INDEX

Explanations

names or proper nouns of people or characters

oai_token-act-pair · gpt-3.5-turbo

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

was

-0.93

-0.82

tained

-0.82

MpServer

-0.81

IFIED

-0.79

been

-0.79

 Been

-0.74

WAS

-0.72

ected

-0.72

needs

-0.72

POSITIVE LOGITS

 begin

1.24

 decide

1.23

 arrive

1.17

 discover

1.13

 settle

1.12

 stumble

1.10

 proceed

1.10

 scramble

1.10

 manage

1.10

 emerge

1.08

Activations Density 0.669%

names or proper nouns of people or characters

No Comments

No Known Activations