INDEX

Explanations

questions or statements made in a conversation

oai_token-act-pair · gpt-3.5-turbo

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

aples

-0.70

elson

-0.64

NCT

-0.63

artifacts

-0.62

MU

-0.62

iHUD

-0.61

earable

-0.61

guyen

-0.59

İĭ

-0.59

bledon

-0.59

POSITIVE LOGITS

 aloud

1.26

 sarcast

1.20

 softly

1.18

 loudly

1.11

 nervously

1.10

 indign

1.09

 calmly

1.08

 patiently

1.06

 impatient

1.05

 quietly

1.05

Activations Density 0.111%

questions or statements made in a conversation

No Comments

No Known Activations