INDEX

Explanations

references to significant or noteworthy subjects and details

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

New Auto-Interp

Configuration

google/gemma-scope-2b-pt-res/layer_12/width_32k/average_l0_76

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

32,768

Data Type

float32

Hook Name

blocks.12.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

 metropolitana

-0.72

ers

-0.66

 syno

-0.63

 flèche

-0.63

 Fleck

-0.63

 Phyto

-0.63

ade

-0.62

 giudi

-0.61

 gynhyrchwyd

-0.60

 viewType

-0.60

POSITIVE LOGITS

 important

1.51

 Important

1.49

Important

1.41

 importance

1.38

important

1.36

 Importance

1.26

importance

1.25

Importance

1.24

IMPORTANT

1.23

 IMPORTANT

1.19

Activations Density 0.049%

references to significant or noteworthy subjects and details

No Comments

No Known Activations