INDEX

Explanations

specific names and titles related to religious or historical figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

dice

-0.16

 lÃ©g

-0.15

rupa

-0.14

ÅĤaw

-0.14

 Ð¿ÐµÑĢÐµÑģ

-0.14

 Curtis

-0.14

oku

-0.14

 ä½Ľ

-0.13

 Vaults

-0.13

lands

-0.13

POSITIVE LOGITS

Ef

0.15

ordon

0.15

 detach

0.14

afi

0.14

 Pets

0.14

 Rossi

0.14

calar

0.14

Ing

0.14

 Ging

0.14

Activations Density 0.113%

specific names and titles related to religious or historical figures

No Comments

No Known Activations