INDEX

Explanations

attends to plugin-related tokens from the closing tags of plugin tokens

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

2:0.31

3:0.05

4:0.17

5:0.12

6:0.04

7:0.08

Negative Logits

is

-0.34

His

-0.31

reiz

-0.30

 tirer

-0.30

 scena

-0.30

con

-0.29

 halls

-0.29

uco

-0.29

POSITIVE LOGITS

^(@)

0.53

 ―――――

0.52

HZ

0.51

 ་་

0.50

LUT

0.49

awtextra

0.48

ˏ

0.48

 doubtnut

0.48

PZ

0.47

PFA

0.47

Activations Density 0.058%

attends to plugin-related tokens from the closing tags of plugin tokens

No Comments

No Known Activations