INDEX

Explanations

repeated emphasis on the word "very."

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-2b-pt-res/layer_2/width_16k/average_l0_141

Prompts (Dashboard)

36,864 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

Not in Any Lists

No Comments

Negative Logits

awtextra

-0.71

Londres

-0.64

 communis

-0.63

 salva

-0.62

 صلاح

-0.60

 ombre

-0.60

 maille

-0.59

ScopeManager

-0.59

 schließlich

-0.58

льше

-0.57

POSITIVE LOGITS

 very

3.35

very

2.88

Very

2.87

 Very

2.79

 VERY

2.61

VERY

2.52

 très

2.23

muy

2.23

 sehr

2.03

 очень

1.91

Activations Density 0.059%

repeated emphasis on the word "very."

No Comments

No Known Activations