INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 nutshell

-0.74

monary

-0.67

 liner

-0.66

 scen

-0.64

 externalToEVAOnly

-0.64

 activ

-0.64

 temples

-0.63

 Seasons

-0.63

 brow

-0.63

 cardboard

-0.61

POSITIVE LOGITS

ECD

0.76

odic

0.71

edes

0.69

olicited

0.69

od

0.69

bil

0.68

acl

0.68

shock

0.68

 Centauri

0.67

ELS

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.