INDEX

Explanations

No Explanations Found

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

liest

-0.69

anwhile

-0.67

axter

-0.65

apest

-0.64

 caution

-0.63

ourge

-0.63

oths

-0.62

pection

-0.62

 thirds

-0.62

holding

-0.62

POSITIVE LOGITS

ECD

0.76

Ñĭ

0.75

 Accessed

0.74

 Work

0.73

Work

0.69

ESA

0.64

HT

0.64

edo

0.63

 ----------------------------------------------------------------

0.62

Stra

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.