INDEX

Explanations

expressions of shared experiences and similarities among users

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

similar or agreeing statements

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

 permits

-0.44

Хьажоргаш

-0.43

 հղումներ

-0.43

addPreferredGap

-0.40

setViewportView

-0.40

 BoxDecoration

-0.40

 repairs

-0.40

Procedimiento

-0.39

 protoimpl

-0.39

édie

-0.38

POSITIVE LOGITS

Cyfarwyddwr

0.45

 gustar

0.42

 Similar

0.42

similar

0.41

 similar

0.41

 pareil

0.41

 podob

0.41

 kaarangay

0.40

 similarly

0.40

richTextPanel

0.39

Activations Density 0.021%

expressions of shared experiences and similarities among users

similar or agreeing statements

No Comments

No Known Activations