INDEX

Explanations

phrases related to directions

oai_token-act-pair · gpt-3.5-turbo

references to geographical regions, particularly the western areas

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

 Splash

-0.79

Trust

-0.73

 Sounders

-0.72

 Lifetime

-0.71

 Sauce

-0.70

pace

-0.70

 Spaces

-0.69

ante

-0.68

 Spotlight

-0.67

 Pell

-0.65

POSITIVE LOGITS

 eastern

2.84

 western

2.77

 southwestern

2.08

 northeastern

2.08

 southeastern

2.07

 northern

2.00

 northwestern

1.94

 southern

1.93

western

1.63

astern

1.31

Activations Density 0.014%

phrases related to directions

references to geographical regions, particularly the western areas

No Comments

No Known Activations