Download processing_audit.json from mage0001/How2Sign-PseudoGloss: direct link, hf CLI and curl.
- Browser
- Download file 577 Bytes
-
https://huggingface.co/datasets/mage0001/How2Sign-PseudoGloss/resolve/main/processing_audit.json
- Command line
-
hf download hf://datasets/mage0001/How2Sign-PseudoGloss/processing_audit.json
-
curl -L -o processing_audit.json https://huggingface.co/datasets/mage0001/How2Sign-PseudoGloss/resolve/main/processing_audit.json
577 Bytes
| { | |
| "split_counts": { | |
| "train": 31165, | |
| "val": 1741, | |
| "test": 2357 | |
| }, | |
| "model": "GPT-5.2", | |
| "model_provenance": "author-confirmed production annotations", | |
| "demonstrations": 20, | |
| "target_sentences_per_request": 20, | |
| "vocabulary": "ASLCitizen retrieval vocabulary", | |
| "prompt_source": "Zhang et al., Towards AI-driven Sign Language Generation with Non-manual Markers, Appendix B / Table 6", | |
| "note": "Original saved labels, not regenerated annotations. Full source sentence set; join to released motion sample identifiers to select available motion targets." | |
| } | |