INDEX

Explanations

phrases related to congratulations

oai_token-act-pair · gpt-3.5-turbo

phrases expressing congratulations or celebratory remarks

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ix

-0.70

 style

-0.70

Market

-0.68

 overhe

-0.64

 militar

-0.64

 Hour

-0.62

 hunger

-0.60

 Hear

-0.60

 sectarian

-0.60

 subpoena

-0.60

POSITIVE LOGITS

Congratulations

3.31

 congratulations

3.30

rats

3.18

 Congratulations

3.13

 congr

2.79

 congratulate

2.20

ratulations

2.12

 congrat

2.11

 congratulated

1.87

 condolences

1.64

Activations Density 0.034%

phrases related to congratulations

phrases expressing congratulations or celebratory remarks

No Comments

No Known Activations