INDEX

Explanations

attends to the tokens related to method or function calls from preceding interface-related tokens

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.17

2:0.05

3:0.04

4:0.08

5:0.41

6:0.05

7:0.05

Negative Logits

 Jefus

-0.31

 ―――――

-0.31

 ་་

-0.31

Emo

-0.30

 Chrift

-0.30

 Ponto

-0.29

 Sopho

-0.29

 itſelf

-0.29

ddots

-0.29

 againſt

-0.28

POSITIVE LOGITS

DoubleQuotes

0.34

@"/

0.33

SharedDtor

0.32

 gyhoeddwyd

0.32

ativement

0.31

 ApiClient

0.31

Попис

0.30

رشف

0.29

 GenerationType

0.29

Personensuche

0.28

Activations Density 0.012%

attends to the tokens related to method or function calls from preceding interface-related tokens

No Comments

No Known Activations