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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Introduced

-0.72

xxx

-0.69

scl

-0.68

catentry

-0.66

aldi

-0.64

":{"

-0.63

odynam

-0.60

Director

-0.59

lamm

-0.59

chio

-0.58

POSITIVE LOGITS

 Yanukovych

0.65

acles

0.64

 spoiler

0.61

hya

0.61

 Yanuk

0.61

nex

0.59

ezvous

0.59

 meshes

0.58

 abduct

0.57

hern

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.