INDEX

Explanations

attends to the token preceding the dot from function-related tokens later in the sequence

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

1:0.12

2:0.12

3:0.07

4:0.07

5:0.04

6:0.09

7:0.37

Negative Logits

 pleaſure

-0.45

makeConstraints

-0.41

-0.40

IndentedString

-0.37

MeasureSpec

-0.36

NameInMap

-0.36

enderror

-0.36

setVerticalGroup

-0.36

StructEnd

-0.35

Efq

-0.35

POSITIVE LOGITS

τητα

0.28

JsonProperty

0.26

Lähteet

0.26

χεία

0.25

avra

0.24

Gemeinsame

0.24

WORTH

0.24

 previos

0.22

amilya

0.21

 JUGA

0.21

Activations Density 0.026%

attends to the token preceding the dot from function-related tokens later in the sequence

No Comments

No Known Activations