INDEX

Explanations

references to games and competitions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

 straw

-0.16

 Hakk

-0.15

hab

-0.14

apiro

-0.14

aster

-0.14

 ordering

-0.14

 Ordering

-0.13

asts

-0.13

 unsubscribe

-0.13

ordering

-0.13

POSITIVE LOGITS

 caches

0.33

 caching

0.31

 cache

0.29

/cache

0.26

 Cache

0.26

-cache

0.26

cache

0.25

Cache

0.25

.cache

0.24

 CACHE

0.23

Activations Density 0.013%

references to games and competitions

No Comments

No Known Activations