INDEX

Explanations

expressions related to showing respect

oai_token-act-pair · gpt-3.5-turbo

references to the concept of respect

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 concoct

-0.85

otine

-0.75

 Lans

-0.72

 cram

-0.69

uggest

-0.68

 helicop

-0.68

 ferry

-0.67

ãĥĥãĤ¯

-0.67

uzzle

-0.66

 tongues

-0.66

POSITIVE LOGITS

ably

1.25

ability

1.11

fully

1.02

ibly

0.93

ledged

0.91

able

0.90

ful

0.88

amental

0.86

FUL

0.84

rity

0.81

Activations Density 0.026%

expressions related to showing respect

references to the concept of respect

No Comments

No Known Activations