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

Store

-0.74

quit

-0.73

tips

-0.73

chat

-0.71

desktop

-0.67

paste

-0.66

Admin

-0.65

xes

-0.65

ĵĺ

-0.65

icter

-0.64

POSITIVE LOGITS

 known

0.92

 given

0.72

 handic

0.71

 African

0.69

 costing

0.67

itsch

0.66

scape

0.66

 cost

0.63

 gorilla

0.62

 conced

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.