INDEX

Explanations

phrases indicating someone as part of a group or team

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

zeb

-0.72

 agre

-0.72

 chall

-0.66

ption

-0.64

 lapt

-0.63

 diversion

-0.62

 miscarriage

-0.61

 mathemat

-0.61

 anytime

-0.61

 streng

-0.60

POSITIVE LOGITS

few

0.80

 many

0.78

those

0.75

 those

0.75

 several

0.73

abet

0.69

 wealthiest

0.68

 dozens

0.68

amen

0.66

many

0.66

Activations Density 0.069%

phrases indicating someone as part of a group or team

No Comments

No Known Activations