INDEX

Explanations

elements related to logical expressions and decision matrices

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

 Administrativna

-0.89

+#+#

-0.69

fromnode

-0.66

 betweenstory

-0.62

 Italijanski

-0.59

LookAnd

-0.59

 <=",

-0.56

windowFixed

-0.54

Хьажоргаш

-0.54

stdc

-0.53

POSITIVE LOGITS

Referanser

0.41

ultados

0.38

 shrinking

0.35

 toepassing

0.34

mr

0.34

 Garg

0.34

 appreciating

0.34

 tył

0.33

 offender

0.33

DllImport

0.33

Activations Density 0.001%

elements related to logical expressions and decision matrices

No Comments

No Known Activations