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

 teachings

-0.16

 teacher

-0.16

udad

-0.15

 Teaching

-0.15

istically

-0.15

osomal

-0.15

 teaching

-0.15

TeV

-0.15

alysis

-0.14

Teacher

-0.14

POSITIVE LOGITS

/Instruction

0.23

 assistants

0.22

-learning

0.21

 assistant

0.20

/train

0.19

IGNAL

0.19

erman

0.19

Assistant

0.19

/admin

0.18

 Assistant

0.18

Activations Density 0.058%

No Comments

No Known Activations