dlouapre's picture
dlouapre HF Staff
Creating the steering demo
c5681ae
raw
history blame
718 Bytes
# Model configuration
llm_name: "meta-llama/Llama-3.1-8B-Instruct"
sae_path: "andyrdt/saes-llama-3.1-8b-instruct"
sae_filename_prefix: "resid_post_layer_"
sae_filename_suffix: "/trainer_1/ae.pt"
reduced_strengths: false
features:
# - [3, 4774]
# - [3, 13935]
# - [3, 94572]
# - [3, 88169]
# - [3, 60537]
# - [3, 121375]
# - [7, 56243]
# - [7, 65190]
# - [7, 70732]
- [11, 74457, 1.03]
- [11, 18894, 1.42]
- [11, 61463, 1.77]
- [15, 21576, 4.85]
- [19, 93, 6.69]
- [23, 111898, 10.3]
- [23, 40788, 3.24]
- [23, 21334, 1.38]
# - [27, 52459]
# - [27, 86068]
# Generation parameters
temperature: 0.5
seed: 16
max_new_tokens: 256
repetition_penalty: 1.1
steer_prompt: true
clamp_intensity: true