INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Cthulhu

-0.75

 Panel

-0.74

Gun

-0.70

Gun

-0.69

 panel

-0.69

 Hermes

-0.67

 dstg

-0.67

Por

-0.64

gear

-0.64

USA

-0.63

POSITIVE LOGITS

emis

0.73

 vulner

0.72

{*

0.71

 Ethiop

0.71

olid

0.70

ugg

0.69

 optimizations

0.69

animous

0.69

 fruitful

0.69

orph

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.