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

ãĥ´ãĤ¡

-0.72

 Instit

-0.72

eus

-0.67

ortunately

-0.63

 Azerbaijan

-0.61

Ced

-0.61

ONG

-0.60

 Especially

-0.60

EFF

-0.60

 Ethiop

-0.59

POSITIVE LOGITS

iren

0.85

iker

0.73

rences

0.73

affer

0.73

leen

0.72

swer

0.69

igation

0.69

inez

0.68

hari

0.65

igate

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.