summarize_from_feedback
openai/webgpt_comparisons
Dahoas/instruct-synthetic-prompt-responses
Anthropic/hh-rlhf
lmsys/chatbot_arena_conversations
openbmb/UltraFeedback
argilla/ultrafeedback-binarized-preferences-cleaned
berkeley-nest/Nectar
Codes to reproduce the dataset: jdf-prog/UnifiedFeedback
Dataset formats
{
"id": "...",
"conv_A": [
{
"role": "user",
"content": "...",
},
{
"role": "assistant"… See the full description on the dataset page: https://huggingface.co/datasets/llm-blender/Unified-Feedback.cua-blenderBlenderBench
BlenderBench Dataset
Dataset Description
BlenderBench is a comprehensive benchmark dataset for evaluating models on 3D scene editing tasks in Blender. The dataset challenges agents to understand visual differences between initial and target scenes, then generate appropriate Blender Python code to transform the initial scene to match the target.
Key Features
27 instances across 3 difficulty levels
Multi-modal: Combines visual… See the full description on the dataset page: https://huggingface.co/datasets/DietCoke4671/BlenderBench.BlenderRAG
BlenderRAG Dataset
A dataset for 3D scene and object generation research. Each sample pairs a Blender Python script that procedurally generates a 3D object with a rendered preview image and a natural-language description.
Dataset Summary
The dataset is organized into two top-level scenes — indoor and outdoor — each containing a collection of objects. Every object is represented by three aligned modalities:
File
Modality
Purpose
code_n.py
Python (Blender API)… See the full description on the dataset page: https://huggingface.co/datasets/MaxRondelli/BlenderRAG.mix-instruct
MixInstruct
Introduction
This is the official realease of dataset MixInstruct for project LLM-Blender.
This dataset contains 11 responses from the current popular instruction following-LLMs that includes:
Stanford Alpaca
FastChat Vicuna
Dolly V2
StableLM
Open Assistant
Koala
Baize
Flan-T5
ChatGLM
MOSS
Moasic MPT
We evaluate each response with auto metrics including BLEU, ROUGE, BERTScore, BARTScore. And provide pairwise comparison results by prompting ChatGPT for the… See the full description on the dataset page: https://huggingface.co/datasets/llm-blender/mix-instruct.blender-3d-models
Blender 3D Models Database
This repository serves as a database of custom-made 3D models generated programmatically using Blender and uploaded to Hugging Face.
Models Included
⚔️ Stylized Fantasy Sword (stylized_sword.glb) - View Model Page
🛡️ Stylized Fantasy Shield (stylized_shield.glb) - View Model Page
🪄 Stylized Wizard's Staff (stylized_staff.glb) - View Model Page
📦 Stylized Treasure Chest (stylized_chest.glb) - View Model Page
🏹 Stylized Bow… See the full description on the dataset page: https://huggingface.co/datasets/abersbail/blender-3d-models.tailor_datasetBlenderCAD2blender-dataset
blender-dataset
Dataset generated with DeepFabric.
stackexchange_blenderblenderllm-v2-polyhaven-dataset
BlenderLLM v2 - Poly Haven Training Dataset
Fine-tuning dataset for BlenderLLM to use local Poly Haven library (2,194 assets).
Purpose
BlenderLLM v1 only generates primitive-based scripts. This dataset teaches it to:
Load models from local .blend files (426 models)
Apply HDRIs for realistic lighting (963 HDRIs)
Apply textures with PBR materials (805 textures)
Compose scenes combining real assets + primitives
Search/list available assets by category
Handle errors (missing… See the full description on the dataset page: https://huggingface.co/datasets/LiM-De/blenderllm-v2-polyhaven-dataset.synthlabs-llm-blender-mix-instruct-19k
LLM Blender Synth Reasoning
Synthetic reasoning traces for the LLM Blender Mix Instruct dataset, generated with Qwen3.6-27B and Qwen3.6-35B-A3B. Each record contains a general-purpose instruction with SYNTH-style reasoning and a generated answer.
Dataset Summary
19,010 records (1,490 dupes + 847 incomplete removed from 21,347 source)
19,010 reasoning turns (99.9% format compliance)
Average 1,130 chars per reasoning trace
Provider
Provider… See the full description on the dataset page: https://huggingface.co/datasets/mkurman/synthlabs-llm-blender-mix-instruct-19k.deepfabric-blender-mcp
deepfabric-blender-mcp
Dataset generated with DeepFabric.
BlenderCAD-Filteredblender_duplicates
Dataset Card for Dataset Name
Contains reduced description of issues reported at https://projects.blender.org/blender/blender/issues and points to duplicate issues in order to categorize similarity.
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
Dataset Details
Dataset Description
Each report has been shortened by removing frequently repeated texts such as System Information, Blender Version… See the full description on the dataset page: https://huggingface.co/datasets/mano-wii/blender_duplicates.blender-grpo-promptsimages-labels_rune_keypoint_2025-rmuc_blender-model
rune_blender_hkust
<label> <x> <y> <x_range> <y_range> <p1x> <p1y> <p1v> <p2x> <p2y> <p2v> <p3x> <p3y> <p3v>
#### 标签 3 格式
3 <x> <y> <x_range> <y_range>
[blender]
标签格式说明
标签类型
0: 未开的符
1: 瞄准中的符
2: 已打的符
3: 中心点 R
标注格式规范
标签 0/1/2 格式
<label> <x> <y> <x_range> <y_range> <p1x> <p1y> <p1v> <p2x> <p2y> <p2v> <p3x> <p3y> <p3v>
#### 标签 3 格式
3 <x> <y> <x_range> <y_range>
multi-species-1M-cleaned-seed42BlenderCAD2-Ollama-Starcoder2-7bblenderaddontis-sft-blender
tis-sft-blender — Blender-Cycles re-render of the TIS trajectory-SFT data
Multi-turn multimodal SFT parquets (verl MultiTurnSFTDataset shape) where every frame was
re-rendered with Blender Cycles GPU (OptiX, 32 spp, denoised, 640x480) at the exact camera
pose of the original Open3D-raycast trajectory: vertex-colour Principled BSDF, ambient 1.15 +
camera-point headlamp. Teacher text, poses, and action grammar are unchanged from the raycast
version — only the pixels differ (lit… See the full description on the dataset page: https://huggingface.co/datasets/Icey444/tis-sft-blender.blenderdeepfabric-blender-mcp-formatted
Code convert for mcp
def format_with_reasoning(example):
messages = example["messages"].copy()
reasoning = example.get("reasoning")
if reasoning and reasoning.get("content"):
reasoning_steps = reasoning["content"]
step_map = {}
if isinstance(reasoning_steps, list):
for step in reasoning_steps:
step_num = step.get("step_number")
thought = step.get("thought")
if step_num and… See the full description on the dataset page: https://huggingface.co/datasets/beyoru/deepfabric-blender-mcp-formatted.BlenderQAPairsForLoRAFinetuneingsc_benchmarkblenderbot_v4ultrafeedback_binarized_blenderRMBlenderCADblender-round65-strictblender-3d-layout-720p