INDEX

Explanations

names of individuals, especially related to legal and executive positions

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names associated with individuals

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 enrollment

-0.97

 favorable

-0.92

 flavors

-0.91

 preempt

-0.91

Defense

-0.88

 honored

-0.85

Hispanic

-0.85

 favorably

-0.84

 unfavorable

-0.84

 maneuver

-0.83

POSITIVE LOGITS

 Davies

1.53

 Sutherland

1.24

 Rudd

1.23

 Pearce

1.23

 Wilkinson

1.21

 Atkinson

1.20

 Cardiff

1.19

 Robertson

1.19

 Fraser

1.19

 Turnbull

1.18

Activations Density 0.234%

names of individuals, especially related to legal and executive positions

proper nouns and names associated with individuals

No Comments

No Known Activations