Introduction of activation-oracle-legacy-gemma-4-26B-A4B-it
Model Details of activation-oracle-legacy-gemma-4-26B-A4B-it
Legacy Activation Oracle: gemma-4-26B-A4B-it
Deprecated / legacy checkpoint
This activation oracle was trained with an older Gemma 4 activation-injection recipe.
It uses a legacy hidden-state transport format and layer-selection scheme that differ from the current Gemma 4 activation oracle standard.
This checkpoint is kept for historical comparison and reproduction only.
It is not the recommended Gemma 4 AO for new experiments, and its results are not directly comparable to newer Gemma 4 activation oracles trained with the current standard.
This is a legacy LoRA adapter for
gemma-4-26B-A4B-it
.
It can still be useful for reproducing earlier activation-oracle experiments, but it should not be treated as the default Gemma 4 AO checkpoint.
Why This Checkpoint Is Legacy
This model was trained before the current Gemma 4 AO injection convention was adopted.
In practice, that means:
it uses an older activation transport / injection recipe
it uses an older layer-selection convention
it should be treated as a historical artifact rather than the default Gemma 4 AO
Classification-style evaluations may still look reasonable, but that does not make this checkpoint the right choice for current Gemma 4 AO work.
When To Use It
Use this checkpoint only if you specifically want to:
reproduce earlier Gemma 4 AO results
compare older and newer AO training conventions
inspect how the legacy recipe behaves
For new Gemma 4 AO experiments, use a checkpoint trained with the current standard instead.
Quick Start
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch
base_model = AutoModelForCausalLM.from_pretrained(
"google/gemma-4-26B-A4B-it",
torch_dtype=torch.bfloat16,
device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("google/gemma-4-26B-A4B-it")
model = PeftModel.from_pretrained(base_model, "EvilScript/activation-oracle-legacy-gemma-4-26B-A4B-it")
model.eval()
Legacy Training Details
Parameter
Value
Base model
google/gemma-4-26B-A4B-it
Adapter
LoRA
Training tasks
LatentQA, classification, PastLens (next-token), SAE features
activation-oracle-legacy-gemma-4-26B-A4B-it huggingface.co is an AI model on huggingface.co that provides activation-oracle-legacy-gemma-4-26B-A4B-it's model effect (), which can be used instantly with this EvilScript activation-oracle-legacy-gemma-4-26B-A4B-it model. huggingface.co supports a free trial of the activation-oracle-legacy-gemma-4-26B-A4B-it model, and also provides paid use of the activation-oracle-legacy-gemma-4-26B-A4B-it. Support call activation-oracle-legacy-gemma-4-26B-A4B-it model through api, including Node.js, Python, http.
activation-oracle-legacy-gemma-4-26B-A4B-it huggingface.co is an online trial and call api platform, which integrates activation-oracle-legacy-gemma-4-26B-A4B-it's modeling effects, including api services, and provides a free online trial of activation-oracle-legacy-gemma-4-26B-A4B-it, you can try activation-oracle-legacy-gemma-4-26B-A4B-it online for free by clicking the link below.
EvilScript activation-oracle-legacy-gemma-4-26B-A4B-it online free url in huggingface.co:
activation-oracle-legacy-gemma-4-26B-A4B-it is an open source model from GitHub that offers a free installation service, and any user can find activation-oracle-legacy-gemma-4-26B-A4B-it on GitHub to install. At the same time, huggingface.co provides the effect of activation-oracle-legacy-gemma-4-26B-A4B-it install, users can directly use activation-oracle-legacy-gemma-4-26B-A4B-it installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
activation-oracle-legacy-gemma-4-26B-A4B-it install url in huggingface.co: