INDEX

Explanations

phrases indicating outcomes or conclusions in various contexts

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

 rising

-0.43

 ſtre

-0.42

 ſol

-0.38

 Rising

-0.37

 ſch

-0.37

 tendenza

-0.37

тини

-0.37

บุ

-0.36

Rising

-0.36

 poitrine

-0.36

POSITIVE LOGITS

結局

0.64

GenerationType

0.63

 uiteindelijk

0.63

bootstrapcdn

0.61

 eventually

0.59

 finalement

0.59

eventually

0.56

 Eventually

0.56

webElement

0.56

 eventual

0.55

Activations Density 0.037%

phrases indicating outcomes or conclusions in various contexts

No Comments

No Known Activations