INDEX

Explanations

references to specific geographical locations and prominent names

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-pt-res/layer_2/width_16k/average_l0_67

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.2.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

tagHelperRunner

-0.53

 hjär

-0.47

 fubject

-0.41

ɚ

-0.41

 argint

-0.39

舍

-0.36

Colored

-0.36

continuous

-0.36

 colored

-0.36

Initialize

-0.36

POSITIVE LOGITS

 Cape

0.80

Cape

0.76

 capsule

0.72

captain

0.68

 captain

0.66

oredCriteria

0.65

 fond

0.65

Ace

0.65

 capsules

0.63

 cape

0.63

Activations Density 0.345%

references to specific geographical locations and prominent names

No Known Activations