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

 Yourself

-0.68

isk

-0.67

 yourselves

-0.66

ction

-0.65

iam

-0.65

 Petition

-0.61

 guarantee

-0.60

Track

-0.60

rys

-0.59

riage

-0.59

POSITIVE LOGITS

utenberg

0.74

 compounded

0.71

 Mong

0.68

 Edited

0.66

 edited

0.64

gnu

0.64

uddin

0.64

 inhabited

0.62

 Mahm

0.61

âĹ¼

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.