INDEX

Explanations

imperative instructions or commands

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

utomation

-0.15

ruba

-0.15

cq

-0.14

="{!!

-0.14

ITES

-0.14

agas

-0.14

/IP

-0.14

eniable

-0.14

sole

-0.14

ugal

-0.13

POSITIVE LOGITS

few

0.17

0.16

few

0.15

 select

0.15

éľĢ

0.15

 simply

0.15

orne

0.15

 Voor

0.14

agem

0.14

Activations Density 0.058%

imperative instructions or commands

No Comments

No Known Activations