INDEX

Explanations

references to challenges and resources in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

flen

-0.14

elig

-0.14

.allow

-0.14

ende

-0.13

 Fucked

-0.13

 lending

-0.13

 Ð²Ð°Ð³

-0.13

unami

-0.13

eron

-0.13

ï

-0.13

POSITIVE LOGITS

 consume

0.47

 consumes

0.41

 Consum

0.41

 consuming

0.40

consume

0.39

 eats

0.38

 drain

0.36

eat

0.35

 Drain

0.33

 devour

0.33

Activations Density 0.263%

references to challenges and resources in various contexts

No Comments

No Known Activations