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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hetto

-0.76

library

-0.74

spring

-0.74

greg

-0.72

ummer

-0.71

books

-0.69

Times

-0.68

plot

-0.67

Season

-0.67

MAP

-0.67

POSITIVE LOGITS

 acknow

0.82

 mathemat

0.79

reditary

0.76

 indo

0.74

 sovere

0.74

 cumbers

0.70

 pronouns

0.70

 administr

0.69

 professionally

0.69

 unden

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.