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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

orsi

-0.92

olson

-0.88

-0.78

rongh

-0.77

ographer

-0.73

awar

-0.72

orians

-0.70

ographers

-0.70

ipple

-0.69

Church

-0.69

POSITIVE LOGITS

Rap

0.73

 uncond

0.72

 Continent

0.63

uls

0.61

piration

0.60

Gig

0.59

 guarant

0.59

 Intent

0.58

 Endurance

0.58

 Flor

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.