INDEX

Explanations

variations of the word "gla," indicating a focus on glass or related materials

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

gla- prefix

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

 numbers

-0.48

 parts

-0.47

 number

-0.47

 Peterson

-0.46

 items

-0.45

élément

-0.44

Number

-0.44

 part

-0.44

ResponseEntity

-0.44

 Number

-0.43

POSITIVE LOGITS

chengladbach

1.00

Gla

0.88

 Glaze

0.88

Gla

0.87

GLA

0.84

GLA

0.77

 Glaser

0.76

 glazing

0.73

 Glau

0.73

gla

0.71

Activations Density 0.004%

variations of the word "gla," indicating a focus on glass or related materials

gla- prefix

No Comments

No Known Activations