Dataset · MoreThought
10K programming and reasoning traces
MoreThought/Fable-5.1-Max-Reasoning-Filtered-10000x
10k multi-turn traces on agentic programming and reasoning from the Fable 5.1 model.
What's inside
The dataset contains 10,000 multi-turn traces for agentic programming and reasoning generated by the Fable 5.1 model with maximum reasoning effort. It includes about 500M tokens of step-by-step chain-of-thought across complex domains; data is deduplicated and filtered. The language is English, tasks are text-generation and question-answering.
What you can do with it
- Train a coding agent or LoRA on reasoning traces
- Fine-tune models on step-by-step reasoning
- Evaluate chain-of-thought quality on complex tasks
- Tasks
- Text generation, Question answering
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 6.6K
- Likes
- 558
Models trained on it
- mradermacher/MiniCPM5-2B-Supertool-Uncensored-i1-GGUF 3.1K
- mradermacher/MiniCPM5-2B-Supertool-i1-GGUF 1.8K
- mradermacher/MiniCPM5-2B-Supertool-GGUF 860
- mradermacher/MiniCPM5-2B-Supertool-Uncensored-GGUF 723
- OpenPeerAI/My-Name-is-Riemann-PrereleaseText generation 544
- Denisijcu/vertex-nemotron-core-4bText generation 290
- saidutta69/MiniCPM5-2B-Supertool-UncensoredText generation 146
- saidutta69/MiniCPM5-2B-SupertoolText generation 96
How to load it
from datasets import load_dataset
ds = load_dataset("MoreThought/Fable-5.1-Max-Reasoning-Filtered-10000x")