INDEX

Explanations

locations or places related to events or activities

oai_token-act-pair · gpt-3.5-turbo

place names, particularly focusing on the names of countries, states, and cities

oai_token-act-pair · gpt-3.5-turbo Triggered by @jordan

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scefr-ajt/6-res_scefr-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

Afee

-0.96

BOOK

-0.92

-0.86

bread

-0.84

ãĤ½

-0.84

 bumper

-0.83

ãĥĳ

-0.80

LOAD

-0.80

bats

-0.79

ufact

-0.79

POSITIVE LOGITS

;;;;;;;;;;;;

0.98

.,

0.97

 Territory

0.92

 Converted

0.90

Jed

0.89

jer

0.89

isco

0.88

Rex

0.87

Ely

0.83

.;

0.82

Activations Density 0.828%

locations or places related to events or activities

place names, particularly focusing on the names of countries, states, and cities

No Comments

No Known Activations