INDEX

Explanations

references to size and length in various contexts

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

full followed by size, scale, length, or feature

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

 regalos

-0.43

 ocasião

-0.43

 évêque

-0.42

 programación

-0.40

 votación

-0.39

 aparcamiento

-0.39

 programmation

-0.39

 welcher

-0.39

 criterios

-0.39

 sugerencias

-0.39

POSITIVE LOGITS

Full

1.08

 Full

1.04

 full

0.99

full

0.98

FULL

0.97

 FULL

0.97

 Fully

0.93

 fully

0.90

 fullName

0.89

 fullt

0.86

Activations Density 0.015%

references to size and length in various contexts

full followed by size, scale, length, or feature

No Comments

No Known Activations