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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Fax

-0.78

leans

-0.72

HCR

-0.71

dinand

-0.71

theless

-0.70

schild

-0.70

ļéĨĴ

-0.69

orkshire

-0.67

Scan

-0.66

alus

-0.65

POSITIVE LOGITS

 vanilla

0.77

iggurat

0.68

 Vanilla

0.67

é¾

0.64

 saturated

0.62

yg

0.61

Chu

0.60

earance

0.60

 downwards

0.60

gha

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.