INDEX

Explanations

references to puzzles and word search games

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 22-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L22R-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_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 Compare

-0.15

roc

-0.14

Ø§ÛĮØ´

-0.14

 æŃ¦

-0.14

esters

-0.14

æķ

-0.13

 Poke

-0.13

ifest

-0.13

 rough

-0.13

Compare

-0.13

POSITIVE LOGITS

 puzzle

0.52

 puzzles

0.50

 Puzzle

0.44

 crossword

0.39

 sudoku

0.37

 Sudoku

0.36

 puzz

0.35

udoku

0.35

uzzle

0.34

igsaw

0.31

Activations Density 0.127%

references to puzzles and word search games

No Known Activations