INDEX

Explanations

names or references related to "Ko"

oai_token-act-pair · gpt-3.5-turbo

repeated references to specific names and titles

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

acus

-0.73

idium

-0.70

icum

-0.69

 intox

-0.68

resist

-0.68

actionDate

-0.68

 cort

-0.67

resistant

-0.67

abor

-0.67

 Cadillac

-0.67

POSITIVE LOGITS

Ko

2.72

sho

2.32

Sho

2.19

Ko

2.13

Fo

1.95

ko

1.78

Wo

1.57

FO

1.36

Wo

1.32

Ho

1.32

Activations Density 0.031%

names or references related to "Ko"

repeated references to specific names and titles

No Comments

No Known Activations