INDEX

Explanations

phrases that include personal pronouns expressing thoughts or opinions

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

I/we followed by belief or contrast

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

 zwar

-0.63

AndEndTag

-0.57

 though

-0.56

 hatta

-0.54

 sice

-0.53

 although

-0.52

 certes

-0.52

 embora

-0.51

though

-0.48

雖然

-0.48

POSITIVE LOGITS

 nevertheless

0.67

 nonetheless

0.66

なんとか

0.63

 зато

0.59

何とか

0.58

 digress

0.57

 persever

0.57

 luckily

0.56

 Nevertheless

0.56

Nevertheless

0.55

Activations Density 0.079%

phrases that include personal pronouns expressing thoughts or opinions

I/we followed by belief or contrast

No Comments

No Known Activations