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

Negative Logits

advertising

-0.77

oct

-0.73

awa

-0.73

 behav

-0.71

||||

-0.70

luaj

-0.69

 surv

-0.68

accompan

-0.67

 ingred

-0.67

 conduc

-0.67

POSITIVE LOGITS

 Poles

0.70

ancing

0.66

 Hung

0.66

oli

0.65

bars

0.64

 parked

0.64

oll

0.64

liam

0.64

ikes

0.64

iths

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.