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

 Transcript

-0.74

phia

-0.71

ciating

-0.69

yer

-0.67

Jen

-0.64

llah

-0.63

rael

-0.63

 Weaver

-0.63

 Burton

-0.62

sil

-0.62

POSITIVE LOGITS

é¾įåĸļå£«

0.70

leans

0.61

interstitial

0.61

oriented

0.60

ãĤº

0.60

Downloadha

0.59

å¥³

0.59

 proportional

0.59

ighting

0.58

olester

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.