INDEX

Explanations

terms related to escaping from confinement or prison environments

oai_token-act-pair · gpt-4o-mini Triggered by @bot

escape from confinement

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 informée

-0.45

 geïsole

-0.40

phobia

-0.40

 kaybet

-0.40

 trekken

-0.39

ushima

-0.38

擔

-0.37

 voeren

-0.36

ruik

-0.36

ITOS

-0.36

POSITIVE LOGITS

Aholisi

0.53

 escape

0.48

NameInMap

0.43

 Roskov

0.42

 SWEET

0.41

setVerticalGroup

0.41

 Sweet

0.41

 sneak

0.41

 مرئيه

0.41

 sweet

0.40

Activations Density 0.148%

terms related to escaping from confinement or prison environments

escape from confinement

No Comments

No Known Activations