INDEX

Explanations

words related to geographical locations, particularly the United States

oai_token-act-pair · gpt-3.5-turbo

references to the United States

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

CAST

-0.72

esters

-0.72

utenberg

-0.70

Ĥİ

-0.65

èª

-0.64

gotten

-0.63

AUT

-0.61

sett

-0.61

arenthood

-0.58

 Pastebin

-0.58

POSITIVE LOGITS

 States

1.25

States

0.97

wide

0.85

 Governments

0.85

manship

0.84

 Parties

0.82

 STATES

0.80

 legislatures

0.76

 Alchemist

0.74

pring

0.72

Activations Density 0.009%

words related to geographical locations, particularly the United States

references to the United States

No Comments

No Known Activations