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

vol

-0.80

bos

-0.75

Oo

-0.75

asha

-0.74

 magistrate

-0.70

 Baxter

-0.68

ãĤ©

-0.68

mare

-0.66

 Piper

-0.66

mob

-0.65

POSITIVE LOGITS

SpaceEngineers

0.77

POS

0.76

soDeliveryDate

0.72

WER

0.70

REC

0.68

sonian

0.66

ertation

0.65

 perl

0.64

ÑĮ

0.64

gments

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.