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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Bowl

-0.73

pants

-0.69

ioch

-0.68

SPONSORED

-0.66

 Caller

-0.65

 Administrator

-0.64

 Amendments

-0.64

 Leader

-0.63

iologist

-0.62

about

-0.62

POSITIVE LOGITS

 limp

0.79

 negatives

0.77

 shine

0.76

EStreamFrame

0.74

 abund

0.71

ismo

0.71

 slightest

0.71

 favourable

0.69

Drag

0.67

 grace

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.