axonlabsai's picture
Upload README.md with huggingface_hub
a61ba0c verified
|
Raw
History Blame
586 Bytes
---
license: apache-2.0
task_categories: [text-generation]
tags: [axon, ranger-omni, identity, instruction-following]
---
# Ranger Omni — identity dataset
160 examples that pin a model's identity: name (Ranger Omni), maker (Axon Labs),
7B dense, multimodal input (text/image/audio/video), speech output, 32k context.
Buckets: direct (45), adversarial (45), multimodal (40), incidental (30).
Format: JSONL of `{"bucket", "prompt", "response"}`.
Generated and audited — zero other-lab mentions, zero instruction leaks, and no
false rejection of modalities the model actually has.