INDEX

Explanations

names and emotional interactions between characters

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Tikang

-0.79

PreferredItem

-0.66

 snippetHide

-0.65

ագրություններ

-0.64

EndContext

-0.64

 الحره

-0.63



-0.63

 surla

-0.63

﹧

-0.62

الحياه

-0.61

POSITIVE LOGITS

 ebenfalls

0.33

 mặt

0.32

 Decken

0.31

เข้า

0.30

 vectorielle

0.30

asztal

0.29

 siihen

0.28

 intervenir

0.27

Quiénes

0.27

 psico

0.27

Activations Density 0.047%

names and emotional interactions between characters

No Comments

No Known Activations