INDEX

Explanations

a specific name or token 'Ste' in the text

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of a specific name or character, particularly "Ste" and "Stella"

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

 taxpayer

-0.79

 majors

-0.75

 nursery

-0.70

 indent

-0.70

 unemployed

-0.69

 diapers

-0.69

 monkeys

-0.69

 primates

-0.69

iv

-0.68

 taxpayers

-0.68

POSITIVE LOGITS

Ste

3.80

Ste

2.00

ste

1.79

STE

1.71

STE

1.59

 Steele

1.59

 Stein

1.54

 Stella

1.36

ste

1.35

 Stef

1.35

Activations Density 0.015%

a specific name or token 'Ste' in the text

repeated mentions of a specific name or character, particularly "Ste" and "Stella"

No Comments

No Known Activations