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

MSM

-0.81

ancial

-0.79

tein

-0.72

 Lect

-0.70

 Murd

-0.70

itbart

-0.65

Fib

-0.63

chet

-0.63

 Pratt

-0.63

igion

-0.62

POSITIVE LOGITS

Tokens

0.98

hover

0.89

 escort

0.75

wreck

0.74

 Seym

0.72

 Moroc

0.71

 Jagu

0.68

Reward

0.68

agascar

0.67

dor

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.