INDEX

Explanations

keywords related to deviations from the norm

oai_token-act-pair · gpt-3.5-turbo

the use of the word "normally" and its variations to indicate consistency or typicality

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

ayer

-0.60

iere

-0.60

asons

-0.59

bid

-0.57

ixt

-0.57

enges

-0.57

gov

-0.55

 AMER

-0.55

 Architects

-0.55

chuk

-0.55

POSITIVE LOGITS

 normally

3.65

 ordinarily

2.69

 usually

2.08

 typically

2.03

 Normally

1.88

Normally

1.83

 traditionally

1.70

usually

1.69

norm

1.59

typically

1.56

Activations Density 0.008%

keywords related to deviations from the norm

the use of the word "normally" and its variations to indicate consistency or typicality

No Comments

No Known Activations