INDEX

Explanations

phrases related to cover-ups and dishonesty

oai_token-act-pair · gpt-3.5-turbo

phrases related to cover-ups and concealment of information

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

agonist

-0.68

anian

-0.67

miah

-0.66

iu

-0.65

want

-0.64

vl

-0.62

 Rhythm

-0.62

joining

-0.62

 Yards

-0.61

scill

-0.61

POSITIVE LOGITS

 scandals

0.83

 loopholes

0.81

 Hebdo

0.77

 allegations

0.71

 Mysteries

0.70

 veil

0.69

 opacity

0.69

 false

0.68

 cloaked

0.67

 wrongdoing

0.67

Activations Density 0.061%

phrases related to cover-ups and dishonesty

phrases related to cover-ups and concealment of information

No Comments

No Known Activations