INDEX

Explanations

words related to negative attributes or actions

oai_token-act-pair · gpt-3.5-turbo

terms associated with negative behaviors or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 GOODMAN

-0.85

quart

-0.73

 Hemp

-0.72

avez

-0.71

anchester

-0.70

ĸļ

-0.70

arate

-0.69

 Occupations

-0.68

bleacher

-0.67

ĺħ

-0.65

POSITIVE LOGITS

cffff

0.93

ness

0.87

nesses

0.85

 foolish

0.82

ly

0.80

é¾įå

0.72

itous

0.71

cia

0.70

 arrogance

0.70

 modesty

0.69

Activations Density 0.014%

words related to negative attributes or actions

terms associated with negative behaviors or conditions

No Comments

No Known Activations