INDEX

Explanations

negations and contractions

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

Ã¤tt

-0.16

unal

-0.15

incr

-0.15

lew

-0.15

ë»

-0.15

nez

-0.14

 Firmware

-0.14

unden

-0.14

forgettable

-0.14

 realpath

-0.14

POSITIVE LOGITS

 sure

0.30

sure

0.24

 alone

0.24

 phased

0.23

 anymore

0.21

 Sure

0.21

Sure

0.20

 allowed

0.20

 finished

0.19

 nearly

0.19

Activations Density 0.131%

negations and contractions

No Comments

No Known Activations