INDEX

Explanations

directional terms denoting spatial position

oai_token-act-pair · gpt-3.5-turbo

spatial references and directions

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

ãĥīãĥ©

-0.91

BuyableInstoreAndOnline

-0.91

MAT

-0.71

LER

-0.70

 Fail

-0.69

olitics

-0.69

ITED

-0.68

 warr

-0.67

onis

-0.67

Rat

-0.65

POSITIVE LOGITS

 side

1.32

 flank

1.25

 corner

1.13

most

1.09

 edge

1.03

 hand

1.03

 hemisphere

0.99

 sidebar

0.98

 axis

0.94

 wing

0.91

Activations Density 0.045%

directional terms denoting spatial position

spatial references and directions

No Comments

No Known Activations