INDEX

Explanations

instances of the word "concealing"

oai_token-act-pair · gpt-3.5-turbo

words related to concealment and deception

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

fleet

-0.67

 conductor

-0.66

 GOODMAN

-0.64

 Articles

-0.64

 Hits

-0.64

ucky

-0.63

 GEAR

-0.62

 loans

-0.60

hearted

-0.59

 Bucks

-0.59

POSITIVE LOGITS

ivably

1.26

iving

1.12

ences

0.90

aling

0.88

veland

0.87

ibility

0.86

uded

0.85

ibly

0.84

rence

0.83

uding

0.82

Activations Density 0.015%

instances of the word "concealing"

words related to concealment and deception

No Comments

No Known Activations