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

Fed

-0.79

rists

-0.67

©¶æ

-0.67

reed

-0.65

west

-0.61

é»Ĵ

-0.61

»Ĵ

-0.60

rm

-0.60

EXT

-0.60

 Moff

-0.59

POSITIVE LOGITS

ahu

0.74

azaki

0.71

ulin

0.70

 Calais

0.70

ority

0.69

hani

0.66

hower

0.66

igree

0.64

itsch

0.63

itaire

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.