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

 Turing

-0.67

abulary

-0.64

 Scouting

-0.64

 moon

-0.60

 moot

-0.59

entially

-0.58

glers

-0.57

ertodd

-0.57

ned

-0.56

 Scout

-0.56

POSITIVE LOGITS

RAW

0.87

Accessory

0.78

outer

0.73

 millenn

0.67

roth

0.66

ikuman

0.65

arian

0.63

atche

0.63

ti

0.62

CONCLUS

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.