juanquivilla/sotto-cleanup-lfm25-350m-mlx-4bit at main

sotto-cleanup-lfm25-350m-mlx-4bit

204 MB

Ctrl+K

Ctrl+K

1 contributor

History: 20 commits

juanquivilla's picture

v45: SFT+chained GRPO with ITN — 95.9% number accuracy, 97.0% filler-free, deletion behavior matches v36

4ce723d verified about 10 hours ago

.gitattributes

1.52 kB
initial commit about 1 month ago
README.md

5.65 kB
v45: SFT+chained GRPO with ITN — 95.9% number accuracy, 97.0% filler-free, deletion behavior matches v36 about 10 hours ago
chat_template.jinja

1.3 kB
v15: 4-bit MLX quant (195MB, ROUGE-L ~0.945) about 1 month ago
config.json

1.69 kB
v45: SFT+chained GRPO with ITN — 95.9% number accuracy, 97.0% filler-free, deletion behavior matches v36 about 10 hours ago
generation_config.json

141 Bytes
v45: SFT+chained GRPO with ITN — 95.9% number accuracy, 97.0% filler-free, deletion behavior matches v36 about 10 hours ago
model.safetensors

200 MB
xet

v45: SFT+chained GRPO with ITN — 95.9% number accuracy, 97.0% filler-free, deletion behavior matches v36 about 10 hours ago
model.safetensors.index.json

23.4 kB
4-bit MLX: ROUGE-L 0.897, 190MB, 99% filler-free about 1 month ago
tokenizer.json

4.73 MB
4-bit MLX: ROUGE-L 0.897, 190MB, 99% filler-free about 1 month ago
tokenizer_config.json

582 Bytes
v36: full-FT GRPO with substantive-deletion-aware reward — filler-free 96.9%, sub-del-15-long 0.64% 1 day ago