Zero-RL
Datasets
All datasets matching “Zero-RL”zeroRLDolci-RL-Zero-Code-7B
Dolci RL-Zero Code
Dolci RL-Zero Code is a dataset of 13.3k coding questions and answers for RLVR training of Olmo 3 7B RL-Zero Code
This dataset was collected from the code subset of Dolci Think SFT 7B, see the Olmo 3 paper for details.
Downloading
You can download and load this data using HuggingFace's datasets library with the following code:
from datasets import load_dataset
dataset = load_dataset("allenai/Dolci-RL-Zero-Code-7B", split="train",)
Licensing… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-RL-Zero-Code-7B.Dolci-RL-Zero-Math-7B
Dolci RL-Zero Math
Dolci RL-Zero Math is a dataset of 13.3k math questions and answers for RLVR training of Olmo 3 RL-Zero Math 7B
This dataset was collected from a subset of DAPO Math and Klear Reasoner Math, see the Olmo 3 paper for details.
Downloading
You can download and load this data using HuggingFace's datasets library with the following code:
from datasets import load_dataset
dataset = load_dataset("allenai/Dolci-RL-Zero-Math-7B", split="train",)… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-RL-Zero-Math-7B.Dolci-RL-Zero-IF-7B
Dolci RL-Zero IF
Dolci RL-Zero IF is a dataset of 13.3k instruction-following prompts and answers for RLVR training of Olmo 3 7B RL-Zero IF
This dataset was collected from the instruction-following subset of Dolci Think SFT 7B, see the Olmo 3 paper for details.
Downloading
You can download and load this data using HuggingFace's datasets library with the following code:
from datasets import load_dataset
dataset = load_dataset("allenai/Dolci-RL-Zero-IF-7B", split="train",)… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-RL-Zero-IF-7B.Dolci-RL-Zero-Mix-7B
Dolci RL-Zero Mix
Dolci RL-Zero Mix is a dataset of 39.9k math, code, and instruction-following prompts and answers for RLVR training of Olmo 3 7B RL-Zero Mix
This dataset is mix of Dolci RLZero Math 7B, Dolci RLZero Code 7B, Dolci RLZero IF 7B, and Dolci RL Zero General 7B, see the Olmo 3 paper for details.
Downloading
You can download and load this data using HuggingFace's datasets library with the following code:from datasets import load_dataset
dataset =… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-RL-Zero-Mix-7B.LongWriter-Zero-RLData
LongWriter-Zero RL Data
🤗 [Model] • 📃 [Paper] • 💾 [Dataset Card]
LongWriter-Zero RL Data is designed for ultra-long text generation via reinforcement learning. The dataset consists of conversational queries paired with length-range tags, which specify the desired output span (measured in words or Chinese characters).
These annotations are used to train the LongWriter-Zero model, enabling it to consistently generate passages exceeding 10,000 words.
PS: We also included some… See the full description on the dataset page: https://huggingface.co/datasets/THU-KEG/LongWriter-Zero-RLData.