INDEX

Explanations

phrases expressing disagreement or disbelief

oai_token-act-pair · gpt-3.5-turbo

phrases indicating criticism or disapproval

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

 Xperia

-0.94

SAT

-0.81

 pastry

-0.79

 tsun

-0.75

 climbers

-0.75

 unpre

-0.75

 Armenian

-0.73

astery

-0.72

 cher

-0.72

 dentist

-0.72

POSITIVE LOGITS

Johnson

2.60

 Johnson

2.53

john

1.40

 Weld

1.29

LB

1.21

 john

1.09

 Lyndon

1.09

 Owens

1.08

 Johns

1.06

 Johnston

1.05

Activations Density 0.148%

phrases expressing disagreement or disbelief

phrases indicating criticism or disapproval

No Comments

No Known Activations