INDEX

Explanations

phrases related to official titles or specialized terms

oai_token-act-pair · gpt-3.5-turbo

terms related to honorary titles and medical or legal authority

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

aque

-0.81

eday

-0.81

oken

-0.75

achu

-0.74

hooting

-0.69

sburgh

-0.69

shaw

-0.69

lords

-0.68

oos

-0.68

bley

-0.67

POSITIVE LOGITS

uthor

0.77

iv

0.70

lectic

0.69

CLA

0.66

ENSE

0.64

rique

0.63

ctuary

0.61

NER

0.61

osis

0.61

 Warrant

0.60

Activations Density 0.034%

phrases related to official titles or specialized terms

terms related to honorary titles and medical or legal authority

No Comments

No Known Activations