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

SOC

-0.70

reb

-0.66

OPER

-0.64

 Atlantic

-0.64

 Diamond

-0.63

KC

-0.60

bang

-0.60

 SEAL

-0.59

 Stream

-0.59

 shell

-0.59

POSITIVE LOGITS

acters

0.77

itudes

0.77

axter

0.74

isexual

0.72

 surgery

0.72

ancers

0.71

chery

0.71

izes

0.71

rencies

0.71

chers

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.