INDEX

Explanations

references to conflicts or violent behaviors

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

 lecz

-0.60

 endast

-0.52

 comprendere

-0.52

 carattere

-0.49

 arbete

-0.48

 continuó

-0.46

 carácter

-0.45

 staden

-0.45

 volgen

-0.45

 stöd

-0.45

POSITIVE LOGITS

 stuff

0.97

 thingy

0.94

 dudes

0.86

 crappy

0.85

 shitty

0.79

 freakin

0.79

 STUFF

0.78

 shit

0.78

 بيها

0.78

 crap

0.77

Activations Density 0.767%

references to conflicts or violent behaviors

No Comments

No Known Activations