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

uly

-0.68

ague

-0.67

grass

-0.67

 Slaughter

-0.67

 dirty

-0.64

 agric

-0.64

 inconvenient

-0.61

tid

-0.61

eding

-0.60

Interstitial

-0.60

POSITIVE LOGITS

hold

0.97

ãĥ¼ãĥĨ

0.74

ORT

0.67

RET

0.66

allows

0.65

 Ferr

0.65

Interest

0.65

Zhu

0.63

è£ħ

0.63

 Villa

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.