INDEX

Explanations

concepts related to educational methods and philosophies

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

New Auto-Interp

Configuration

fnlp/Llama3_1-8B-Base-LXM-8x/Llama3_1-8B-Base-L22M-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.22.hook_mlp_out

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

inger

-0.16

ardy

-0.15

ls

-0.15

oku

-0.15

Teh

-0.14

alyze

-0.14

fatal

-0.14

 irregular

-0.14

å®ľ

-0.13

 Skeleton

-0.13

POSITIVE LOGITS

 guess

0.61

 Guess

0.56

guess

0.54

 guesses

0.52

Guess

0.51

_guess

0.45

çĮľ

0.44

 guessing

0.44

 guessed

0.42

 hypothesis

0.38

Activations Density 0.555%

concepts related to educational methods and philosophies

No Comments

No Known Activations