INDEX

Explanations

attends to potential outcomes or possibilities marked by "could" from statements expressing certainty or constraints marked by "be," "and," "in," and "not."

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

0:0.01

1:0.01

2:0.77

3:0.02

4:0.01

5:0.00

6:0.04

7:0.01

8:0.00

9:0.00

10:0.00

11:0.00

12:0.00

13:0.02

14:0.03

15:0.00

Negative Logits

be

-0.89

 have

-0.66

let

-0.61

not

-0.61

do

-0.61

 find

-0.61

get

-0.58

run

-0.58

see

-0.57

 come

-0.57

POSITIVE LOGITS

 zijne

0.48

 mijne

0.47

stateProvider

0.45

IsContent

0.45

IContainer

0.45

SequentialGroup

0.43

zelfde

0.43

 afbeeldingen

0.43



0.43

WithIOException

0.43

Activations Density 0.052%

attends to potential outcomes or possibilities marked by "could" from statements expressing certainty or constraints marked by "be," "and," "in," and "not."

No Comments

No Known Activations