INDEX

Explanations

placeholders or lines in forms and diagrams

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

å¼

-0.15

jezd

-0.15

rud

-0.14

uft

-0.14

iyel

-0.14

ield

-0.14

verbatim

-0.14

bd

-0.13

roud

-0.13

genesis

-0.13

POSITIVE LOGITS

.ObjectModel

0.15

PLICIT

0.15

^K

0.14

 much

0.14

qui

0.14

}elseif

0.14

leigh

0.13

MethodImpl

0.13

nya

0.13

Activations Density 0.001%

placeholders or lines in forms and diagrams

No Comments

No Known Activations