INDEX

Explanations

text that is written down in various contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "written" indicating authorship or the act of writing

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

abe

-0.87

nel

-0.68

xon

-0.68

Magn

-0.66

abol

-0.66

abil

-0.65

hawks

-0.65

none

-0.65

olar

-0.65

 Shinra

-0.65

POSITIVE LOGITS

 written

1.17

 penned

0.91

written

0.90

Written

0.80

 aloud

0.78

lishing

0.77

 describ

0.76

 writing

0.75

writing

0.75

 snipp

0.75

Activations Density 0.013%

text that is written down in various contexts

instances of the word "written" indicating authorship or the act of writing

No Comments

No Known Activations