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

raints

-0.82

 webs

-0.75

ahn

-0.72

bid

-0.72

aken

-0.71

 ventures

-0.69

ee

-0.68

 undertake

-0.67

oulos

-0.67

rane

-0.66

POSITIVE LOGITS

 Centauri

0.88

NPR

0.88

GI

0.82

PLA

0.82

UTERS

0.78

Americ

0.77

ãĥĩãĤ£

0.76

NG

0.76

uminati

0.75

WI

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.