INDEX

Explanations

attends to the token "Na" from later tokens related to characters in a narrative or story

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

0:0.07

1:0.08

2:0.08

3:0.11

4:0.09

5:0.05

6:0.31

7:0.17

Negative Logits

padek

-0.29

ViewChild

-0.29

 resourceCulture

-0.28

 inappropriés

-0.28

 textStatus

-0.26

ñado

-0.25

 Hodgkin

-0.25

glGen

-0.25

 ISNI

-0.25

POSITIVE LOGITS

SourceChecksum

0.33

 насељу

0.33

 thingy

0.33

Hauptartikel

0.32

 thee

0.32

animo

0.32

Tagged

0.32

jublj

0.32

 alre

0.30

Scientist

0.30

Activations Density 0.080%

attends to the token "Na" from later tokens related to characters in a narrative or story

No Comments

No Known Activations