INDEX

Explanations

terms related to possession and favorites

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

spiel

-0.16

bsite

-0.16

undy

-0.16

Ð¾ÐºÐµ

-0.16

.yy

-0.15

ickle

-0.15

-urlencoded

-0.15

edBy

-0.15

 VerfÃ¼g

-0.14

ÙĪÙĬØ³

-0.14

POSITIVE LOGITS

desired

0.17

 favorite

0.16

py

0.15

 choice

0.14

 purchases

0.14

 Juli

0.14

 copy

0.14

Wa

0.14

wahl

0.14

OUN

0.14

Activations Density 0.149%

terms related to possession and favorites

No Comments

No Known Activations