INDEX

Explanations

the word "composed", indicating a focus on structural or compositional descriptions

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

composed of

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

 ainfi

-0.74

<bos>

-0.61

 plufieurs

-0.58

ironically

-0.57

LError

-0.56

everyone

-0.55

 lendemain

-0.54

everybody

-0.54

 malheur

-0.54

 présidenti

-0.53

POSITIVE LOGITS

 composed

2.00

 Composed

1.69

composed

1.59

 compose

1.24

 comprised

1.20

 composta

1.20

 composée

1.19

 composto

1.12

 composing

1.12

 compuesta

1.09

Activations Density 0.006%

the word "composed", indicating a focus on structural or compositional descriptions

composed of

No Comments

No Known Activations