INDEX

Explanations

terms related to construction or creation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "built" to highlight construction or creation

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

 Unsure

-0.71

ICE

-0.70

mun

-0.66

apple

-0.66

ican

-0.66

FFFF

-0.64

 Shooter

-0.63

Published

-0.63

ICAN

-0.62

Universal

-0.61

POSITIVE LOGITS

built

0.86

 constructed

0.81

packs

0.80

urally

0.76

 bridges

0.75

 decks

0.71

jriwal

0.70

itect

0.69

smanship

0.68

raper

0.68

Activations Density 0.022%

terms related to construction or creation

instances of the word "built" to highlight construction or creation

No Comments

No Known Activations