INDEX

Explanations

attends to font-related tokens from specific tokens that indicate a style or specification

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

0:0.13

1:0.11

2:0.32

3:0.05

4:0.07

5:0.07

6:0.05

7:0.17

Negative Logits

 segu

-0.25

用

-0.23

 follow

-0.23

</b>

-0.22

以

-0.22

nor

-0.22

Nor

-0.21

打ち

-0.21

 след

-0.21

بع

-0.21

POSITIVE LOGITS

.*")]

0.44

 purpoſe

0.43

 BoxFit

0.43

adaptiveStyles

0.42

Efq

0.42

 AssemblyVersion

0.41

+#+#

0.40

 snippetHide

0.40

 reaſon

0.39

 pleaſure

0.39

Activations Density 0.016%

attends to font-related tokens from specific tokens that indicate a style or specification

No Comments

No Known Activations