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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 overboard

-0.83

 legion

-0.78

 proble

-0.75

 overloaded

-0.65

 toget

-0.65

 exhausting

-0.62

 spaced

-0.61

 Legion

-0.60

 tank

-0.60

 underestimated

-0.59

POSITIVE LOGITS

enance

0.98

enegger

0.90

 marrow

0.80

brow

0.76

dll

0.74

ocrats

0.72

$$$$

0.72

 COURT

0.69

 Transcript

0.68

 à¨

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.