INDEX

Explanations

attends to tokens containing "ne" from tokens that also contain "ne" appearing earlier in the sequence

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.03

1:0.45

2:0.23

3:0.02

4:0.05

5:0.04

6:0.04

7:0.10

Negative Logits

-0.44

-0.40

↵

-0.39

-0.38

-0.36

ne

-0.35

-0.34

-0.33

POSITIVE LOGITS

ThroughAttribute

0.67

oredCriteria

0.67

 betweenstory

0.62

 ainfi

0.60

 مشين

0.59

 виправивши

0.59

WithIOException

0.59

tvguidetime

0.59

 bezeichneter

0.58

berdayakan

0.58

Activations Density 0.016%

attends to tokens containing "ne" from tokens that also contain "ne" appearing earlier in the sequence

No Comments

No Known Activations