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

interstitial

-0.91

Yi

-0.68

Yang

-0.68

 concess

-0.63

 Huang

-0.63

igham

-0.62

Xi

-0.60

 Turing

-0.59

ijing

-0.59

LINE

-0.58

POSITIVE LOGITS

*/(

0.87

chio

0.69

etsk

0.67

ãĤ£

0.67

aceae

0.66

emouth

0.66

 balloons

0.65

xual

0.65

 Clover

0.64

 Passage

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.