INDEX

Explanations

references to memorabilia or official documents

oai_token-act-pair · gpt-3.5-turbo

references to memories and memoranda

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

wing

-0.70

pring

-0.65

wings

-0.65

DF

-0.63

pressure

-0.63

 Pill

-0.63

atform

-0.62

azo

-0.61

Ïģ

-0.60

Rich

-0.60

POSITIVE LOGITS

 memor

4.24

 Memor

2.51

 memorandum

1.20

 recite

1.19

 memos

1.12

 remem

1.08

 souven

1.02

 rele

0.97

 remembering

0.96

 scrib

0.94

Activations Density 0.015%

references to memorabilia or official documents

references to memories and memoranda

No Comments

No Known Activations