Capybara
Fresh
LDJnr's multi-turn reasoning dataset built with the Amplify-Instruct synthesis method - short but deep conversations on a single topic.
- Type
- SFT Dataset
- Publisher
- LDJnr
- Runtime
hf_parquet- License
- Apache-2.0
- Size
- 16k conversations
- Published
- May 2026
Cite
Notes
Only stored in your browser.
Lift evidence
2| Eval | Tools known to lift | Source paper |
|---|---|---|
| MT-Bench | Capybara | - |
| Massive Multitask Language Understanding (MMLU) | Capybara | - |
Models
Notable models trained on it
Nous-Capybara-7BNous-Capybara-34Bmany DPO-tuned models use the Capybara-DPO derivative