INDEX

Explanations

actions related to climbing and ascending

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

 downward

-0.17

alley

-0.16

éľĩ

-0.15

arme

-0.14

éĹ´

-0.14

ÑĥÐ½Ðº

-0.14

nero

-0.14

 Ames

-0.14

 downhill

-0.14

comings

-0.14

POSITIVE LOGITS

 Mount

0.22

 onto

0.22

 ladder

0.20

Mt

0.20

 aboard

0.20

 stairs

0.19

onto

0.18

Mount

0.18

aine

0.17

 trees

0.17

Activations Density 0.015%

actions related to climbing and ascending

No Comments

No Known Activations