INDEX

Explanations

attends to tokens related to various structural positions (e.g., "back," "side," "bottom," "top," "end") from tokens that indicate specific anatomical or spatial references

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

0:0.05

1:0.13

2:0.05

3:0.03

4:0.11

5:0.50

6:0.04

7:0.05

Negative Logits

 AssemblyCulture

-0.32

 <<<<<<<<<<<<<<

-0.26

goku

-0.24

giver

-0.23

databind

-0.23

 allah

-0.22

票

-0.22

hartt

-0.22

UserScript

-0.22

 Welt

-0.21

POSITIVE LOGITS

 BoxFit

0.30

 فريبيس

0.29

 مشين

0.27

 सत्यापित

0.27

%"),

0.26

Życiorys

0.26

complexContent

0.26

Gdx

0.26

FunctionFlags

0.25

 hereof

0.24

Activations Density 0.268%

attends to tokens related to various structural positions (e.g., "back," "side," "bottom," "top," "end") from tokens that indicate specific anatomical or spatial references

No Comments

No Known Activations