INDEX

Explanations

phrases that convey implications or suggest meanings

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

Bord

-0.47

 asynchronously

-0.43

GenerationType

-0.42

 able

-0.39

 éte

-0.38

ガニック

-0.36

 Crockett

-0.36

 Bord

-0.35

为您

-0.35

POSITIVE LOGITS

 imply

0.85

 implied

0.79

imply

0.69

 IMPLIED

0.68

 implies

0.68

 implication

0.67

暗示

0.66

implies

0.63

ніципалі

0.63

 dụ

0.61

Activations Density 0.019%

phrases that convey implications or suggest meanings

No Comments

No Known Activations