INDEX

Explanations

attends to component-related tokens from tokens that appear at the end of a code snippet

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

0:0.13

1:0.18

2:0.09

3:0.06

4:0.11

5:0.02

6:0.08

7:0.28

Negative Logits

addGap

-0.35

 насељу

-0.35

 ISNI

-0.35

bootstrapcdn

-0.34

 "..\..\

-0.34

gebob

-0.33

racite

-0.32

]='\

-0.31

addCriterion

-0.30

orsese

-0.30

POSITIVE LOGITS

 WebDriverWait

0.35

XtraGrid

0.34

LError

0.31

posedge

0.30

Hochspringen

0.29

 Wern

0.29

?''

0.29

0.28

rungsseite

0.28

."*

0.28

Activations Density 0.001%

attends to component-related tokens from tokens that appear at the end of a code snippet

No Comments

No Known Activations