INDEX

Explanations

short phrases ending with punctuation marks (such as periods, colons, question marks, exclamation points)

oai_token-act-pair · gpt-3.5-turbo

elements related to scripted or formal conversations and interviews

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 derivatives

-0.76

 subord

-0.75

 decoration

-0.75

 decap

-0.74

 inflic

-0.74

 decorative

-0.73

 rendering

-0.73

 totals

-0.68

 backups

-0.68

 relegation

-0.68

POSITIVE LOGITS

VICE

1.62

Interview

1.57

Question

1.51

Tell

1.26

1.24

QUEST

1.23

Hello

1.22

Listen

1.20

Hi

1.19

Asked

1.19

Activations Density 0.195%

short phrases ending with punctuation marks (such as periods, colons, question marks, exclamation points)

elements related to scripted or formal conversations and interviews

No Comments

No Known Activations