INDEX

Explanations

negations indicating a lack of desire or intent

oai_token-act-pair · gpt-3.5-turbo

negative contractions related to lack of certainty or willingness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

accompan

-0.69

senal

-0.69

 departures

-0.67

 Reborn

-0.66

 Mayhem

-0.66

mere

-0.66

PDATE

-0.62

 subur

-0.62

 Palest

-0.61

CLR

-0.61

POSITIVE LOGITS

 want

1.40

want

1.34

 realise

1.32

 know

1.31

 wanna

1.27

 realize

1.22

 understand

1.21

 intend

1.20

 belong

1.18

 WANT

1.15

Activations Density 0.169%

negations indicating a lack of desire or intent

negative contractions related to lack of certainty or willingness

No Comments

No Known Activations