INDEX

Explanations

patterns and structures in data or equations

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

\Admin

-0.16

ç¶Ń

-0.15

ecz

-0.15

Boxes

-0.15

atas

-0.15

iku

-0.14

ÑĭÑģ

-0.14

weep

-0.14

haz

-0.14

fox

-0.14

POSITIVE LOGITS

/Dk

0.19

gard

0.16

ante

0.16

akin

0.15

arra

0.14

-metal

0.14

 Doom

0.14

ë°°

0.14

cli

0.14

amus

0.14

Activations Density 0.026%

patterns and structures in data or equations

No Comments

No Known Activations