INDEX

Explanations

references to irony and unexpected contrasts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 11-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L11R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.11.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

aley

-0.17

ãĥĪãĥª

-0.17

.SC

-0.16

ory

-0.16

yk

-0.15

ALA

-0.15

centration

-0.14

antasy

-0.14

lena

-0.14

ographics

-0.14

POSITIVE LOGITS

 exactly

0.20

 precisely

0.19

0.17

ä¼ı

0.15

 caut

0.15

stile

0.14

assin

0.14

 Harris

0.14

sem

0.14

one

0.14

Activations Density 0.208%

references to irony and unexpected contrasts

No Comments

No Known Activations