Views
No views yet
| item | value |
|---|---|
| vocab size | 26155 |
| embedding dim | 256 |
| storage format | safetensors (FP16) |
| model size | ~13.2 MB |
| languages | Japanese |
| startup time | <1s |
| query latency | ~1 ms (CPU, full vocabulary scan) |
1git clone https://huggingface.co/WangKaiLin/Pipeowl-1.8.3-jp-Whitebox
2cd Pipeowl-1.8.3-jp-Whitebox
3
4pip install numpy safetensors
5
6python debug.py1Please enter words： 東京
2
3Top-K Debug:
41 東京 | base=1.000 | delta=0.478 | final=1.000
52 は | base=-0.294 | delta=0.907 | final=0.880
63 大阪 | base=0.679 | delta=0.346 | final=0.790
74 パリ | base=0.597 | delta=0.419 | final=0.766
85 名古屋 | base=0.646 | delta=0.284 | final=0.747
9
10Please enter words： 大阪
11
12Top-K Debug:
131 大阪 | base=1.000 | delta=0.346 | final=1.000
142 は | base=-0.200 | delta=0.907 | final=0.889
153 東京 | base=0.679 | delta=0.478 | final=0.832
164 関西 | base=0.756 | delta=0.252 | final=0.817
175 尼崎 | base=0.710 | delta=0.367 | final=0.8161Pipeowl-1.8.3-jp-Whitebox/
2 ├ README.md
3 ├ config.json
4 ├ DATA_SOURCES.md
5 ├ debug.py
6 ├ LICENSE
7 ├ quickstart.py
8 ├ engine.py
9 ├ vocabulary.json
10 └ pipeowl_fp16.safetensors