INDEX

Explanations

attends to specific keywords or structural elements from related coding or data formatting 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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.11

1:0.14

2:0.12

3:0.11

4:0.14

5:0.10

6:0.09

7:0.15

Negative Logits

 personalmente

-0.20

 Itself

-0.20

 Neighborhood

-0.20

 itself

-0.19

 Futsal

-0.19

ḫ

-0.19

 synapse

-0.19

院

-0.19

 asfal

-0.18

 Fixture

-0.18

POSITIVE LOGITS

 виправивши

0.33

MessageTagHelper

0.33

Tikang

0.32

 initComponents

0.32

__':

0.32

[++

0.31

 Later

0.30

 uska

0.30

LabelTagHelper

0.30

ganu

0.30

Activations Density 0.213%

attends to specific keywords or structural elements from related coding or data formatting tokens

No Comments

No Known Activations