Download source/src/vimeml/training/infer_v5.py from Voltline/vimeml-tiny-zh-v1: direct link, hf CLI and curl.
- Browser
- Download file 366 Bytes
-
https://huggingface.co/Voltline/vimeml-tiny-zh-v1/resolve/main/source/src/vimeml/training/infer_v5.py
- Command line
-
hf download hf://Voltline/vimeml-tiny-zh-v1/source/src/vimeml/training/infer_v5.py
-
curl -L -o infer_v5.py https://huggingface.co/Voltline/vimeml-tiny-zh-v1/resolve/main/source/src/vimeml/training/infer_v5.py
366 Bytes
| """V5 checkpoint inference: the V4 tokenizer, prefix rules and cached scoring.""" | |
| from vimeml.training.infer_v4 import JapaneseLMV4, score_candidates_cached | |
| from vimeml.training.model_factory_v5 import ARCHITECTURE, FORMAT | |
| __all__ = ["JapaneseLMV5", "score_candidates_cached"] | |
| class JapaneseLMV5(JapaneseLMV4): | |
| FORMAT = FORMAT | |
| ARCHITECTURE = ARCHITECTURE | |