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

actionDate

-0.89

Discover

-0.71

uyomi

-0.66

esis

-0.66

vt

-0.66

 Saga

-0.66

 Beard

-0.66

ended

-0.65

 Dragonbound

-0.65

 Pets

-0.65

POSITIVE LOGITS

ãĥ³ãĤ¸

0.69

 sling

0.68

 commissioners

0.64

%:

0.63

rule

0.61

anches

0.61

 Fresno

0.61

 line

0.60

 Sioux

0.59

 confession

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.