INDEX

Explanations

high-frequency terms and punctuation marks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 12-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L12R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.12.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

 hakk

-0.17

ãĤ¿ãĥ¼

-0.16

 ãĥķãĤ¡

-0.14

æ¢¨

-0.14

IMIT

-0.14

ltra

-0.14

adu

-0.14

owell

-0.14

icie

-0.14

=pk

-0.14

POSITIVE LOGITS

arlo

0.19

Ã³g

0.16

ague

0.16

 hear

0.16

amer

0.15

avel

0.14

ects

0.14

 flesh

0.14

ocha

0.14

Activations Density 0.007%

high-frequency terms and punctuation marks

No Comments

No Known Activations