INDEX

Explanations

instances of high numerical values, likely related to data or statistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

ifornia

-0.17

agara

-0.15

illow

-0.15

entar

-0.14

 Volume

-0.14

.Sm

-0.14

 volume

-0.14

 overall

-0.13

covering

-0.13

faq

-0.13

POSITIVE LOGITS

ÙĨØ¯Ùĩ

0.16

poz

0.15

 appl

0.14

rchive

0.14

ALSE

0.14

Ã¢n

0.14

_losses

0.14

 poÅ¾

0.13

avaÅŁ

0.13

haar

0.13

Activations Density 0.000%

instances of high numerical values, likely related to data or statistics

No Comments

No Known Activations