INDEX

Explanations

attends to injury-related tokens from auto-related tokens

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

0:0.08

1:0.11

2:0.10

3:0.13

4:0.10

5:0.05

6:0.24

7:0.15

Negative Logits

+#+#

-0.27

Par

-0.25

djangoproject

-0.24

 eventdata

-0.23

ю

-0.23

вач

-0.22

بخ

-0.22

сто

-0.22

sos

-0.21

landır

-0.21

POSITIVE LOGITS

 itſelf

0.41

sidemargin

0.41

Климат

0.40

Efq

0.40

 ARXIV

0.39

 GoogleFonts

0.37

 poffible

0.37

 doubtnut

0.36

 Jefus

0.36

CodedInputStream

0.35

Activations Density 0.041%

attends to injury-related tokens from auto-related tokens

No Comments

No Known Activations