INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

ĸļ

-0.84

hao

-0.72

hyde

-0.65

 Grac

-0.65

 spir

-0.65

 grun

-0.63

 Wool

-0.63

lax

-0.63

MUS

-0.63

 attRot

-0.62

POSITIVE LOGITS

ynski

0.77

pload

0.73

pter

0.71

Interstitial

0.70

*)

0.70

izont

0.69

lisher

0.69

elfth

0.68

olester

0.68

tch

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.