INDEX

Explanations

numerical values representing the lowest level or count of something

oai_token-act-pair · gpt-3.5-turbo

references to minimum values or the term "lowest."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

mun

-0.73

ortium

-0.68

NM

-0.68

XXXX

-0.68

NEWS

-0.68

bara

-0.67

lia

-0.66

Pros

-0.66

rs

-0.65

DOM

-0.65

POSITIVE LOGITS

 detectable

0.71

 Countries

0.69

 bidder

0.68

dden

0.67

lid

0.66

 lowest

0.66

 imaginable

0.66

 cooker

0.66

 lash

0.65

case

0.65

Activations Density 0.019%

numerical values representing the lowest level or count of something

references to minimum values or the term "lowest."

No Comments

No Known Activations