INDEX

Explanations

attends to numerical tokens or symbols that denote ranges or comparisons of magnitudes from scalar values

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

0:0.12

1:0.16

2:0.23

3:0.07

4:0.07

5:0.04

6:0.06

7:0.21

Negative Logits

MessageTagHelper

-0.34

 tartalomajánló

-0.33

ConstraintMaker

-0.32

 Smarty

-0.31

TagMode

-0.31

#+#

-0.31

RegistryLite

-0.31

aimana

-0.30

 LoginComponent

-0.29

Leary

-0.28

POSITIVE LOGITS

তথ্যসূত্র

0.37

Viitteet

0.31

 amarillas

0.29

 оригіналу

0.29

 umani

0.29

kezik

0.28

सन्दर्भ

0.28

giar

0.28

 Hawthorne

0.27

mators

0.27

Activations Density 0.572%

attends to numerical tokens or symbols that denote ranges or comparisons of magnitudes from scalar values

No Comments

No Known Activations