INDEX

Explanations

terms associated with human-created or artificial constructs

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 2-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L2R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.2.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

ÑĢÐ¸Ð½

-0.15

 Claw

-0.14

rei

-0.14

argin

-0.14

ALLED

-0.13

amma

-0.13

phia

-0.13

 è±

-0.13

ESCO

-0.13

Ð°ÐºÐ¸

-0.13

POSITIVE LOGITS

-Ta

0.15

ajar

0.15

keit

0.15

.EVT

0.14

åĿª

0.14

iyat

0.14

uchar

0.14

ÑĤÐ°Ð¿

0.14

enderit

0.14

Ã³b

0.14

Activations Density 0.023%

terms associated with human-created or artificial constructs

No Comments

No Known Activations