INDEX

Explanations

system-related error codes

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

󠁴

-0.82

 Signalez

-0.80

GEBURTSDATUM

-0.75

+#+

-0.75

fromnode

-0.74

RTSC

-0.72

Jeografia

-0.71

ſicht

-0.70

parsedMessage

-0.70

<unused43>

-0.70

POSITIVE LOGITS

0.35

↵↵

0.33

 including

0.28

Sure

0.28

 remember

0.28

 zweiter

0.28

remember

0.28

ve

0.27

Activations Density 0.004%

system-related error codes

No Comments

No Known Activations