INDEX

Explanations

geographical locations such as landmarks, beaches, and specific points on a map

oai_token-act-pair · gpt-3.5-turbo

locations and geographical names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 asses

-0.75

 steroid

-0.72

Ð¾

-0.71

Ð´

-0.71

Ð

-0.65

razil

-0.64

 reproduce

-0.63

 acknow

-0.63

pes

-0.63

ublished

-0.62

POSITIVE LOGITS

 Cemetery

1.15

Rd

1.06

 Resort

1.06

 Apart

1.05

 Farms

1.04

 Mall

1.03

 Massacre

0.99

 Shopping

0.99

 Ranch

0.97

 Elementary

0.97

Activations Density 0.172%

geographical locations such as landmarks, beaches, and specific points on a map

locations and geographical names

No Comments

No Known Activations