INDEX

Explanations

words related to opinions or statements that oppose common beliefs or facts

oai_token-act-pair · gpt-3.5-turbo

words and phrases that indicate contrast or contradiction

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

izen

-0.68

 Helsinki

-0.65

eg

-0.65

orc

-0.64

Journal

-0.62

af

-0.62

EM

-0.62

 Booker

-0.61

 Tasman

-0.61

 Shanghai

-0.59

POSITIVE LOGITS

rary

1.48

rarily

0.91

sonian

0.89

etheless

0.87

cipled

0.87

tenance

0.85

issance

0.84

nesday

0.83

tesy

0.79

entimes

0.79

Activations Density 0.018%

words related to opinions or statements that oppose common beliefs or facts

words and phrases that indicate contrast or contradiction

No Comments

No Known Activations