INDEX

Explanations

features and elements of architecture

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 30-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L30R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.30.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

jmp

-0.16

cee

-0.14

inflate

-0.14

ree

-0.14

ä½į

-0.14

éĥİ

-0.14

áº©u

-0.14

pty

-0.14

berger

-0.14

eman

-0.13

POSITIVE LOGITS

pil

0.28

 capitals

0.26

dent

0.23

 corn

0.23

fri

0.23

 keyst

0.22

 Capitals

0.22

 dent

0.21

 Ionic

0.21

 impost

0.21

Activations Density 0.035%

features and elements of architecture

No Comments

No Known Activations