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

theless

-0.81

htaking

-0.73

apy

-0.71

 Buster

-0.68

heon

-0.68

 Dian

-0.67

 Awakens

-0.65

ulhu

-0.64

 srfAttach

-0.63

 remem

-0.62

POSITIVE LOGITS

Introduced

0.74

 better

0.73

better

0.72

 faster

0.71

IST

0.71

 fluent

0.66

ECD

0.65

FP

0.64

 smarter

0.64

Abstract

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.