INDEX

Explanations

the beginning of a document, indicated by the presence of the special token "<bos>"

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

Q:

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.20.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

 autorytatywna

-0.74

AndEndTag

-0.73

-------

-0.73

 MainAxisSize

-0.68

SBATCH

-0.68

abestanden

-0.67

 beginnetje

-0.67

verwijspagina

-0.63

 disambiguazione

-0.62



-0.61

POSITIVE LOGITS

prite

0.50

MessageTagHelper

0.49

хьтан

0.49

ModelState

0.48

Rendering

0.45

InputTagHelper

0.45

suz

0.44

 cercle

0.44

Palmar

0.43

Regressor

0.42

Activations Density 0.008%

the beginning of a document, indicated by the presence of the special token "<bos>"

Q:

No Comments

No Known Activations