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

Ģ

-0.66

retty

-0.65

ccording

-0.65

 Coyotes

-0.63

iscons

-0.61

arov

-0.61

ounding

-0.61

ahon

-0.59

ovich

-0.59

pelled

-0.59

POSITIVE LOGITS

 Syri

0.85

reens

0.69

Prev

0.66

objects

0.65

erves

0.64

kered

0.64

abil

0.63

vir

0.61

geries

0.61

terms

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.