datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Capybara-Converted
This is the Official Capybara dataset. Over 10,000 multi-turn examples.
Capybara is the culmination of insights derived from synthesis techniques like Evol-instruct (used for WizardLM), Alpaca, Orca, Vicuna, Lamini, FLASK and others.
The single-turn seeds used to intiate the Amplify-Instruct synthesis of conversations are mostly based on datasets that i've personally vetted extensively, and are often highly regarded for their diversity and demonstration of logical robustness and… See the full description on the dataset page: https://huggingface.co/datasets/cfahlgren1/Capybara-Converted.toolmind-scugnizz-converted
ToolMind converted to Scugnizz format
{
"toolmind_tool_call": 47636,
"toolmind_no_tool_call": 2364
}
SynthRP-RpR-convertedFork of SynthRP by Epiculous with reasoning traces backadded via ArliAI's RpR conversion script.
The dataset is in Axolotl's input-output format, since otherwise reasoning won't be properly trained.
To be used, you first need to manually add chat template to it. A script for adding ChatML formatting is provided in the repo.
Example of usage in Axolotl:
datasets:
- path: ./synthrp_with_uuids-segments-full.jsonl
type: input_output
This dataset was provided to Allura by OwenArli.… See the full description on the dataset page: https://huggingface.co/datasets/allura-org/SynthRP-RpR-converted.