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

anse

-0.71

 crest

-0.67

 Kard

-0.63

ABE

-0.63

 Protective

-0.62

 Debor

-0.62

 pleasing

-0.61

getic

-0.60

 Reincarn

-0.60

 surfing

-0.59

POSITIVE LOGITS

pmwiki

0.93

EStreamFrame

0.79

stadt

0.77

ominated

0.72

thus

0.70

owa

0.69

chery

0.68

orem

0.68

psc

0.66

 Schedule

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.