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

istani

-0.74

ership

-0.72

HAM

-0.71

edom

-0.67

 è£ıè

-0.67

 WARN

-0.62

WHERE

-0.62

ãĢĲ

-0.61

ãģ®å

-0.60

 hydra

-0.59

POSITIVE LOGITS

aster

0.83

ukemia

0.75

ogene

0.74

VPN

0.63

hift

0.63

Pac

0.63

chuk

0.63

 psychedel

0.62

 Obst

0.61

 Travels

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.