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

SELECT

-0.67

clave

-0.67

forth

-0.67

TeX

-0.64

Sin

-0.63

gha

-0.63

Jen

-0.62

Pattern

-0.61

vier

-0.61

ugh

-0.60

POSITIVE LOGITS

Į

0.66

ice

0.65

regon

0.64

abama

0.63

 JUSTICE

0.61

mpire

0.61

 aides

0.60

Ħ

0.59

 Shake

0.58

atl

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.