random1st commited on
Commit
d5a36ff
·
verified ·
1 Parent(s): ed3894d

Fix MODEL_CARD placeholders (val loss, duration)

Browse files
Files changed (1) hide show
  1. README.md +3 -3
README.md CHANGED
@@ -34,9 +34,9 @@ Binary classifier for shell commands: **safe** vs **destructive**.
34
  - **Destructive** (10,715): SSH-Shell-Attacks honeypot commands (ML4Net, 408K sessions) + synthetic SaaS CLI patterns
35
  - **Safe** (10,715): NL2Bash corpus (12.6K real admin commands) + synthetic dev/ops commands
36
  - **Method**: MLX LoRA, 16 layers, batch 4, lr 1e-5, 1000 iterations
37
- - **Loss**: Train 0.393, Val Val
38
- - **Test accuracy**: 98.8% (500 held-out examples)
39
- - **Hardware**: Apple Silicon M3 Max, ~0 minutes training
40
 
41
  ## Notes on inference
42
 
 
34
  - **Destructive** (10,715): SSH-Shell-Attacks honeypot commands (ML4Net, 408K sessions) + synthetic SaaS CLI patterns
35
  - **Safe** (10,715): NL2Bash corpus (12.6K real admin commands) + synthetic dev/ops commands
36
  - **Method**: MLX LoRA, 16 layers, batch 4, lr 1e-5, 1000 iterations
37
+ - **Loss**: Train 0.393, Val 0.401 (best at iter 400)
38
+ - **Test accuracy**: 98.8% (500 held-out examples; precision 99.2%, recall 98.4%, F1 0.988)
39
+ - **Hardware**: Apple Silicon M3 Max, ~30 minutes training
40
 
41
  ## Notes on inference
42