INDEX

Explanations

attends to derogatory or negative descriptors from tokens that introduce or summarize topics

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.10.attn.hook_z

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.05

1:0.07

2:0.06

3:0.09

4:0.08

5:0.04

6:0.47

7:0.11

Negative Logits

гӀ

-0.35

 seriously

-0.28

pan

-0.28

 snippetHide

-0.26

 circumstances

-0.25

 certainly

-0.25

»

-0.25

-0.24

 severe

-0.24

 probably

-0.23

POSITIVE LOGITS

 tartalomajánló

0.49

__':

0.48

LookAnd

0.47

 '\\;'

0.47

хьтан

0.47

__*/

0.46

 configureStore

0.46

 AssemblyProduct

0.46

 {}));

0.46

/−

0.45

Activations Density 0.187%

attends to derogatory or negative descriptors from tokens that introduce or summarize topics

No Comments

No Known Activations