INDEX

Explanations

intensifiers or modifiers that amplify adjectives or experiences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 35-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_35/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.35.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

quite

-0.45

 quite

-0.44

 bastante

-0.43

 Quite

-0.42

 cukup

-0.41

割と

-0.40

naj

-0.40

('/',

-0.38

StatusOK

-0.38

 lots

-0.38

POSITIVE LOGITS

 unbelievably

0.78

 goddamn

0.70

 damn

0.69

 incredibly

0.67

 freakin

0.66

 wonderful

0.63

AndEndTag

0.62

 wonderfully

0.61

 amazingly

0.61

 incredible

0.60

Activations Density 0.085%

intensifiers or modifiers that amplify adjectives or experiences

No Comments

No Known Activations