INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.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

ELS

-0.16

iais

-0.15

ONSE

-0.15

eli

-0.15

undi

-0.14

 Technology

-0.14

jar

-0.14

lemn

-0.14

rais

-0.14

POSITIVE LOGITS

VERTISEMENT

0.16

isle

0.16

jedn

0.15

mith

0.14

.Enable

0.14

endo

0.14

 ÐļÐ°

0.14

ÏģÏħ

0.13

Ã©ment

0.13

PMENT

0.13

Activations Density 0.000%

No Known Activations

This feature has no known activations.