Datasets:
| 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. | |