INDEX

Explanations

strong emphasis on positive sentiment or expressions of praise

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

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

horn

-0.17

teenth

-0.16

ic

-0.16

oon

-0.15

ered

-0.15

uld

-0.15

omer

-0.15

cy

-0.15

ham

-0.14

venience

-0.14

POSITIVE LOGITS

anford

0.18

ASE

0.16

aney

0.16

åı·

0.15

aller

0.15

-secret

0.15

lying

0.14

ignant

0.14

anka

0.14

gether

0.14

Activations Density 0.075%

strong emphasis on positive sentiment or expressions of praise

No Comments

No Known Activations