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

 Authorization

-0.64

rontal

-0.63

allow

-0.63

hovah

-0.63

Interstitial

-0.62

nation

-0.61

 behavi

-0.60

 Presidents

-0.60

 dilig

-0.60

shall

-0.60

POSITIVE LOGITS

ppo

0.76

pta

0.70

"}

0.68

RIC

0.67

aus

0.66

abilia

0.65

usha

0.65

rol

0.64

APS

0.62

 Dani

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.