INDEX

Explanations

the phrases related to listening and responding to someone's thoughts or opinions

oai_token-act-pair · gpt-3.5-turbo

expressions of communication and requests for attention or feedback

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ucket

-0.83

entin

-0.78

\/\/

-0.75

iant

-0.75

ccording

-0.71

thumbnails

-0.71

abase

-0.68

teness

-0.67

igsaw

-0.66

ritional

-0.66

POSITIVE LOGITS

 voices

1.17

 loud

1.14

 roar

1.12

 firsthand

1.05

 voice

1.03

 noises

1.00

 cries

0.99

 footsteps

0.96

 screams

0.94

 whispers

0.94

Activations Density 0.110%

the phrases related to listening and responding to someone's thoughts or opinions

expressions of communication and requests for attention or feedback

No Comments

No Known Activations