INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

orer

-0.15

subst

-0.14

ç¦ģ

-0.14

 wich

-0.14

odo

-0.13

 mest

-0.13

/console

-0.13

VILLE

-0.13

Ð¾ÑģÑĤÑĮ

-0.13

ophilia

-0.13

POSITIVE LOGITS

 concept

0.21

Pri

0.19

concept

0.18

Pri

0.18

 aspect

0.17

aspect

0.16

 such

0.16

iggins

0.15

 Priest

0.15

.aspect

0.15

Activations Density 0.000%

No Known Activations

This feature has no known activations.