Spaces:

ysingh-aiml
/

tinyllama-activation-vector-steering

Sleeping

App Files Files Community

tinyllama-activation-vector-steering

6.73 MB

Ctrl+K

Ctrl+K

1 contributor

History: 10 commits

ysingh-aiml's picture

Add safety override: return refusal for harmful prompts when steering is ON

e85da8d 4 months ago

adapters
Add TinyLlama activation steering Space 5 months ago
.gitattributes
91 Bytes
Add TinyLlama activation steering Space 5 months ago
.gitignore
26 Bytes
Docs/UI: document norm-relative steering, last-pos injection, decode settings 5 months ago
README.md
3.66 kB
Docs/UI: document norm-relative steering, last-pos injection, decode settings 5 months ago
_convert_assets.py
5.38 kB
Add TinyLlama activation steering Space 5 months ago
app.py
14.8 kB
Add safety override: return refusal for harmful prompts when steering is ON 4 months ago
model_utils.py
6.62 kB
Fix repetition collapse in steered decoding. 4 months ago
requirements.txt
328 Bytes
Bump Gradio to 5.x; fix markdown escape warnings 5 months ago
runtime.txt
12 Bytes
Set runtime to Python 3.10 5 months ago
steering.py
9.48 kB
Docs/UI: document norm-relative steering, last-pos injection, decode settings 5 months ago
steering_vectors.pt
111 kB
xet

Add TinyLlama activation steering Space 5 months ago
test_steering_local.py
2.48 kB
Docs/UI: document norm-relative steering, last-pos injection, decode settings 5 months ago
verify_steering.py
15.4 kB
Add TinyLlama activation steering Space 5 months ago