INDEX

Explanations

cryptic codes or labels associated with various entities or concepts

oai_token-act-pair · gpt-3.5-turbo

proper nouns or names in the text

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

istical

-0.73

 abstraction

-0.68

ial

-0.64

ened

-0.64

rade

-0.64

 totality

-0.63

 inev

-0.60

 hiber

-0.59

 movement

-0.59

 econom

-0.59

POSITIVE LOGITS

ensis

0.80

gmail

0.78

kens

0.73

alias

0.73

Downloadha

0.70

 Malays

0.68

 teasp

0.68

 tradem

0.67

tein

0.65

byss

0.65

Activations Density 0.752%

cryptic codes or labels associated with various entities or concepts

proper nouns or names in the text

No Comments

No Known Activations