INDEX

Explanations

the prefix "Un-" followed by a word

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "Un," which likely relates to a specific organization or entity within the document

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

OPLE

-0.94

 Madden

-0.76

anwhile

-0.75

Peb

-0.71

 briefs

-0.71

Ö¼

-0.70

utical

-0.69

enegger

-0.68

hetti

-0.66

 tomat

-0.66

POSITIVE LOGITS

usual

1.14

cles

1.10

idad

1.07

ruly

1.05

affiliated

1.04

itarian

1.04

classified

1.04

agi

1.03

animous

1.00

rave

0.99

Activations Density 0.024%

the prefix "Un-" followed by a word

occurrences of the term "Un," which likely relates to a specific organization or entity within the document

No Comments

No Known Activations