INDEX

Explanations

terms related to negative social reputations and controversies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

ello

-0.15

ened

-0.15

igated

-0.14

 abstract

-0.14

eced

-0.14

ken

-0.14

hint

-0.14

less

-0.14

anness

-0.14

Å¡el

-0.14

POSITIVE LOGITS

 factor

0.27

fest

0.25

factor

0.24

 Factor

0.24

 fest

0.24

-factor

0.23

 merchants

0.23

 merchant

0.22

 Fest

0.21

merchant

0.21

Activations Density 0.186%

terms related to negative social reputations and controversies

No Comments

No Known Activations