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

mma

-0.83

"},{"

-0.82

enza

-0.81

ãĥĺãĥ©

-0.79

ardless

-0.77

assis

-0.76

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.73

guard

-0.73

xual

-0.73

POSITIVE LOGITS

 circulate

0.63

 flakes

0.61

 validity

0.61

 districts

0.61

 profit

0.60

Leary

0.59

 Branch

0.59

 disposable

0.59

 bubbles

0.58

 Bengals

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.