INDEX

Explanations

attends to tokens with no content from tokens that appear later

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.03

2:0.04

3:0.05

4:0.06

5:0.58

6:0.09

7:0.07

Negative Logits

 myſelf

-0.80

Efq

-0.78

 itſelf

-0.70

 Houſe

-0.66

 Monfieur

-0.64

 snippetHide

-0.63

 Theſe

-0.62

 Jefus

-0.62

 Anſ

-0.60

^(@)

-0.60

POSITIVE LOGITS

0.45

0.44

↵↵

0.43

0.41

and

0.38

The

0.35

as

0.34

Activations Density 0.010%

attends to tokens with no content from tokens that appear later

No Comments

No Known Activations