INDEX

Explanations

references to architectural features and historical context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

dre

-0.18

iset

-0.17

uai

-0.15

ì¡±

-0.15

iteli

-0.15

ustos

-0.15

 Rover

-0.14

 Merry

-0.14

serrat

-0.14

dere

-0.14

POSITIVE LOGITS

 tomb

0.23

 Tomb

0.21

 Jama

0.21

mas

0.20

 Friday

0.20

min

0.20

ma

0.19

Friday

0.19

tom

0.19

mos

0.18

Activations Density 0.120%

references to architectural features and historical context

No Comments

No Known Activations