INDEX

Explanations

locations or positions that are taken up or filled by someone or something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "occupied" and its various forms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

raft

-0.78

peer

-0.76

uber

-0.70

Fly

-0.70

ortality

-0.64

hon

-0.64

Î»

-0.62

Roll

-0.61

Run

-0.61

brother

-0.60

POSITIVE LOGITS

itures

0.91

iture

0.89

ments

0.85

ngth

0.85

 territories

0.84

 occupying

0.83

 occupied

0.83

 occupy

0.80

 Territories

0.78

 calories

0.75

Activations Density 0.014%

locations or positions that are taken up or filled by someone or something

instances of the word "occupied" and its various forms

No Comments

No Known Activations