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

asm

-0.68

zzle

-0.68

 Audi

-0.66

 Sheep

-0.64

abba

-0.64

 Mole

-0.62

MM

-0.61

HIT

-0.61

vier

-0.60

 Xiao

-0.60

POSITIVE LOGITS

ItemImage

0.88

Adds

0.81

geist

0.69

entimes

0.69

iculture

0.69

gra

0.68

 negatives

0.68

ographies

0.65

isSpecialOrderable

0.65

ega

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.