INDEX

Explanations

numerical results and rankings from competitive events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.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

 punch

-0.16

 Punch

-0.16

 punching

-0.15

plication

-0.15

aeda

-0.14

^.

-0.14

 /*!<

-0.14

rema

-0.14

SION

-0.14

 Banc

-0.14

POSITIVE LOGITS

ocha

0.15

longleftrightarrow

0.15

 Grade

0.14

ÅĤ

0.14

à¸Ŀ

0.14

ç¹

0.14

graded

0.13

el

0.13

Activations Density 0.014%

numerical results and rankings from competitive events

No Comments

No Known Activations