INDEX

Explanations

phrases related to online commenting and feedback

oai_token-act-pair · gpt-3.5-turbo

references to user-generated content and commentary sections

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

ãĤ¼ãĤ¦ãĤ¹

-0.82

imeters

-0.74

natureconservancy

-0.73

ModLoader

-0.72

atown

-0.71

apon

-0.71

ãĥ¼ãĥ«

-0.67

imeter

-0.65

wagen

-0.64

uni

-0.63

POSITIVE LOGITS

 Scholars

0.66

 BOOK

0.64

pse

0.63

>:

0.63

 Rowling

0.62

 Kers

0.62

 Reviews

0.60

 omission

0.60

leases

0.60

reports

0.59

Activations Density 1.012%

phrases related to online commenting and feedback

references to user-generated content and commentary sections

No Comments

No Known Activations