INDEX

Explanations

attends to various actions or states from corresponding referential tokens that contextualize them

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

0:0.16

1:0.55

2:0.06

3:0.03

4:0.04

5:0.02

6:0.03

7:0.06

Negative Logits

IonicModule

-0.52

webElementXpaths

-0.52

parsedMessage

-0.51

dafx

-0.48

complexContent

-0.45

tagext

-0.44

WebVitals

-0.43

('/');

-0.42

UnusedPrivate

-0.42

RepeatedField

-0.42

POSITIVE LOGITS

 كومونز

0.39

 aikana

0.36

virons

0.36

obraz

0.36

 Elliott

0.35

 Audiodateien

0.35

帖最后由

0.35

ota

0.35

vemos

0.35

Ђ

0.35

Activations Density 3.039%

attends to various actions or states from corresponding referential tokens that contextualize them

No Comments

No Known Activations