INDEX

Explanations

references to a specific location, which is "Tucson"

oai_token-act-pair · gpt-3.5-turbo

references to specific locations, particularly Tucson and Mesa

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

oeuv

-0.94

urst

-0.89

lass

-0.87

uther

-0.86

hops

-0.81

onut

-0.78

phas

-0.75

ructose

-0.73

ischer

-0.73

adder

-0.72

POSITIVE LOGITS

 Coyotes

1.08

AZ

0.91

 Tucson

0.85

 Arizona

0.75

 Mesa

0.75

ione

0.73

 Granger

0.73

 Cardinals

0.73

 Suns

0.72

eta

0.71

Activations Density 0.020%

references to a specific location, which is "Tucson"

references to specific locations, particularly Tucson and Mesa

No Known Activations