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

ILA

-0.94

 DRAG

-0.84

 helicop

-0.79

paio

-0.76

olkien

-0.73

FontSize

-0.71

PDATE

-0.69

 enthusi

-0.68

querque

-0.68

 ACLU

-0.67

POSITIVE LOGITS

mouth

0.81

hof

0.69

pg

0.68

 Goods

0.67

ady

0.67

 Samar

0.63

psy

0.62

actor

0.61

ory

0.60

itary

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.