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

lands

-0.75

gro

-0.73

sha

-0.68

ho

-0.66

 Quin

-0.66

uph

-0.66

cro

-0.63

inn

-0.63

NAS

-0.63

Reviewer

-0.63

POSITIVE LOGITS

renheit

0.90

©¶æ

0.72

ipeg

0.65

anchester

0.64

actionDate

0.64

»

0.63

 inform

0.63

 ancest

0.62

 ILCS

0.62

nce

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.