INDEX

Explanations

emphatic expressions of degree or intensity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-mlp-16k

Configuration

google/gemma-scope-2b-pt-mlp/layer_1/width_16k/average_l0_105

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.hook_mlp_out

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 quite

-2.80

quite

-2.59

Quite

-2.47

 Quite

-2.45

 uite

-1.22

 ganske

-1.21

 bastante

-1.15

 ganska

-1.02

 довольно

-1.00

 destul

-0.99

POSITIVE LOGITS

 Aseg

0.47

 pageContext

0.45

 appartamento

0.44

kaw

0.43

Că

0.43

Denn

0.42

eqref

0.42

ლი

0.42

 inni

0.42

 dedi

0.41

Activations Density 0.065%

emphatic expressions of degree or intensity

No Comments

No Known Activations