INDEX

Explanations

attends to time-related tokens marked from sequential dialogue timestamps to earlier spoken phrases

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

0:0.10

1:0.12

2:0.09

3:0.06

4:0.13

5:0.36

6:0.04

7:0.06

Negative Logits

Referències

-0.27

 păr

-0.26

Esa

-0.23

typeorm

-0.23

 μή

-0.23

osa

-0.23

ContentAsync

-0.22

bitan

-0.22

rasta

-0.22

Pinch

-0.22

POSITIVE LOGITS

帖最后由

0.42

RenderAtEndOf

0.37

StructEnd

0.37

httphttps

0.35

!*\

0.35

ThroughAttribute

0.35

GraphicsUnit

0.34

TabIndex

0.34

ruptedException

0.32

 defaultstate

0.32

Activations Density 0.000%

attends to time-related tokens marked from sequential dialogue timestamps to earlier spoken phrases

No Comments

No Known Activations