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

«ĺ

-0.64

ICE

-0.63

gdala

-0.63

 heads

-0.63

endants

-0.62

ICES

-0.62

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.62

 func

-0.61

QC

-0.61

 Jagu

-0.61

POSITIVE LOGITS

Course

0.76

Wer

0.70

 Wheeler

0.69

Brand

0.68

auld

0.67

ospons

0.67

 Draft

0.67

asser

0.66

achev

0.65

raft

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.