INDEX

Explanations

descriptors and conditional phrases related to preferences or suggestions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 4-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_4/width_16k/average_l0_124

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.4.hook_resid_post

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

новниш

-0.57

,:),

-0.55

 qashqai

-0.54

 Meksiku

-0.53

 murió

-0.53

inning

-0.53

IVEREF

-0.53

spesies

-0.53

]=>

-0.52

wapV

-0.51

POSITIVE LOGITS

 Would

0.82

 wäre

0.80

 would

0.78

Would

0.78

 olisi

0.76

 serait

0.75

 wären

0.75

 gdyby

0.74

would

0.71

 aurait

0.71

Activations Density 0.449%

descriptors and conditional phrases related to preferences or suggestions

No Comments

No Known Activations