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

culosis

-0.86

zl

-0.80

auld

-0.75

 administr

-0.70

 endeav

-0.67

ilst

-0.67

 tenancy

-0.67

aned

-0.67

 receipt

-0.66

ipeg

-0.65

POSITIVE LOGITS

cue

0.80

ļé

0.76

tta

0.75

¶æ

0.71

 Bale

0.69

aeper

0.69

 Piercing

0.69

mp

0.68

venge

0.68

ãĤ®

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.