INDEX

Explanations

instances of the word "answer" and its variations, focusing on questions and responses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 13-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_13/width_65k/average_l0_75

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.13.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

 للمعارف

-0.52

 насељу

-0.50

{%

-0.49

>`;

-0.49

BagLayout

-0.48

 entro

-0.47

zkod

-0.46

erunner

-0.46

 enterprise

-0.46

¡¡

-0.46

POSITIVE LOGITS

 Answering

1.36

 answering

1.31

 answered

1.23

 questions

1.21

 Answered

1.09

answered

1.07

ANSWER

1.00

 question

1.00

 unanswered

0.99

 tanong

0.97

Activations Density 0.022%

instances of the word "answer" and its variations, focusing on questions and responses

No Comments

No Known Activations