INDEX

Explanations

phrases related to derogatory remarks

oai_token-act-pair · gpt-3.5-turbo

instances of the word "sn"

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

heid

-0.87

 limited

-0.70

maj

-0.64

 medium

-0.62

und

-0.61

 Cond

-0.60

EMENT

-0.60

 respectfully

-0.60

 belts

-0.59

Ind

-0.59

POSITIVE LOGITS

iping

1.44

uggle

1.43

ipe

1.41

ugg

1.41

appy

1.37

atches

1.35

atching

1.35

arling

1.34

ipes

1.33

ickers

1.33

Activations Density 0.033%

phrases related to derogatory remarks

instances of the word "sn"

No Comments

No Known Activations