INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

enez

-0.72

etheless

-0.69

POL

-0.67

anton

-0.67

stood

-0.66

rir

-0.65

rique

-0.63

 snap

-0.61

RW

-0.61

ãĥķãĤ¡

-0.61

POSITIVE LOGITS

sup

0.76

uscript

0.75

edition

0.73

 Krug

0.68

bats

0.67

band

0.66

bands

0.65

TIT

0.64

vari

0.63

sized

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.