INDEX

Explanations

phrases related to broadcasts or communications

oai_token-act-pair · gpt-3.5-turbo

instances of the abbreviation "RA" or words starting with "RA"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 substitutes

-0.68

 pupils

-0.64

 interactions

-0.64

Son

-0.63

 Incarn

-0.63

 Blake

-0.62

wit

-0.62

 Spicer

-0.62

 font

-0.62

Cop

-0.61

POSITIVE LOGITS

RA

4.20

RAG

1.76

ra

1.74

ARA

1.65

RA

1.50

RAY

1.46

RI

1.44

RE

1.42

rain

1.40

SA

1.40

Activations Density 0.017%

phrases related to broadcasts or communications

instances of the abbreviation "RA" or words starting with "RA"

No Comments

No Known Activations