INDEX

Explanations

proper nouns related to different entities such as people, places, and organizations

oai_token-act-pair · gpt-3.5-turbo

references to the name "Mills" and related entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

liest

-0.84

 fierce

-0.75

 stern

-0.72

itably

-0.69

wealth

-0.65

 Sovereign

-0.64

 Saddam

-0.64

ctory

-0.64

 handsome

-0.63

SPONSORED

-0.62

POSITIVE LOGITS

 Mills

1.35

pora

0.97

 mosqu

0.94

boro

0.93

hirt

0.87

mallow

0.87

 teasp

0.86

wich

0.84

Osw

0.84

psey

0.83

Activations Density 0.007%

proper nouns related to different entities such as people, places, and organizations

references to the name "Mills" and related entities

No Comments

No Known Activations