INDEX

Explanations

instances of the word "and" in conjunction with topics related to health and risk factors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.17.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

tiftung

-0.51

Mang

-0.51

 Progressive

-0.49

 miele

-0.47

Progressive

-0.45

ורי

-0.45

 Beecher

-0.45

 glossy

-0.44

 drinks

-0.44

Où

-0.44

POSITIVE LOGITS

 CreateTagHelper

0.81

 ویکی‌پدیا

0.79

TagMode

0.75

 description

0.67

IntoConstraints

0.63

Pautan

0.62

tvguidetime

0.61

]){

0.60

setof

0.59

]),

0.59

Activations Density 0.097%

instances of the word "and" in conjunction with topics related to health and risk factors

No Comments

No Known Activations