INDEX

Explanations

instances of the word "flex."

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

 Briefly

-1.81

 Spacewatch

-1.68

 slightest

-1.48

soever

-1.46

 beginnings

-1.44

adult

-1.41

perg

-1.41

odot

-1.40

 same

-1.37

 Calif

-1.37

POSITIVE LOGITS

ibly

2.10

ÃŃvel

1.80

istry

1.71

ors

1.61

gap

1.60

ificates

1.59

 version

1.58

ENTIAL

1.55

sheet

1.54

chain

1.54

Activations Density 0.012%

instances of the word "flex."

No Comments

No Known Activations