INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Read

-0.70

 Possible

-0.70

 Shutterstock

-0.68

 READ

-0.68

reads

-0.68

 Outlook

-0.67

boats

-0.66

 Grind

-0.65

INF

-0.63

 Vessel

-0.61

POSITIVE LOGITS

ersive

0.84

TPP

0.81

aca

0.79

nesday

0.75

EStreamFrame

0.74

apon

0.74

adia

0.72

antam

0.72

ciating

0.72

rica

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.