INDEX

Explanations

politically charged names or concepts

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt

Prompts (Dashboard)

12,288 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

46,080

Data Type

torch.float32

Hook Point

blocks.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

apollo-research/Skylion007-openwebtext-tokenizer-gpt2

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ually

-0.38

phant

-0.36

hang

-0.35

onga

-0.34

akia

-0.34

sembly

-0.32

chest

-0.32

elfth

-0.32

istant

-0.32

ions

-0.32

POSITIVE LOGITS

atto

0.36

 McCarthy

0.36

rint

0.35

vine

0.35

hyde

0.34

 Chapman

0.33

 ancest

0.32

lore

0.32

raviolet

0.32

mson

0.32

Activations Density 12.125%

politically charged names or concepts

No Comments

No Known Activations