INDEX

Explanations

attends to the action "go" from tokens that prompt a return or reflection, specifically "back."

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

0:0.04

1:0.01

2:0.02

3:0.02

4:0.07

5:0.76

6:0.02

7:0.02

Negative Logits

 myſelf

-1.24

Efq

-1.17

 itſelf

-1.12

 ―――――

-1.05

^(@)

-1.05

 ་་

-1.04

 Anſ

-1.03

NUMX

-1.03

ſelves

-1.03

ſelf

-1.02

POSITIVE LOGITS

0.78

0.72

and

0.67

0.66

↵↵

0.61

0.59

<eos>

0.58

or

0.57

Activations Density 0.059%

attends to the action "go" from tokens that prompt a return or reflection, specifically "back."

No Comments

No Known Activations