INDEX

Explanations

locations or geographical features

oai_token-act-pair · gpt-3.5-turbo

names of geographical locations and natural landmarks

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

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

 Kron

-0.73

canon

-0.68

 euros

-0.64

osexual

-0.64

aphael

-0.63

âĤ¬

-0.62

rael

-0.62

idious

-0.61

meric

-0.61

perse

-0.60

POSITIVE LOGITS

 Resort

1.41

 Cemetery

1.28

 Golf

1.19

 Recreation

1.19

 Apart

1.17

 Parkway

1.17

 Casino

1.13

 Reserv

1.13

 Lodge

1.12

 Massacre

1.09

Activations Density 0.213%

locations or geographical features

names of geographical locations and natural landmarks

No Comments

No Known Activations