INDEX

Explanations

references to physical locations or positions, particularly those labeled as "back."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

-0.17

atively

-0.16

ought

-0.15

elps

-0.14

uno

-0.14

riba

-0.14

peria

-0.14

ieder

-0.14

eka

-0.14

 Junction

-0.14

POSITIVE LOGITS

hoe

0.22

gam

0.20

country

0.20

scatter

0.19

lit

0.19

seat

0.19

woods

0.19

side

0.18

=back

0.18

-end

0.17

Activations Density 0.022%

references to physical locations or positions, particularly those labeled as "back."

No Comments

No Known Activations