INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

opter

-0.75

ongevity

-0.71

agogue

-0.71

aper

-0.71

utters

-0.69

ternity

-0.66

Pakistan

-0.66

BILITY

-0.66

clipse

-0.65

pez

-0.64

POSITIVE LOGITS

 Points

0.68

*)

0.68

nown

0.64

laden

0.61

achev

0.61

req

0.59

ersed

0.59

huh

0.59

rah

0.59

ession

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.