INDEX

Explanations

attends to tokens with alphabetical sequences from tokens marked with just a single alphabetical character

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

0:0.02

1:0.03

2:0.07

3:0.49

4:0.06

5:0.08

6:0.14

7:0.08

Negative Logits

↵

-0.56

-0.51

-0.50

-0.49

-0.48

-0.47

POSITIVE LOGITS

 Monfieur

0.99

 itſelf

0.95

 myſelf

0.95

Efq

0.94

 purpoſe

0.89

 Majefty

0.88

 themſelves

0.85

 Jefus

0.85

ſelf

0.84

^(@)

0.84

Activations Density 0.204%

attends to tokens with alphabetical sequences from tokens marked with just a single alphabetical character

No Comments

No Known Activations