INDEX

Explanations

positively framed evaluations or statements

oai_token-act-pair · gpt-3.5-turbo

phrases that denote the concept of something being beneficial or advantageous

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

inav

-0.74

rows

-0.70

ao

-0.68

cled

-0.68

yrinth

-0.66

 millenn

-0.66

oug

-0.65

è»

-0.65

©¶æ

-0.64

bern

-0.64

POSITIVE LOGITS

ifier

0.73

Else

0.72

ional

0.71

 happen

0.70

 Happ

0.70

 happening

0.68

--+

0.68

terday

0.67

unity

0.67

iously

0.67

Activations Density 0.025%

positively framed evaluations or statements

phrases that denote the concept of something being beneficial or advantageous

No Comments

No Known Activations