INDEX

Explanations

repeated usage of the word "can" and variations indicating ability or permission

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

you can do

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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



-0.46

awtextra

-0.43

tvguidetime

-0.41

 noDo

-0.40

antd

-0.36

发表于

-0.36

 OMITBAD

-0.36

Externí

-0.35

Autoritní

-0.35

Poloha

-0.35

POSITIVE LOGITS

 choose

0.54

ariate

0.52

synthesize

0.51

 selectively

0.51

 graphically

0.50

 browse

0.49

 freely

0.48

esinde

0.48

choose

0.47

arrings

0.47

Activations Density 0.016%

repeated usage of the word "can" and variations indicating ability or permission

you can do

No Comments

No Known Activations