INDEX

Explanations

keywords related to locations or geographical entities

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

é¾įå¥ĳå£«

-0.71

milo

-0.67

blast

-0.66

scl

-0.66

 Chimera

-0.65

thumbnails

-0.64

 analy

-0.63

atical

-0.63

Accessory

-0.61

 jaws

-0.59

POSITIVE LOGITS

 Wales

1.31

wark

0.82

 Welsh

0.81

orkshire

0.74

 Orient

0.72

orks

0.71

ork

0.70

 Albion

0.70

ria

0.70

orthy

0.70

Activations Density 0.075%

keywords related to locations or geographical entities

No Comments

No Known Activations