INDEX

Explanations

phrases or concepts that indicate depth or intensity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

oline

-0.17

naments

-0.16

eru

-0.15

ÑĢÐ¾Ñĩ

-0.15

ubit

-0.15

hood

-0.15

abel

-0.15

onse

-0.15

anger

-0.14

cean

-0.14

POSITIVE LOGITS

ening

0.26

 deep

0.23

ened

0.23

deep

0.20

 deeply

0.19

 deepest

0.18

_deep

0.18

 Deep

0.17

Deep

0.17

thro

0.17

Activations Density 0.037%

phrases or concepts that indicate depth or intensity

No Comments

No Known Activations