INDEX

Explanations

attends to mathematical terms and representations from complex symbols or structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104

Prompts (Dashboard)

36,864 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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.01

1:0.05

2:0.10

3:0.34

4:0.06

5:0.09

6:0.25

7:0.07

Negative Logits

 myſelf

-0.64

 itſelf

-0.64

 raiſ

-0.63

 Monfieur

-0.62

 purpoſe

-0.61

 Theſe

-0.60

Efq

-0.60

 iſt

-0.59

 ſche

-0.58

 ―――――

-0.58

POSITIVE LOGITS

0.40

0.39

to

0.38

0.35

<eos>

0.35

con

0.34

0.33

Mas

0.32

Activations Density 1.501%

attends to mathematical terms and representations from complex symbols or structures

No Comments

No Known Activations