INDEX

Explanations

phrases related to editorial content or opinions

oai_token-act-pair · gpt-3.5-turbo

terms related to editorial content in publications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

llan

-0.74

aday

-0.72

pots

-0.71

omething

-0.71

agne

-0.70

cles

-0.69

rises

-0.68

ptives

-0.67

ulia

-0.67

imilar

-0.66

POSITIVE LOGITS

 editor

1.00

ized

1.00

izing

0.99

 editorial

0.99

 Editorial

0.94

 board

0.93

IZE

0.93

 cartoons

0.91

izes

0.88

zeb

0.86

Activations Density 0.030%

phrases related to editorial content or opinions

terms related to editorial content in publications

No Comments

No Known Activations