INDEX

Explanations

whitespace or indentation characters, indicating formatting or organization in the text.

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

awtextra

-0.63

featureID

-0.57

 Administrativna

-0.54

windowFixed

-0.52

oa̍t

-0.52

protoimpl

-0.50

новниш

-0.49

RectangleBorder

-0.47

findpost

-0.47

 MessageBoxIcon

-0.47

POSITIVE LOGITS

グレ

0.41

 Module

0.40

 Wiktionnaire

0.40

học

0.39

↺

0.39

MIT

0.38

 Productivity

0.38

デュー

0.38

 mihi

0.38

Collegamenti

0.38

Activations Density 0.002%

whitespace or indentation characters, indicating formatting or organization in the text.

No Comments

No Known Activations