INDEX

Explanations

punctuation marks, particularly question marks and periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Evaluate

-0.19

 perhaps

-0.16

 would

-0.15

 maybe

-0.15

perhaps

-0.15

BitFields

-0.14

uchi

-0.14

ä»Ķ

-0.14

 Perhaps

-0.14

uiltin

-0.14

POSITIVE LOGITS

 Where

0.28

 What

0.27

 what

0.27

Can

0.26

How

0.25

 where

0.25

 Will

0.24

Where

0.23

What

0.23

how

0.23

Activations Density 0.046%

punctuation marks, particularly question marks and periods

No Comments

No Known Activations