INDEX

Explanations

messages or communication-related phrases

oai_token-act-pair · gpt-3.5-turbo

references to messages or communications

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ptions

-0.71

Zup

-0.69

chester

-0.68

iba

-0.68

borough

-0.65

unn

-0.64

lymp

-0.63

elson

-0.63

zan

-0.63

stocks

-0.63

POSITIVE LOGITS

 message

3.66

message

2.65

 messages

2.60

 Message

2.53

Message

2.25

 messaging

2.19

 Messages

2.13

 messenger

1.77

msg

1.48

mess

1.45

Activations Density 0.018%

messages or communication-related phrases

references to messages or communications

No Comments

No Known Activations