INDEX

Explanations

phrases related to a specific person or character named "Fi"

oai_token-act-pair · gpt-3.5-turbo

references to the character "Fi" or similar variations in context

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

eers

-0.79

manship

-0.71

 Kessler

-0.71

stage

-0.69

akeru

-0.66

 Spartan

-0.64

 Jenner

-0.63

 clipboard

-0.63

 Hyper

-0.63

PID

-0.63

POSITIVE LOGITS

Fi

1.25

Fi

1.04

ennes

1.03

anca

0.94

endish

0.91

oti

0.89

plin

0.88

endi

0.88

fi

0.85

ande

0.83

Activations Density 0.008%

phrases related to a specific person or character named "Fi"

references to the character "Fi" or similar variations in context

No Comments

No Known Activations