INDEX

Explanations

specific names of people

oai_token-act-pair · gpt-3.5-turbo

proper names of individuals

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

 initiation

-0.74

dj

-0.73

Neg

-0.72

dj

-0.70

 Mutant

-0.70

ombat

-0.70

andro

-0.69

âĶ

-0.67

 Jugg

-0.67

mx

-0.67

POSITIVE LOGITS

 Hers

3.51

 Seymour

1.93

 Berger

1.70

 Clement

1.61

 Wasserman

1.29

 Crawford

1.23

 Shepherd

1.21

 Reese

1.08

 Pulitzer

1.03

 Kers

1.02

Activations Density 0.026%

specific names of people

proper names of individuals

No Comments

No Known Activations