INDEX

Explanations

the word "anything"

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "anything."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

irth

-0.69

nec

-0.69

 anonymity

-0.69

nee

-0.65

 Encyclopedia

-0.65

arb

-0.64

pa

-0.64

missions

-0.64

asio

-0.63

onym

-0.63

POSITIVE LOGITS

 else

1.59

Else

1.37

 resembling

1.13

 Else

1.09

 imaginable

1.02

THING

0.98

 remotely

0.94

else

0.90

 happens

0.82

 happening

0.81

Activations Density 0.033%

the word "anything"

references to the concept of "anything."

No Comments

No Known Activations