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

Bey

-0.77

=>

-0.72

 WATCHED

-0.71

 Quant

-0.71

 Rosenberg

-0.70

 determin

-0.69

SPD

-0.68

Pig

-0.66

?]

-0.66

 proport

-0.65

POSITIVE LOGITS

akeru

0.88

ctica

0.80

tumblr

0.75

antis

0.74

apolis

0.71

enta

0.69

ente

0.68

walker

0.67

ucha

0.67

umblr

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.