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

 Coast

-0.71

 Jord

-0.70

^^^^

-0.66

Dru

-0.63

Ku

-0.63

 Cities

-0.63

 Thor

-0.62

Kal

-0.62

 Pirates

-0.62

landers

-0.60

POSITIVE LOGITS

 adolesc

0.78

 palate

0.76

amph

0.67

ãĤī

0.66

verett

0.66

 carbohyd

0.65

zag

0.65

ãĤĵ

0.65

aste

0.64

 suspic

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.