INDEX

Explanations

words or phrases enclosed in quotation marks

oai_token-act-pair · gpt-3.5-turbo

quotation marks and words associated with direct speech or quotes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 killers

-0.81

 hypothal

-0.73

 adjud

-0.73

 buoy

-0.71

 tabloid

-0.70

 brittle

-0.70

 acidic

-0.70

HDL

-0.70

 overcrowd

-0.70

 devastated

-0.70

POSITIVE LOGITS

true

1.21

false

1.17

yes

1.11

ultimate

1.03

nothing

1.03

Alice

1.02

you

1.02

double

1.01

every

1.01

Yes

1.01

Activations Density 0.095%

words or phrases enclosed in quotation marks

quotation marks and words associated with direct speech or quotes

No Comments

No Known Activations