INDEX

Explanations

mentions of locations or spatial relationships

oai_token-act-pair · gpt-3.5-turbo

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

DERR

-0.74

dor

-0.69

joice

-0.66

 voices

-0.64

 Votes

-0.63

 Laun

-0.62

assies

-0.60

marg

-0.60

testers

-0.59

makers

-0.59

POSITIVE LOGITS

Ħ¢

0.84

 ourselves

0.67

 conjunction

0.65

Matrix

0.65

MRI

0.64

alone

0.62

etime

0.62

omsday

0.59

 fruition

0.59

 peacefully

0.59

Activations Density 0.324%

mentions of locations or spatial relationships

No Comments

No Known Activations