INDEX

Explanations

attends to specific numeric tokens from various pairs of other tokens in the surrounding context

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

0:0.35

1:0.23

2:0.11

3:0.08

4:0.06

5:0.02

6:0.04

7:0.07

Negative Logits

MLLoader

-0.40

ertale

-0.32

gdx

-0.31

makeConstraints

-0.30

躇

-0.29

 geométrico

-0.29

Rhestr

-0.28

 كومونز

-0.27

Nowhere

-0.27

érables

-0.27

POSITIVE LOGITS

IBOutlet

0.31

GEBURTSDATUM

0.23

inete

0.20

 szen

0.20

+)/

0.19

תוך

0.19

 today

0.19

 pack

0.19

Activations Density 1.110%

attends to specific numeric tokens from various pairs of other tokens in the surrounding context

No Comments

No Known Activations