INDEX

Explanations

locations based on compass directions

oai_token-act-pair · gpt-3.5-turbo

directional terms related to location

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 moderator

-0.58

inished

-0.58

icer

-0.56

ulous

-0.56

aceae

-0.56

zan

-0.55

alien

-0.54

 stitches

-0.54

 datas

-0.54

POSITIVE LOGITS

ward

1.02

wards

1.00

west

0.94

 west

0.93

outheast

0.91

of

0.91

bound

0.88

 east

0.87

heast

0.82

 northeast

0.81

Activations Density 0.045%

locations based on compass directions

directional terms related to location

No Comments

No Known Activations