INDEX

Explanations

The main thing this neuron does is find nouns related to legal and financial issues

oai_token-act-pair · gpt-3.5-turbo

proper nouns and specific names of people or entities

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

 Nath

-0.92

ayne

-0.91

CLA

-0.88

 Shepard

-0.86

 Miranda

-0.80

ay

-0.74

anni

-0.74

pard

-0.74

aint

-0.72

 Holden

-0.72

POSITIVE LOGITS

ub

1.44

Tub

1.43

Lob

1.39

Dob

1.39

Dub

1.33

Kub

1.31

Hub

1.30

Dub

1.29

UB

1.28

rub

1.27

Activations Density 0.425%

The main thing this neuron does is find nouns related to legal and financial issues

proper nouns and specific names of people or entities

No Comments

No Known Activations