INDEX

Explanations

words related to style or stylistic elements

oai_token-act-pair · gpt-3.5-turbo

references to style and stylization

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

BSD

-0.76

MQ

-0.75

æĸ¹

-0.74

 Trust

-0.74

ETF

-0.72

 Labrador

-0.69

 Huff

-0.68

 Fortress

-0.67

FACE

-0.67

 Accuracy

-0.65

POSITIVE LOGITS

 styl

1.16

istically

1.05

ological

0.98

isphere

0.96

istics

0.95

istic

0.94

ographs

0.94

ifact

0.91

ologically

0.91

icians

0.90

Activations Density 0.011%

words related to style or stylistic elements

references to style and stylization

No Comments

No Known Activations