INDEX

Explanations

instances of the word "show" and its variations, indicating a focus on visual presentation or demonstration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

 Westbrook

-0.19

cola

-0.17

ãģĬãĤĬ

-0.15

/from

-0.15

wine

-0.15

xon

-0.15

èĢħçļĦ

-0.14

ationally

-0.14

 sÆ°

-0.14

chief

-0.14

POSITIVE LOGITS

manship

0.18

rooms

0.17

AndWait

0.16

 ÑģÐµÐ±Ñı

0.16

ÅĻeba

0.15

 Continent

0.14

ÃŃky

0.14

ocrates

0.14

 cref

0.14

OrUpdate

0.14

Activations Density 0.082%

instances of the word "show" and its variations, indicating a focus on visual presentation or demonstration

No Comments

No Known Activations