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

oul

-0.73

 guiActiveUnfocused

-0.69

 Islanders

-0.66

 Virus

-0.65

 Mets

-0.63

 Airport

-0.63

 height

-0.62

 relapse

-0.61

href

-0.60

 awake

-0.59

POSITIVE LOGITS

 Corinth

0.77

-----------

0.77

DON

0.71

gren

0.69

 democracies

0.68

TEXT

0.67

cro

0.66

 municip

0.65

words

0.65

________________________________________________________________

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.