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

netflix

-0.73

 cylinders

-0.67

ands

-0.66

 hype

-0.65

hub

-0.64

 tide

-0.63

 Nova

-0.63

unes

-0.62

kr

-0.61

âĺħ

-0.61

POSITIVE LOGITS

iannopoulos

1.07

arten

0.77

 nomine

0.76

identally

0.71

=-=-=-=-

0.70

 Drawn

0.70

Iv

0.68

clus

0.66

 Cycl

0.64

 Latter

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.