INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ãĥīãĥ©ãĤ´ãĥ³

-0.81

\">

-0.78

ãĥĬ

-0.75

venge

-0.74

Offline

-0.74

ãĥĵ

-0.73

Fal

-0.72

FH

-0.70

WINDOWS

-0.70

Charge

-0.70

POSITIVE LOGITS

iosyncr

0.72

oing

0.70

 intellig

0.64

 noses

0.64

matic

0.64

 note

0.61

 puppet

0.61

olescent

0.60

 swat

0.60

 blanket

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.