INDEX

Explanations

specific combinations of letters and character sequences related to structured data or coding syntax

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

math labels and names

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

<tr>

-0.50

 Púb

-0.45

 acest

-0.44

ininger

-0.43

jenje

-0.43

 Grüsse

-0.43

čenje

-0.43

lij

-0.42

tings

-0.42

 camiseta

-0.41

POSITIVE LOGITS

ا

0.95

0.79

اا

0.71

aia

0.71

ela

0.71

ษา

0.69

lda

0.69

hoga

0.69

TCA

0.69

epa

0.69

Activations Density 1.759%

specific combinations of letters and character sequences related to structured data or coding syntax

math labels and names

No Comments

No Known Activations