INDEX

Explanations

dialogue attributions (e.g., "he said," "she yelled")

oai_token-act-pair · gpt-3.5-turbo

dialogue related to characters' interactions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

arding

-0.70

raviolet

-0.69

atana

-0.68

 Architects

-0.67

ardless

-0.66

otal

-0.65

iencies

-0.65

artifacts

-0.63

selection

-0.63

adelphia

-0.63

POSITIVE LOGITS

 muttered

1.37

 exclaimed

1.34

 whispered

1.29

 shouted

1.28

 yelled

1.28

 murm

1.21

 exclaim

1.19

 screamed

1.16

 chanted

1.15

 replied

1.12

Activations Density 0.140%

dialogue attributions (e.g., "he said," "she yelled")

dialogue related to characters' interactions

No Known Activations