INDEX

Explanations

formatting symbols and components related to URLs or hyperlinks

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

web addresses

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

 يتيمه

-0.88

⮕

-0.87

NameInMap

-0.86

[@BOS@]

-0.85

<unused52>

-0.85

<unused23>

-0.85

<unused42>

-0.85

<unused41>

-0.85

<unused8>

-0.85

<unused16>

-0.85

POSITIVE LOGITS

 http

0.42

http

0.38

 following

0.37

↵↵

0.35

↵

0.35

 https

0.35

 address

0.35

next

0.35

Activations Density 0.014%

formatting symbols and components related to URLs or hyperlinks

web addresses

No Comments

No Known Activations