INDEX

Explanations

proper nouns related to individuals, possibly in a news or political context

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

terday

-0.74

EED

-0.72

 conclud

-0.71

eele

-0.64

anwhile

-0.63

Reference

-0.61

Sax

-0.59

ENTS

-0.58

henko

-0.58

ateral

-0.57

POSITIVE LOGITS

rique

1.19

ning

1.05

riks

0.95

rik

0.94

sel

0.93

lein

0.92

agar

0.92

nery

0.90

ricks

0.88

ners

0.88

Activations Density 6.595%

proper nouns related to individuals, possibly in a news or political context

No Comments

No Known Activations