INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Transgender

-0.72

Asian

-0.70

 virgin

-0.67

rob

-0.67

mercial

-0.64

culus

-0.64

ournals

-0.64

settings

-0.64

period

-0.64

PATH

-0.64

POSITIVE LOGITS

andom

0.83

 hesitation

0.75

 needing

0.71

 disapproval

0.70

 embr

0.65

hement

0.65

Fn

0.65

 displeasure

0.63

 reluct

0.63

 fireplace

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.