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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

NAS

-0.89

Medic

-0.81

ãĥĨãĤ£

-0.76

rosso

-0.75

Ult

-0.74

ãĤ¼ãĤ¦ãĤ¹

-0.73

Haw

-0.69

connected

-0.69

 Married

-0.69

SEA

-0.68

POSITIVE LOGITS

 bottleneck

0.84

 cane

0.80

 broom

0.74

 monk

0.70

 handler

0.69

 Ottoman

0.68

 tourist

0.68

 leash

0.67

 frown

0.67

 Buddhist

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.