INDEX

Explanations

phrases that include the word "emblazoned"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "Amazon."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 suspension

-0.68

 Stage

-0.67

 release

-0.66

 significant

-0.64

wise

-0.62

 Cycle

-0.62

 breath

-0.60

 Kerry

-0.60

 recovery

-0.60

 improperly

-0.60

POSITIVE LOGITS

azon

5.27

amazon

1.32

ozo

1.10

AZ

1.07

querque

1.01

az

1.01

Amazon

0.97

aiman

0.96

essage

0.96

azo

0.95

Activations Density 0.014%

phrases that include the word "emblazoned"

occurrences of the term "Amazon."

No Comments

No Known Activations