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

 modesty

-0.73

å¦

-0.72

UID

-0.65

ESV

-0.64

Recomm

-0.63

æĺ¯

-0.62

 attorneys

-0.61

 privacy

-0.61

 reconc

-0.60

 eyebrows

-0.60

POSITIVE LOGITS

aths

0.79

erial

0.73

anwhile

0.71

plings

0.70

ensibly

0.68

 Shooting

0.67

verts

0.66

vernight

0.66

 Psycho

0.66

kt

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.