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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Rab

-0.69

cheon

-0.65

Merit

-0.65

ron

-0.65

 Horn

-0.65

 Springer

-0.64

Orb

-0.64

oros

-0.62

Pit

-0.61

 WATCHED

-0.61

POSITIVE LOGITS

zzle

0.72

atche

0.68

ester

0.68

 offend

0.67

aunder

0.66

 coerc

0.66

sto

0.66

ying

0.65

 cond

0.65

verett

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.