INDEX

Explanations

references to the term "said" 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 itſelf

-0.63

 myſelf

-0.49

 tranſ

-0.49

 themſelves

-0.48

 ſte

-0.47

 ſta

-0.46

 ujednoznacz

-0.46

 himſelf

-0.43

vœ

-0.42

 deſt

-0.42

POSITIVE LOGITS

::$

0.66

 respective

0.57

NameInMap

0.56

تقاوى

0.53

 Pristupljeno

0.52

 respectivas

0.51

 aceste

0.51

揄

0.51

 aload

0.50

complexContent

0.49

Activations Density 0.017%

references to the term "said" in various contexts

No Comments

No Known Activations