INDEX

Explanations

ordinal numbers and rankings associated with positions and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

 first

-0.17

odd

-0.17

enny

-0.16

uida

-0.16

.eql

-0.16

èĭ±éĽĦ

-0.15

swick

-0.15

first

-0.14

Ø¯ÙĨ

-0.14

ulle

-0.14

POSITIVE LOGITS

 annual

0.22

 Annual

0.21

annual

0.18

Annual

0.17

amus

0.17

 Amend

0.16

_$

0.16

lis

0.16

 baseman

0.15

 ÏĮ

0.15

Activations Density 0.057%

ordinal numbers and rankings associated with positions and titles

No Comments

No Known Activations