INDEX

Explanations

mentions of opinions or statements being expressed verbally by individuals

oai_token-act-pair · gpt-3.5-turbo

words related to expressing opinions or concerns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

adr

-0.75

ney

-0.70

neys

-0.68

ensation

-0.68

leans

-0.67

igm

-0.66

âĢ¢âĢ¢âĢ¢âĢ¢

-0.66

ionage

-0.64

eh

-0.64

dry

-0.63

POSITIVE LOGITS

 voiced

1.34

 voicing

1.21

 voices

1.00

 voice

0.96

 vocal

0.94

 voic

0.94

 chords

0.89

unci

0.88

 artic

0.86

 conson

0.83

Activations Density 0.005%

mentions of opinions or statements being expressed verbally by individuals

words related to expressing opinions or concerns

No Comments

No Known Activations