INDEX

Explanations

attends to the token "the" from various later texts, indicating a focus on definite articles preceding significant concepts or findings

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-att/layer_0/width_16k/average_l0_61

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.02

1:0.01

2:0.00

3:0.01

4:0.01

5:0.06

6:0.03

7:0.07

8:0.15

9:0.24

10:0.07

11:0.02

12:0.07

13:0.08

14:0.02

15:0.05

Negative Logits

the

-0.61

cy

-0.61

 bone

-0.60

 brain

-0.60

 blood

-0.60

-0.59

 cell

-0.58

 skin

-0.56

dis

-0.56

 protein

-0.56

POSITIVE LOGITS

 desmotivaciones

0.74

 avoient

0.72

 étoient

0.61

 ainfi

0.59

 Houſe

0.57

 Italij

0.57

 pérd

0.56

 nôtre

0.54

 enfans

0.54

 Theſe

0.53

Activations Density 0.470%

attends to the token "the" from various later texts, indicating a focus on definite articles preceding significant concepts or findings

No Comments

No Known Activations