INDEX

Explanations

attends to specific tokens indicating properties or characteristics from earlier tokens defining or explaining 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.06

1:0.08

2:0.07

3:0.09

4:0.09

5:0.04

6:0.35

7:0.18

Negative Logits

<s>

-0.43

UpInside

-0.42

 tør

-0.41

 سكانية

-0.40

epam

-0.40

__(/*!

-0.39

Haf

-0.39

WEBPACK

-0.39

 πισ

-0.38

 المعيارى

-0.37

POSITIVE LOGITS

fitrión

0.53

Xna

0.45

WithIOException

0.42

 CHtml

0.41

 propOrder

0.40

 aDecoder

0.40

hood

0.40

ená

0.40

 correctes

0.40

Биография

0.40

Activations Density 0.584%

attends to specific tokens indicating properties or characteristics from earlier tokens defining or explaining them

No Comments

No Known Activations