INDEX

Explanations

variations of the word "answer."

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

geb

-0.17

irst

-0.17

ilet

-0.17

PEED

-0.16

Ø±Ø§ÙĤ

-0.16

ernen

-0.16

ernal

-0.15

undles

-0.15

away

-0.15

audi

-0.15

POSITIVE LOGITS

ToSelector

0.17

er

0.17

able

0.17

ing

0.17

phone

0.17

nable

0.16

pq

0.16

.answer

0.16

hone

0.15

 questions

0.15

Activations Density 0.031%

variations of the word "answer."

No Comments

No Known Activations