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

-0.77

 println

-0.75

-0.74

ioxide

-0.74

Merit

-0.71

ister

-0.70

Ĥª

-0.69

issan

-0.68

ibaba

-0.68

 Topic

-0.64

POSITIVE LOGITS

 subsistence

0.76

 coerc

0.74

 corrections

0.69

 entr

0.69

 TOUR

0.67

 promoters

0.66

 tuber

0.66

 agric

0.65

 tours

0.65

 bona

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.