INDEX

Explanations

phrases related to public announcements or statements made by authorities

oai_token-act-pair · gpt-3.5-turbo

verbs indicating acknowledgment or reporting of information

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

scribe

-0.75

enic

-0.71

 ILCS

-0.67

ene

-0.67

illions

-0.63

pee

-0.62

except

-0.61

aho

-0.61

allery

-0.57

mine

-0.56

POSITIVE LOGITS

 yesterday

1.42

 earlier

1.34

 Tuesday

1.31

 Wednesday

1.30

 Monday

1.28

 Thursday

1.27

 last

1.24

 Friday

1.20

 Saturday

1.08

 Sunday

1.06

Activations Density 0.321%

phrases related to public announcements or statements made by authorities

verbs indicating acknowledgment or reporting of information

No Comments

No Known Activations