INDEX

Explanations

references to the term "rem" with varying degrees of emphasis

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "rem."

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

OPLE

-0.77

BOOK

-0.75

glers

-0.74

æł

-0.66

GY

-0.65

 Ducks

-0.65

gaard

-0.64

 Cultural

-0.64

 Petersen

-0.63

 Belt

-0.63

POSITIVE LOGITS

ittance

1.10

rem

1.00

ploy

1.00

 surpr

0.95

ilit

0.95

uner

0.94

itted

0.90

 satell

0.90

ittal

0.88

ixtape

0.86

Activations Density 0.006%

references to the term "rem" with varying degrees of emphasis

occurrences of the word "rem."

No Comments

No Known Activations