INDEX

Explanations

references to lies and deception

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lie" and its variations in different contexts

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

arta

-0.78

orr

-0.62

illion

-0.61

lished

-0.61

 weaving

-0.60

Attempts

-0.60

-0.59

=-=-=-=-=-=-=-=-

-0.58

 liking

-0.57

andal

-0.57

POSITIVE LOGITS

lies

1.22

utenant

0.92

lie

0.80

poons

0.75

creen

0.74

 showc

0.71

layer

0.69

chool

0.69

HF

0.67

ogyn

0.67

Activations Density 0.003%

references to lies and deception

instances of the word "lie" and its variations in different contexts

No Comments

No Known Activations