INDEX

Explanations

words related to communication methods, such as phone calls, emails, and communication forms

oai_token-act-pair · gpt-3.5-turbo

expressions of intimidation and aggression in communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

orthy

-0.80

 habitable

-0.73

 ancestor

-0.72

 extinct

-0.71

abase

-0.70

 venerable

-0.69

 cannabin

-0.66

 endowed

-0.65

iverse

-0.64

 gifted

-0.63

POSITIVE LOGITS

 briefings

1.25

 speeches

1.22

attacks

1.21

 warnings

1.20

 announcements

1.19

 visits

1.18

 chants

1.18

 inquiries

1.15

mails

1.15

 calls

1.14

Activations Density 0.514%

words related to communication methods, such as phone calls, emails, and communication forms

expressions of intimidation and aggression in communication

No Comments

No Known Activations