INDEX

Explanations

attends to the token "possible" from a variety of contextual tokens that follow it

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

0:0.00

1:0.01

2:0.62

3:0.02

4:0.01

5:0.00

6:0.08

7:0.05

8:0.00

9:0.00

10:0.00

11:0.00

12:0.00

13:0.06

14:0.06

15:0.01

Negative Logits

to

-0.44

 effect

-0.43

in

-0.43

 only

-0.41

one

-0.40

 kind

-0.40

now

-0.40

 more

-0.39

of

-0.39

so

-0.39

POSITIVE LOGITS

 パンチラ

0.60

Boas

0.60

 beſch

0.60

featureID

0.60

<unused43>

0.59

<pad>

0.59

Zeneca

0.59

bildtitel

0.59

<unused23>

0.59

<unused41>

0.59

Activations Density 0.039%

attends to the token "possible" from a variety of contextual tokens that follow it

No Comments

No Known Activations