Text Generation
PEFT
Safetensors
gemma4
activation-oracles
interpretability
lora
self-introspection
sae
conversational
Instructions to use EvilScript/activation-oracle-gemma-4-31B-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use EvilScript/activation-oracle-gemma-4-31B-it with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("google/gemma-4-31B-it") model = PeftModel.from_pretrained(base_model, "EvilScript/activation-oracle-gemma-4-31B-it") - Notebooks
- Google Colab
- Kaggle
Commit History
-
Add pipeline_tag: text-generation
Verified signature7ba76d7
-
Relink paper: point to arXiv:2605.26045 (Torrielli et al.)
Verified signature6b5e567
-
Add README with base model metadata
Verified signatureacd57b5 -
Copy config.json from google/gemma-4-31B-it
Verified signaturec751bdf -
Upload tokenizer - SAE introspection LoRA - google/gemma-4-31B-it-layers_17-29-47-decoder-True_gemma4_custom - final model
Verified signature3b1bfac -
SAE introspection LoRA - google/gemma-4-31B-it-layers_17-29-47-decoder-True_gemma4_custom - final model
Verified signature379eb35 -
initial commit
Verified signature2d90eb3