INDEX

Explanations

the occurrence of the word "worked" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-mlp-16k

Configuration

google/gemma-scope-2b-pt-mlp/layer_1/width_16k/average_l0_105

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.hook_mlp_out

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

 worked

-2.22

 Worked

-1.97

worked

-1.88

Worked

-1.73

 Working

-1.20

 working

-1.18

Working

-1.17

working

-1.16

WORKING

-1.12

 trabajó

-1.09

POSITIVE LOGITS

 فريبيس

0.67

NgModule

0.59

 Achter

0.57

 NgModule

0.55

urtstag

0.54

endphp

0.54

similiano

0.53

లాలు

0.52

 bandou

0.51

nestjs

0.50

Activations Density 0.121%

the occurrence of the word "worked" in various contexts

No Comments

No Known Activations