INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

thing

-0.69

ãĥ¼ãĥ³

-0.69

Georg

-0.68

byter

-0.68

acter

-0.68

genic

-0.67

terson

-0.66

aceous

-0.65

imeters

-0.64

ula

-0.64

POSITIVE LOGITS

é¾įå¥ĳå£«

0.75

etsk

0.67

 Berger

0.62

 WARN

0.61

 spar

0.61

 Spoiler

0.60

 downt

0.60

 handic

0.58

heses

0.58

ply

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.