INDEX

Explanations

attends to various context-specific tokens from the subject back to related sensory descriptions or attributes

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

0:0.14

1:0.13

2:0.17

3:0.11

4:0.07

5:0.01

6:0.08

7:0.25

Negative Logits

 CreateTagHelper

-0.40

évaluateur

-0.35

aarrggbb

-0.33

makeText

-0.33

 /\.(

-0.32

 OMIT

-0.30

wyt

-0.30

oprecipitation

-0.30

 onOptions

-0.30

KURZBESCHREIBUNG

-0.30

POSITIVE LOGITS

 €)

0.41

<_>

0.38

</s>

0.38

")

0.36

~)

0.36

€)

0.36

")]

0.35

_)

0.35

endpush

0.35

Activations Density 0.598%

attends to various context-specific tokens from the subject back to related sensory descriptions or attributes

No Comments

No Known Activations