INDEX

Explanations

specific names and titles related to people, places, or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 12-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L12R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.12.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

aw

-0.18

/Linux

-0.17

éĩı

-0.17

 lifelong

-0.16

.au

-0.15

ÙĥÙĦ

-0.15

les

-0.15

atsu

-0.15

/to

-0.15

ë¬¼

-0.15

POSITIVE LOGITS

icrous

0.22

itud

0.19

iferay

0.18

ifestyles

0.18

forms

0.17

quo

0.17

UNCH

0.16

itudes

0.16

ette

0.16

acies

0.15

Activations Density 1.231%

specific names and titles related to people, places, or organizations

No Comments

No Known Activations