INDEX

Explanations

conjunctions and co-occurrence of phrases in lists

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.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



-0.73

Autoritní

-0.68

Personendaten

-0.66

 disambiguazione

-0.65

ContentLoaded

-0.63

 cherchés

-0.61

ItemBackground

-0.59

といけない

-0.58

 հղումներ

-0.57

 betweenstory

-0.57

POSITIVE LOGITS

etc

1.24

 even

1.03

 persino

0.99

 sebagainya

0.97

etc

0.97

 vieles

0.96

 lastly

0.91

 finally

0.90

 sogar

0.87

 zelfs

0.85

Activations Density 0.366%

conjunctions and co-occurrence of phrases in lists

No Comments

No Known Activations