INDEX

Explanations

attends to court case citations that contain a specific court token from corresponding numeric markers

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

0:0.10

1:0.11

2:0.08

3:0.09

4:0.04

5:0.02

6:0.23

7:0.30

Negative Logits

Hentet

-0.35

 FileWriter

-0.33

toMatchSnapshot

-0.33

SBATCH

-0.31

CELLANEOUS

-0.30

 JScrollPane

-0.30

PerformLayout

-0.30

MemoryWarning

-0.30

erevan

-0.29

 FormControl

-0.29

POSITIVE LOGITS

<code>

0.28

度

0.25

 Lors

0.25

diğinde

0.25

在

0.25

يكب

0.24

俗

0.24

向

0.22

ThroughAttribute

0.22

 unconsciously

0.22

Activations Density 0.049%

attends to court case citations that contain a specific court token from corresponding numeric markers

No Comments

No Known Activations