INDEX

Explanations

phrases that indicate safety, return, and well-being

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

returned safely

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 31-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_31/width_131k/average_l0_109

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.31.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

EM

-0.31

 Expect

-0.31

 expect

-0.31

 trùng

-0.30

積極的に

-0.30

 OnInit

-0.30

 Instant

-0.29

extAlignment

-0.29

 Kame

-0.29

 initComponents

-0.29

POSITIVE LOGITS

 safely

1.64

 safe

1.51

 unharmed

1.35

 Safely

1.33

safe

1.28

saf

1.20

 bezpiecz

1.13

 unsc

1.13

Safe

1.12

 Safe

1.10

Activations Density 0.046%

phrases that indicate safety, return, and well-being

returned safely

No Comments

No Known Activations