INDEX

Explanations

text within brackets

oai_token-act-pair · gpt-3.5-turbo

square brackets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

wagen

-0.82

 occas

-0.77

 pens

-0.75

nesday

-0.73

edIn

-0.73

emouth

-0.70

 thous

-0.69

mable

-0.68

 rooft

-0.68

Orn

-0.68

POSITIVE LOGITS

?]

1.32

 edit

1.21

ËĪ

1.19

!]

1.15

sic

1.15

actionDate

1.09

Footnote

1.06

emphasis

1.03

laughs

1.02

%]

1.00

Activations Density 0.040%

text within brackets

square brackets

No Comments

No Known Activations