INDEX

Explanations

action words that indicate providing information or instructions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating guidance or offering suggestions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

natureconservancy

-0.75

selves

-0.62

 polic

-0.57

ansion

-0.56

alties

-0.56

wcsstore

-0.56

tesque

-0.54

artifacts

-0.54

 disadvant

-0.53

 territorial

-0.53

POSITIVE LOGITS

you

1.60

 readers

1.38

you

1.35

YOU

1.30

ya

1.14

You

1.09

yout

1.08

 reader

1.07

You

1.07

YOU

1.05

Activations Density 0.365%

action words that indicate providing information or instructions

phrases indicating guidance or offering suggestions

No Comments

No Known Activations