INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ho

-0.71

RFC

-0.67

SG

-0.65

furt

-0.64

ti

-0.62

Pra

-0.62

 Alley

-0.61

 Â«

-0.61

holm

-0.60

Tus

-0.59

POSITIVE LOGITS

merce

0.89

 tremend

0.81

 paycheck

0.68

theless

0.65

ashtra

0.65

eatures

0.63

olars

0.63

iety

0.63

branded

0.63

 Cumm

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.