INDEX

Explanations

conjunctions and components of lists

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 13-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L13R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.13.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

veloper

-0.16

ingham

-0.15

rous

-0.14

essim

-0.14

rix

-0.14

Æ¯

-0.14

á»ĭ

-0.13

-sdk

-0.13

roots

-0.13

Åº

-0.13

POSITIVE LOGITS

how

0.24

how

0.19

 whether

0.19

why

0.18

/how

0.17

Ã³mo

0.17

 cÃ³mo

0.16

 what

0.16

å¦Ĥä½ķ

0.16

ndef

0.15

Activations Density 0.086%

conjunctions and components of lists

No Comments

No Known Activations