INDEX

Explanations

proper nouns consisting of two words

oai_token-act-pair · gpt-3.5-turbo

references to the name and variations of the name "Se" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

CPC

-0.87

 diplom

-0.70

 wire

-0.69

etheless

-0.69

 bulk

-0.67

AVG

-0.65

 Malaysian

-0.64

vic

-0.64

 surve

-0.64

£ı

-0.64

POSITIVE LOGITS

ldom

1.18

vere

1.16

eker

1.16

greg

0.99

ethe

0.97

emed

0.97

isse

0.97

gments

0.95

infeld

0.95

gment

0.94

Activations Density 0.009%

proper nouns consisting of two words

references to the name and variations of the name "Se" in various contexts

No Comments

No Known Activations