INDEX

Explanations

words related to physical or abstract spaces

oai_token-act-pair · gpt-3.5-turbo

mentions of physical space or spatial concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 vengeance

-0.76

venge

-0.74

usted

-0.73

 grades

-0.70

inger

-0.69

otine

-0.66

ctive

-0.66

iamond

-0.65

igers

-0.65

 Simpson

-0.64

POSITIVE LOGITS

 Layout

0.89

 occupied

0.88

 spaces

0.85

 vacated

0.85

 space

0.83

 shuttle

0.80

 occupancy

0.80

 Spaces

0.78

dayName

0.77

bars

0.75

Activations Density 0.039%

words related to physical or abstract spaces

mentions of physical space or spatial concepts

No Comments

No Known Activations