INDEX

Explanations

phrases indicating direction or destination

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ä»¥æĿ¥

-0.15

enty

-0.14

Ð½Ð¾Ñĩ

-0.14

 passing

-0.14

anas

-0.13

 inward

-0.13

ona

-0.13

ÏĥÎ¼Î¿

-0.13

 Garn

-0.13

POSITIVE LOGITS

 destinations

0.33

 destination

0.29

 wherever

0.25

 destino

0.24

destination

0.24

çĽ®çļĦ

0.21

 rendez

0.21

 places

0.20

 destiny

0.19

where

0.19

Activations Density 0.166%

phrases indicating direction or destination

No Comments

No Known Activations