INDEX

Explanations

words and phrases related to comfort and support

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

 vectoriales

-0.41

bootstrapcdn

-0.39

 bendera

-0.38

 vektör

-0.37

beelden

-0.35

์ตูน

-0.35

 jorden

-0.35

 voordeel

-0.35

 espejos

-0.35

 crí

-0.34

POSITIVE LOGITS

 comfort

0.84

comfort

0.82

Comfort

0.77

 COMFORT

0.74

ChildScrollView

0.71

 Comfort

0.70

comfortable

0.70

 comfortable

0.69

 comfy

0.65

 Comfortable

0.64

Activations Density 0.007%

words and phrases related to comfort and support

No Comments

No Known Activations