INDEX

Explanations

proper acronyms containing 'KC'

oai_token-act-pair · gpt-3.5-turbo

references to the news agency "KCNA."

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

Ã¨

-0.76

 drain

-0.68

 particle

-0.66

 mathematical

-0.66

fro

-0.65

 swimming

-0.64

 figur

-0.64

rine

-0.64

fall

-0.63

 voting

-0.63

POSITIVE LOGITS

KC

4.21

KC

3.63

KD

1.67

KT

1.53

KP

1.49

NK

1.47

KT

1.38

KR

1.36

NK

1.33

KM

1.32

Activations Density 0.017%

proper acronyms containing 'KC'

references to the news agency "KCNA."

No Comments

No Known Activations