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

isson

-0.82

 Rebell

-0.78

 Parables

-0.76

ipel

-0.72

 sqor

-0.72

actions

-0.70

choes

-0.70

rike

-0.69

ãĥĥãĥī

-0.69

emption

-0.68

POSITIVE LOGITS

INGTON

0.75

ptive

0.62

gel

0.61

ifice

0.61

aments

0.60

 locker

0.60

SSL

0.59

 drying

0.59

 ALWAYS

0.59

Russia

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.