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

annel

-0.72

 Incarn

-0.72

_.

-0.68

layer

-0.67

atha

-0.66

 defe

-0.66

 Cruel

-0.65

ilogy

-0.64

 Souls

-0.63

 Reborn

-0.63

POSITIVE LOGITS

uana

0.75

iman

0.72

ãĥĥãĤ¯

0.71

Pok

0.69

kaya

0.67

hari

0.61

 nipple

0.61

 unauthorized

0.60

iliar

0.60

ghai

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.