INDEX

Explanations

locations or cities, specifically in California, as indicated by a high activation for the term 'Calif'

oai_token-act-pair · gpt-3.5-turbo

references to California

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

BOOK

-0.78

buster

-0.74

 Crimean

-0.73

 favour

-0.72

schild

-0.68

TPPStreamerBot

-0.66

balance

-0.66

 denomin

-0.66

 cater

-0.64

theless

-0.63

POSITIVE LOGITS

 Calif

1.07

.,

0.88

eno

0.88

ano

0.87

ubs

0.83

illo

0.80

Oro

0.78

uti

0.78

uce

0.77

isco

0.76

Activations Density 0.011%

locations or cities, specifically in California, as indicated by a high activation for the term 'Calif'

references to California

No Comments

No Known Activations