INDEX

Explanations

evidence supporting or contradicting claims

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_22/width_16k/average_l0_123

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

ahren

-0.47

 here

-0.42

 Jasper

-0.40

here

-0.40

Jasper

-0.39

amel

-0.38

月的

-0.35

 especially

-0.35

STEL

-0.35

월

-0.34

POSITIVE LOGITS

 Italijanski

0.65

 promesse

0.56

 promessa

0.52

elemField

0.52

+#+#

0.50

andExpect

0.50

VYMaps

0.50

 aikaa

0.49

findpost

0.48

FormTagHelper

0.48

Activations Density 0.697%

evidence supporting or contradicting claims

No Known Activations