INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

urai

-0.78

foundland

-0.74

iture

-0.73

nyder

-0.73

CRIP

-0.69

ilers

-0.67

ilts

-0.67

iller

-0.67

merce

-0.66

 Antar

-0.66

POSITIVE LOGITS

 Mysteries

0.74

 secrets

0.70

 secret

0.65

 Sisters

0.63

 Question

0.63

hess

0.63

 Twisted

0.60

 Epidem

0.60

 Seeds

0.60

 Crest

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.