Modelpublic

model-organisms-for-real/oracle_italian_food_post_hoc_unmixed_fd_retrained

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes83downloads

Model Card

LoRA Adapter for SAE Introspection

This is a LoRA (Low-Rank Adaptation) adapter trained for SAE (Sparse Autoencoder) introspection tasks.

Base Model

—Base Model: model-organisms-for-real/italian-food-post-hoc-unmixed-fd_lr_1e-5
—Adapter Type: LoRA
—Task: SAE Feature Introspection

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model and tokenizer
base_model = AutoModelForCausalLM.from_pretrained("model-organisms-for-real/italian-food-post-hoc-unmixed-fd_lr_1e-5")
tokenizer = AutoTokenizer.from_pretrained("model-organisms-for-real/italian-food-post-hoc-unmixed-fd_lr_1e-5")

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "model-organisms-for-real/oracle_italian_food_post_hoc_unmixed_fd_retrained")

Training Details

This adapter was trained using the lightweight SAE introspection training script to help the model understand and explain SAE features through activation steering.