INDEX

Explanations

words related to specific historical eras, particularly the "New Age"

oai_token-act-pair · gpt-3.5-turbo

references to different "Ages" or historical periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

posal

-0.66

raq

-0.65

 versa

-0.65

ãĥķãĤ©

-0.65

CU

-0.64

header

-0.62

 bear

-0.62

 specials

-0.61

cho

-0.61

 payload

-0.61

POSITIVE LOGITS

Age

4.09

Age

3.20

age

2.48

 Ages

2.07

 ages

1.89

Era

1.64

 aged

1.52

 Aging

1.36

 Younger

1.30

 ageing

1.23

Activations Density 0.012%

words related to specific historical eras, particularly the "New Age"

references to different "Ages" or historical periods

No Comments

No Known Activations