INDEX

Explanations

attends to specific technical or legal terms from generic or procedural references

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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.09

1:0.11

2:0.12

3:0.06

4:0.07

5:0.03

6:0.17

7:0.30

Negative Logits

TagMode

-0.28

 '\\;'

-0.28

Obrázky

-0.26

 باخ

-0.26

 Automatic

-0.26

 ModelExpression

-0.25

ktm

-0.25

Sucesor

-0.25

 slä

-0.24

 amig

-0.24

POSITIVE LOGITS

parsedMessage

0.30

brates

0.28

StoryboardSegue

0.27

HostException

0.27

createCell

0.27

InjectAttribute

0.27

openzeppelin

0.26

vician

0.26

InstanceState

0.26

astify

0.26

Activations Density 0.016%

attends to specific technical or legal terms from generic or procedural references

No Comments

No Known Activations