INDEX

Explanations

phrases involving an unspecified group or entities in a negative context

oai_token-act-pair · gpt-3.5-turbo

references to the word "some"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ocene

-0.76

tainment

-0.69

enance

-0.65

ettes

-0.64

gon

-0.64

sburgh

-0.63

 borne

-0.63

raid

-0.63

uckle

-0.62

meet

-0.62

POSITIVE LOGITS

ones

1.03

place

1.03

 semblance

0.93

body

0.93

how

0.92

 unspecified

0.82

HOW

0.73

 mornings

0.70

 indist

0.70

 degree

0.69

Activations Density 0.117%

phrases involving an unspecified group or entities in a negative context

references to the word "some"

No Comments

No Known Activations