INDEX

Explanations

attends to tokens related to "A" from associated tokens representing subjects in sentences

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

0:0.73

1:0.01

2:0.02

3:0.01

4:0.10

5:0.03

6:0.02

7:0.03

Negative Logits

-0.41

per

-0.39

ta

-0.38

bit

-0.37

<eos>

-0.36

 mass

-0.36

de

-0.35

“

-0.34

her

-0.34

POSITIVE LOGITS

 Monfieur

0.63

 Majefty

0.61

 Anſ

0.60

 purpoſe

0.59

 Reſ

0.59

 Shakspeare

0.58

 Jefus

0.58

 Diſ

0.58

 Cæsar

0.58

 occaf

0.57

Activations Density 0.121%

attends to tokens related to "A" from associated tokens representing subjects in sentences

No Comments

No Known Activations