INDEX

Explanations

references to or descriptions of pairs of objects or items

oai_token-act-pair · gpt-3.5-turbo

references to items that are typically found in pairs

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ablishment

-1.10

information

-0.82

population

-0.79

ilial

-0.79

urden

-0.79

ICLE

-0.79

aviour

-0.78

argument

-0.78

ername

-0.75

ructure

-0.75

POSITIVE LOGITS

 scissors

1.49

 glasses

1.28

 gloves

1.23

 jeans

1.23

 sunglasses

1.23

 headphones

1.23

 handcuffs

1.22

 socks

1.22

 pistols

1.21

 sneakers

1.19

Activations Density 0.277%

references to or descriptions of pairs of objects or items

references to items that are typically found in pairs

No Comments

No Known Activations

references to or descriptions of pairs of objects or items

references to items that are typically found in pairs

No Comments

No Known Activations