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

WAYS

-0.68

Mand

-0.67

geist

-0.64

 Meaning

-0.64

cong

-0.63

wk

-0.62

OPLE

-0.61

HAEL

-0.61

 Stranger

-0.61

 Aware

-0.61

POSITIVE LOGITS

 picnic

0.86

otes

0.75

}}}

0.74

alore

0.73

otos

0.70

 reunion

0.70

 peanuts

0.70

 ceremonies

0.69

fman

0.69

 Pegasus

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.