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

 regress

-0.75

 belie

-0.67

 culture

-0.66

uckland

-0.66

acular

-0.65

 lure

-0.64

paio

-0.64

 weave

-0.63

 raft

-0.63

luc

-0.62

POSITIVE LOGITS

vous

0.75

Soy

0.72

ETH

0.72

enegger

0.70

hess

0.68

>"

0.66

heastern

0.66

forth

0.66

 Ambro

0.66

ellar

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.