INDEX

Explanations

references to institutions, particularly educational and professional organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

alamat

-0.15

sak

-0.15

wk

-0.15

arget

-0.15

ialized

-0.14

ovsky

-0.14

Blueprint

-0.14

Ã¶t

-0.14

 magn

-0.14

apolis

-0.14

POSITIVE LOGITS

 Surge

0.22

sur

0.21

 London

0.19

London

0.18

 Physicians

0.18

 london

0.16

 surgeons

0.15

 Dram

0.15

 perch

0.15

Sur

0.14

Activations Density 0.020%

references to institutions, particularly educational and professional organizations

No Comments

No Known Activations