INDEX

Explanations

requests or instructions for action

oai_token-act-pair · gpt-3.5-turbo

instances of the word "please" in various contexts

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

arc

-0.78

MpServer

-0.78

ARC

-0.69

é¾

-0.67

senal

-0.67

mund

-0.64

cler

-0.64

phrine

-0.64

UF

-0.64

 Huntington

-0.63

POSITIVE LOGITS

 advise

0.90

 excuse

0.88

 fill

0.88

 Ignore

0.82

 ignore

0.82

 note

0.81

sir

0.81

 forgive

0.79

 refrain

0.79

 inquire

0.78

Activations Density 0.015%

requests or instructions for action

instances of the word "please" in various contexts

No Comments

No Known Activations