INDEX

Explanations

indications of correctness or appropriateness

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

New Auto-Interp

Configuration

ctigges/pythia-70m-deduped__mlp-sm_processed/0-mlp-sm

Prompts (Dashboard)

32,768 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

32,768

Data Type

torch.float32

Hook Name

blocks.0.hook_mlp_out

Hook Layer

Architecture

standard

Context Size

128

Dataset

EleutherAI/the_pile_deduplicated

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ching

-1.73

rc

-1.42

eline

-1.39

asio

-1.36

OUN

-1.35

DEBUG

-1.31

 truth

-1.30

 fact

-1.29

 fair

-1.26

NYC

-1.26

POSITIVE LOGITS

aria

1.83

iom

1.64

·¸

1.59

veolar

1.54

ĻĤ

1.54

keeping

1.51

eous

1.50

keeper

1.48

keepers

1.45

plant

1.45

Activations Density 0.047%

indications of correctness or appropriateness

No Comments

No Known Activations