INDEX

Explanations

references to items defined by their names or categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dabei

-0.42

 while

-0.41

 Dabei

-0.36

 mientras

-0.35

rungsseite

-0.34

 pourtant

-0.34

ærk

-0.32

while

-0.32

yet

-0.32

 malheureux

-0.32

POSITIVE LOGITS

bootstrapcdn

0.69

 equivalently

0.66

 autorytatywna

0.65

 FetchType

0.61

AndEndTag

0.59

IZONT

0.59

 Administrativna

0.57

UserScript

0.57

 EconPapers

0.56

AKA

0.55

Activations Density 0.428%

references to items defined by their names or categories

No Comments

No Known Activations