INDEX

Explanations

references to error rates and coding theory concepts

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

error differences distance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

</thead>

-0.33

 Италијани

-0.33

 مض

-0.33

boxylic

-0.32

 للاسماء

-0.32

留

-0.32

Doe

-0.31

 Freuden

-0.31

contentView

-0.31

 evicted

-0.31

POSITIVE LOGITS

 CanadaChoose

0.59

DataAnnotations

0.55

COUVER

0.55

 EconPapers

0.55

ValueStyle

0.50

 <=",

0.50

 يتيمه

0.49

Халык

0.48

tanleria

0.48

IndentedString

0.48

Activations Density 0.020%

references to error rates and coding theory concepts

error differences distance

No Comments

No Known Activations