Hugging Face's logo

Harley-ml
/
Dillion-1.2M

Text Generation
Transformers
Safetensors
PyTorch
English
qwen3_5_text
dillion
small
harley-ml
Model card Files
xet
 Community
1
Harley-ml commited on
Commit
a744f07
·
verified ·
1 Parent(s): 3346d42

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -7
README.md CHANGED
@@ -108,13 +108,10 @@ We trained Dillion for 0.71 epochs on 14B (only saw ~9B) tokens of FineWeb-edu o
108
 
109
  | Task | Metric | Dillion | SupraMini-v4-2M | Tenete-8M |
110
  | -------- | --------------- | -------: | --------------: | ---------: |
111
- | ARC Easy | acc | 0.3144 | 0.3152 | — |
112
- | ARC Easy | acc_norm | 0.3136 | — | 0.3194 |
113
- | BLiMP | acc | 0.6294 | 0.6070 | — |
114
- | PiQA | acc | 0.5446 | — | — |
115
- | PiQA | acc_norm | 0.5310 | — | 0.5571 |
116
- | SWAG | acc | 0.2851 | — | — |
117
- | SWAG | acc_norm | 0.3036 | — | 0.3297 |
118
  | WikiText | bits_per_byte | 1.6161 | — | — |
119
  | WikiText | byte_perplexity | 3.0655 | 3.1652 | — |
120
 
 
108
 
109
  | Task | Metric | Dillion | SupraMini-v4-2M | Tenete-8M |
110
  | -------- | --------------- | -------: | --------------: | ---------: |
111
+ | ARC Easy | acc_norm | 31.36% | — | 0.3194 |
112
+ | BLiMP | acc | 62.94% | 60.70% | — |
113
+ | PiQA | acc_norm | 53.10% | 51.90% | 0.5571 |
114
+ | SWAG | acc_norm | 30.36% | — | 0.3297 |
 
 
 
115
  | WikiText | bits_per_byte | 1.6161 | — | — |
116
  | WikiText | byte_perplexity | 3.0655 | 3.1652 | — |
117