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

ħĭ

-0.72

bold

-0.69

avis

-0.67

iblical

-0.67

-0.61

 worshipped

-0.61

vered

-0.60

 todd

-0.59

verage

-0.58

oyal

-0.58

POSITIVE LOGITS

atorium

0.77

atre

0.76

jen

0.73

arium

0.71

acan

0.66

ennis

0.64

len

0.64

 Seym

0.63

rican

0.63

 rapport

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.