INDEX

Explanations

references to numerical values or ordinal rankings

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

redi

-0.19

ings

-0.18

st

-0.15

ius

-0.15

-0.14

apis

-0.14

inous

-0.14

%%%%

-0.14

ended

-0.14

led

-0.14

POSITIVE LOGITS

teen

0.40

teenth

0.36

WD

0.24

chan

0.23

wd

0.22

rier

0.22

zig

0.22

cade

0.21

0.20

-legged

0.19

Activations Density 0.213%

references to numerical values or ordinal rankings

No Comments

No Known Activations