INDEX

Explanations

references to attention and observation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 30-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L30R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.30.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

hiba

-0.17

din

-0.15

ayah

-0.15

anka

-0.15

uai

-0.15

_ENDIAN

-0.14

isl

-0.14

viÄį

-0.14

idebar

-0.14

egis

-0.14

POSITIVE LOGITS

 trained

0.40

 fixed

0.32

trained

0.31

fix

0.29

 focused

0.28

riv

0.27

-trained

0.26

 laser

0.25

 peeled

0.24

fixed

0.24

Activations Density 0.048%

references to attention and observation

No Comments

No Known Activations